research in signal processing, computing · 2015-05-16 · research in signal processing, computing...

Research in Signal Processing, Computing &Communication-System Design

Dr. VSK Reddy Dr. D. RaghuRami ReddyProfessor & Principal Dean AcademicsMRCET, JNTUH MRECW, JNTUHHyderabad, Telangana, India. Hyderabad,Telangana, India.

Dr. Syed Abdul Sattar Dr. PVY JayasreeDean, RITS Professor, ECEJNTUH, Hyderabad, Telangana, India. GITAM University

Vishakapatnam, A.P. India.

Dr. B. Vijayakumar Prof. J. Krishna ReddyDean R&D Dean IIPCMRECW, JNTUH MRECW, JNTUHHyderabad, Telangana, India. Hyderabad, Telangana, India.

Dr. S. Srinivasa Rao Prof. K.Ramesh BabuProfessor & HOD, ECE Professor & HOD, CSEMRCET, JNTUH MRECW, JNTUHHyderabad, Telangana,. India. Hyderabad, Telangana,India.

Prof. P. Sanjeeva Reddy Prof. A. Radha RaniDirector, School of Electronics Professor & HOD, ITMRCET, JNTUH MRECW, JNTUHHyderabad, Telangana,. India. Hyderabad, Telangana,. India.

Dr.V. Malleswara Rao Prof. K.KAILASA RAOProfessor & HOD, ECE Dean Placements, MRGIGITAM University Hyderabad, Telangana, IndiaVishakapatnam, A.P. India

Editor – in - Chief Managing EditorDr. Y. Madhaveelatha K. Niranjan ReddyProfessor & Principal Professor & HOD, ECEMRECW, JNTUH MRECW, JNTUHHyderabad, Telangana, India. Hyderabad, Telangana, India.

InterInterInterInterInternananananational Jtional Jtional Jtional Jtional Jourourourourournal ofnal ofnal ofnal ofnal ofRRRRResearesearesearesearesearccccch in Signal Prh in Signal Prh in Signal Prh in Signal Prh in Signal Processingocessingocessingocessingocessing,,,,,Computing & CommComputing & CommComputing & CommComputing & CommComputing & Communicaunicaunicaunicaunication-tion-tion-tion-tion-System DesignSystem DesignSystem DesignSystem DesignSystem Design

AN INTERNATIONAL JOURNAL

MALLA REDDY ENGINEERING COLLEGE FOR WOMEN MALLA REDDY EDUCATIONAL SOCIETY

Accredited by NAAC with ‘ A’ Grade Permanently Affiliated to JNTUH, Approved by AICTE,

ISO 9001:2008 Certified InstitutionMaisammaguda, Dhulapally, Secunderabad

I J R S C S DI J R S C S DI J R S C S DI J R S C S DI J R S C S DVolume : 01, Issue: 01, Jan-June, 2015

International Journal ofResearch in Signal Processing, Computing &

Communication-System DesignVolume : 01, Issue: 01, Jan-June, 2015

CONTENTS

Video Watermarking in Motion Vectors Based on Visual Masking ………………………….… 01 - 06K.Swaraja, Dr. Y.Madhavee Latha and Dr.V.S.K.Reddy

NLM for Improvement of Sensing Capability in Femtocells …………………………………… 07 - 10Dr S.Srinivasa Rao

A Novel Digital Image Tamper Detection and Recovery by ………………………………….…11 - 15Grouping Most Significant Bits

Dr. B. Vijay Kumar

Verification of GPIO Core Functions Using Universal Verification …………………………….16 - 19Methodology

K. Niranjan Reddy, U.DhanaLakshmi and Dr. PVY Jaya Sree

Intelligent Wireless Embedded System for Vehicle Control in Transportation ……….…………20 - 23System Based on GSM and GPS Technology

S. Srivani, M.Haritha and K. Sumalatha

Multiple frequency Polyhedron Ring Slot Microstrip Antenna ..................................................... 24 - 26Srilakshmi. A, N.V. Koteswara Rao and D.Sreenivasa Rao

Provision of Trustworthy Associations in Peer to Peer Structures ................................................ 27 - 29Ramesh Babu and A. Radha Rani

Features Extraction and Selection based on Rough Set in Shot Detection ................................... 30 - 35GS Naveen,VSK Reddy and S. Srinivas Kumar

Mobile Element Routing, Data Gathering and Energy Efficient Data TransmissioninWireless Sensor Networks ............................................................................................................. 36 - 43

S. Ravi Kumar, D. Srinivasa Rao

International Journal of Research in Signal Processing, Computing & Communication-System DesignISSN: 2395 – 3187

Vol: 01, Issue: 01, Jan-June, 2015

1

Video Watermarking in Motionvectors Based on visual masking

K.Swaraja Dr. Y.Madhavee Latha Dr.V.S.K.Reddy

[email protected] , [email protected] [email protected]

Gokaraju Rangaraju Institute of Engineering &Technology, Hyderabad, T.S, India1

Malla Reddy Engineering College for women, Hyderabad, T.S ,India2,

Malla Reddy College of Engineering &Technology, Hyderabad, T.S, India3,

Abstract— Digital watermarking is an effective technologyfor digital products copyright protection and data securitymaintenance as well as an important branch of informationhiding technology. By virtue of this technology, copyrightinformation is embedded in the video data to provideownership verification. Several watermarking schemeshave been proposed in recent years, but most of them dealwith still images, only some being extended over to thetemporal domain for video watermarking. But again mostof those approaches are applied to uncompressed videoprocessing domain. In this paper, a new compressed videowatermarking procedure is explained. We propose a novelhybrid digital video watermarking scheme embeddingwatermark in P-frames. Search its best match block bythe watermark while embedding and the embeddingstrength of every block in the video sequence is calculatedwith the set of non-linear formulas that have beenproposed, according to the entropy of motion informationof every macro-block and the human visual maskingsystem. The experimental results demonstrate that thismethod impacts the video quality slightly but bit rate iscontrolled to a large extent.

Index Terms— Introduction, MPEG-4 Video Compression,Related work, Visual masking, Video Watermarking in P-Frames, Experimental results & performance evaluation,conclusion.

I. INTRODUCTION

Digital video watermarking refers to techniques for embeddingadditional data into host video by utilizing the redundancy ofthe video due to the limitations of the human visual system(HVS). The information hidden in the host video can be usedas an invisible label for copyright protection, or as auxiliaryinformation for video segmentation, retrieval, annotation,indexing, error concealment, etc. The main requirements oninvisible digital watermark normally include imperceptibility,

robustness, and capacity. Encryption techniques are commonlyused to control access of the multimedia Contents. However,they do not provide any protection after the digital contentshave been decrypted. The copyright should be such that itwould be both easy to detect, yet hard to remove, and withthis kind of information being embedded in the object, themultimedia source would also be well protected. It wouldthen be possible to prove copyright ownership if something isused illegally. This would make the multimedia providers feelmore comfortable in supplying Copyrighted materials, andalso benefit users by being able to share more of theinformation. Digital watermarking technology has emerged asan effective means to hide copyright information in the originalcontent to protect the authenticity of the intellectual property.There are three principal processes involved in robustwatermarking: watermark embedding, attack, and watermarkdetection. In watermark embedding, a watermark is constructedand then embedded into an original signal to produce thewatermarked signal. For security, watermark embedding usuallyrequires knowledge of a secret embedding key. In addition,some watermarks also allow auxiliary information to beencoded in the watermark, known as the message or payload.Once the watermark has been embedded, the watermarkedsignal may be subjected to attack. There are many differenttypes of attacks, including those which attempt to remove thewatermark, make the watermark more difficult to detect, orsubvert the security of the watermark. In watermark detection,a test signal is provided to the watermark detector. The testsignal may be watermarked and possibly attacked, or may nothave been watermarked at all. The watermark detectorexamines its input signal and reports whether the watermarkis present or not, and if applicable, extracts the payload. Ifthe watermark detector does not require access to the original(unwatermarked) signal, the watermarking technique is knownas a blind technique.

The video watermarking methods mainly have two kinds:Spatial domain methods and transformation domain method



2

Compared with the spatial domain, the watermarking methodson the transform domain, especially on the DCT (discretecosine transformation) domain, have the following advantages:First of all, the characteristic of humanity vision System canbe used effectively in the transformation domain; secondly,the watermarking methods can be compatible with the videocompression standards (for example MPEG and so on); Lastly,the computation complexity of the watermarking algorithm inthe DCT domain is often low. Typically the watermarkinginclude transaction tracking, copy control, authentication,legacy system enhancement and database linking etc. Growingpopularity of video based applications such as Internetmultimedia, wireless video, personal video recorders, video-on-demand, set-top box, videophone and videoconferencinghave a demand for much higher Compression to meetbandwidth criteria and best video quality as possible. Differentvideo Encoder Decoders (Codec’s) such as MPEG-1, MPEG-2, MPEG-4, ITU-T H.261, and H.263 have evolved to meetthe current requirements of video application based products.Among various available Standards H.264 / Advanced VideoCodec (AVC) is becoming an important alternative regardingreduced band width, better image quality in terms of peak-signal-to-noise-ratio (PSNR) and network friendliness, but itrequires higher computational complexity.

Current video watermarking algorithms can be divided intothree classes: watermarking in the raw, video, Watermarkingin the compressed video streams and Watermarking in theencoding process [4-6]. Watermarking in the raw video canuse many algorithms for still images, but it need ‘large amountsof calculations and may lose some Watermark messages aftervideo compression. Watermarking in the compressed videostreams has small amounts of calculations so that the watermarkcan be real-time embedded. The disadvantage of this approachis that the amount of embedded message can’t be too largedue to the compression bit rate limit. The watermarkingalgorithms in the encoding process are robust against MPEGcompression and will not increase bit rate of the video streams.It may farther be categorized into two kinds: watermark in Iframe and watermark in P or B frame. During the encodingprocess, MPEG-2 deals with I frames similarly to JPEG dealingwith still images, so the Watermark in I frame is oftenembedded in DCT coefficients.

But the’ total number of I frames of a video is smaller, so thetotal message embedded in I frames is comparatively small.The video sequences contain a significant amount of P and BFrames. There are two positions in P or B frames that can beused to embed watermark. One is the prediction error; theother is the motion information. When embedding watermarkin the prediction errors, the loss of prediction error dataincreases with the increase of compression ratio.

Comparatively, embedding the watermark in the motioninformation is more robust.

This paper is organized into seven sections. The next sectiongives brief introduction of MPEG-4 Compressed. Section IIIdescribes the details of related works regarding compressedvideo watermarking schemes. Section IV describes about thevisual masking and Section V describes the Videowatermarking in P-Frames (vwm-p frames). The experimentalresults and performance evaluation are shown in section VI.Section VII presents a conclusion.

II. MPEG-4 VIDEO COMPRESSION

As the most advanced video compression standard at present,MPEG-4 is widely used. To take advantage of temporalredundancy, MPEG-4 standard includes three kinds of frames:

1) Intra picture frames (I-frames);

2) forward-predicted frames (P-frames);

3) Bidirectional-predicted frames (B-frames).

I-frames are coded without reference to other frames. P-frameapplies motion prediction by referencing an I-frame or P-frame in front of it, motion vector points to the block in thereferenced frame. B-frame applies motion prediction,referencing a frame in front of it and (or) a frame behind it.Each of the two referenced frames may be I-frame or P-frame. Macro block (MB) in video stream is represented asa 16x16 sample area. Each MB contains six 8x8 blocks, fourfor luminance and two for chrominance. A block of I-framecontains simply values of luminance or chrominance of itsown. A block of P-frame or B-frame contains the differencebetween the values of itself and the referenced block. Thisprocess is called motion compensation. Each frame is dividedinto MBs. Coding process of each block includes DCT,quantization, run-level coding and entropy coding in order.The resulting video stream consists of entropy codes, motionvectors and control information about the structure of videoand characteristics of coding. The structure of frame sequence,coding process and decoding process of MPEG-4 video canbe described as following figures:

Figure 1. MPEG-4 frame sequence



3

Figure 3. MPEG-4 decoding process

Figure 2. MPEG-4 encoding process

III. RELATED WORKS

Watermarking has received much attention due to thepopularity of data communication through the internet. Amongvarious media data, digital video is the one that carries themost amount of data. Hence, it is not easy and realistic toembed watermarks directly in a raw video in real time. Usually,a raw video has to be compressed first and then watermarkedbefore it is transmitted through the network. However, themajor concern is how to design a feasible compressed videowatermarking scheme such that the hidden watermarks couldbe detected in real time.

In the literature, only a few compressed video watermarkingschemes [7, 8] were proposed. In [7], the header/sideinformation and motion vectors of MPEG2 bit stream are notchanged during watermarking. They arranged a watermarksequence to be 2D and have the same size with video frame.Then the watermark signal is 8x8 DCT transformed and addedinto DCT coefficients of video streams. In other words theircompressed domain video watermarking is infact performedin DCT domain. Therefore some processing operations suchas inverse entropy coding and inverse quantization arerequired. Besides no attacks were tested in their experiments.In [8] Langelaar et al proposed a video watermarking schemeperformed in compressed domain based on VLC codewords.At first, they divided run-level pairs into many groups withthe same VLC codeword length under the constraint that thelevel difference in each group should be exactly one. During

watermark embedding, a run-level pair was either unchangedor replaced depending on the incoming watermark value. Theirmethod was basically a least significant bit (LSB) type.Recently, Langelaar et al. proposed a differential energywatermarking (DEW) algorithm performed in the DCT domain.DEW means that watermark bits are inserted by removing thehigh-frequency DCT coefficients. The authors claimed that itis not possible to remove the DEW watermark without causingperceptual degradation.

The pioneering work was reported by Hartung et al. [9].Theyproposed embedding a watermark into DCT coefficients inMPEG-2 compressed data. Their method was based on spreadspectrum, and a modulated pseudo random pattern was addedon the DCT coefficients. They also raised the drift problemin an inter-frame prediction loop between a video encoderand a decoder, and proposed a drift compensation algorithm.However, the compensation which is almost the same as videotranscoding. Alattar, et al. [10] proposed a similarwatermarking method, in which an embedded watermark signalhas a kind of geometrical structure in order to provideresilience to geometrical attacks. The embedding process isalmost the same as [9] with some adjustment to MPEG-4encoding, but the detection process requires MPEG-4 decodingto analyze pixel data. Another approach is proposed by Ghosh,et al. [11]. Their method embeds a watermark during videocompression. The watermark signal is constructed from a pixelpattern of a reference frame designated by a motion vector,and it is modulated by the messages to be embedded. It canbe detected from the compressed data directly; however thewatermark cannot be embedded into compressed video data.Sakazawa et al. [12] proposed a method that can detect thewatermark from MPEG-2 encoded data directly. It employsalteration of DCT coefficients at spatiotemporally distributedlocations, and the watermark can be detected by observingDCT coefficients in the MPEG bit stream. But it has to embedthe watermark on the uncompressed domain. Thus, theconventional method does not satisfy the requirements forlow complexity native watermarking.

IV. VISUAL MASKING

A. The temporal visual masking of video JNDt(x, y)

The great difference between the video watermarking and thestill image watermarking is that the video exist a great mountof the redundancy on temporal axis. We should apply theinter-frame and temporal special motion redundancy of videosequences to calculate the temporal visual masking of videos.In [13] Zhi Li used the block-matching techniques to acquirethe motion vector MV (u,v) in the corresponding macro-blockM(x, y). Through the motion vector MV (u,v), we could obtainthe motion velocity, direction of moving and the degree of the



4

deformation for each macro-block. The following section wedescribe the criterion for calculating the temporal visualmasking based on these moving characteristics in the video:a) the visual masking of motion velocity; b) the visual maskingof deformation;

Visual masking of velocity. When embedding watermarkingsignals into moving objects in video sequence we considerthe factor of velocity. The length of the motion vector ofevery macro-block M (u, v) indicates velocity of motion vector.When the value of velocity is large, macro-block M(x, y)moves fast, Watermarking signals embedded in this block willbe more imperceptible.

| MV(u,v) | = √(MVc(u,v))2 + (MV

c(u,v))2

Visual masking of deformation. At the same time we considerthe influence of the deformation in the corresponding blockbetween every two frames. If there has high degree of macro-block deformation, the HVS decrease its sensitivity to thedeformation, it means embedded watermark in the macro-block will be more difficult to perceive. We calculated thedegree of deformation Def (x, y) of macro-block M (x, y)according to the following formula.

15

D(x,y) = ∑|Yc (x+i,y+j)-Y

p (x+k+i, y+1+j)| - 15≤k, 1≤15

i,j=0

isual masking of moving direction. Finally, we consider theinter-frame motion characteristics randomness by statistic theinformation of moving direction for every two frames. Weused temporal angle è (u, v) of motion vector in correspondingblock as the degree of randomness in two frames. If the objectsmove randomly between frames, the value of temporal angleis large. Objects with a large temporal angle value are inducedthe embedded watermarking signals are more difficult toperceive than that with small temporal angle value. We usethe following formula to calculate the temporal movingdirection through using motion vector in corresponding blockof every two frames

θ(u,v) = across MV

c(u,v) - MV

p(u,v)

| MVc(u,v) | * | MV

p(u,v) |

Then, we use the method below to acquire the JND t (u, v)with the formula

JNDt (u,v) = D (x,y)* θ(u,v) * | MV(u,v) |

B. The motion entropy

Entropy of a source could indicate the amount of theinformation come from the source. As the entropy increases,

we can say the source increase its amount of information tothe perceiver. We extend the concept of the entropy to thevideo sequence to get the motion entropy of the videosequence. The motion entropy increase and decrease isrespected to the increase and decrease in the motion complexityof the video. We calculate the motion entropy of the |MV (u,v) |, θ (u, v) of every macro-block to indicate whether thearea where much motion occurs in the video sequence or not,by using the below formula. To get the accurate and reallymotion information we did not quantify the motion entropy ofthe |MV (u, v) |, θ (u, v) in every macro-block. Where, Hp(|MV (u, v)|), Hp (θ (u, v)) is the motion entropy of the |MV(u, v)|, θ (u, v) in the pth frame. After getting Hp(|MV(u, v)|),Hp(θ (u, v) ) of every p frame in the video, we project all ofHp(|MV(u, v)|) and Hp( θ (u, v) ) to the histogram to get thethreshold Tv, Tθ. We could adjust whether the motion entropyof the video sequence is larger than the thresholds Tv, Tθ ornot. So we could know the macro block is motion area orstatic area.

C. Non-linear formula calculate visual masking JND(x,y)

We conclude the visual masking of the video is consisted ofthe spatial and temporal visual masking. We get the accuratemotion information according to the motion entropy of everymacro-block. So we apply the non-linear formula and themotion entropy to calculate the visual masking of videosequence that depends on the temporal visual masking of themoving information as well as the spatial visual masking ofstill image properties. So we get the two kinds of areas by thefollowing describe.

Motion areas: block in which the motion entropy of lengthof motion vector Hp(|MV(u, v)|) is not less than threshold Tvor the motion entropy of moving direction Hp(θ (u, v) ) is notless than threshold Tθ. Objects in these areas are moving fastor random. To the moving fast or random areas we use thefollowing formula to calculate the visual masking of video

JND(x,y)=JNDs(x,y)+JNDt(x,y)á*min{JNDs(x,y),JNDt(x,y)}0<α≤1

Static areas: blocks in which the motion entropy of length ofmotion vector Hp(|MV(u, v)|) is less than threshold Tv andthe motion entropy of moving direction Hp(θ (u, v) ) is lessthan threshold Tθ. Objects in these areas are static. To thestatic areas we use the below formula to calculate the visualmasking of video

JND (x, y) =β*JNDs (x, y), 0<β≤1

We use visual masking JND (x, y) of every block in video asthe maximum embedding strength, which could guarantee theexcellence imperceptibility of this scheme.



5

V. VIDEO WATERMARKING IN P-FRAMES(VWM-P FRAMES)

In the MPEG compression algorithms, a good search algorithmwill lead to good video quality and less time spent in motionestimation. FS (full search algorithm) is one of the commonestalgorithms. FS can gain the best video quality, but it spendsthe most time. So many fast search algorithms are researchedto take the place of FS such as 3SS (three step search), N3SS(new three step search), DS (diamond search), HS (hexagonsearch), etc. But new problem appeared that it is easy for themacro block searched to get into the local best, which leadsthat the bit rate is increased and the video quality may bedecreased. From many researches, we know that as long aswe use fast search algorithm, it is difficulty to avoid theoccurrence of local best. This local character arouses ourinterest.

A. Embedding scheme

Watermarking information can be easily embedded into MPEGvideo stream by the following steps.

1. Confirm which search algorithm is used in MPEG-4

2. We will select only p-frames and restrict the search regionin search algorithm to embed the watermarkinginformation

3. Execute the search algorithm to find the motion vectorof the best matching block.

4. Different information can be embedded into motion vectoras mentioned above and by making use of motion entropyand visual masking JND (x, y) of every block in thevideo as the maximum embedding strength, will guaranteethe excellent imperceptibility of this scheme.

B. Retrieving scheme

Watermarking information can be easily extracted from MPEGvideo stream by the following steps.

1. We will take the mpeg-4 watermarked video.

2. We will take only p-frames and restrict the search regionin search algorithm to extract the watermarkinginformation.

3. Execute the search algorithm to find the motion vectorof the best matching block.

4. by making use of motion entropy and visual masking ofevery block in the video the watermark will be extracted.

VI. EXPERIMENTAL RESULTS AND PERFORMANCEEVALUATION

As a measure of Imperceptibility, the peak signal to noiseratio is typically used. From the result, we can see that thereis no much change in PSNR and bit rate impacted iscomplicated when algorithm [14] is used but bit rate iscontrolled to a large extent in our algorithm as we areembedding only in the motion vectors which are having highimperceptibility based on motion entropy of visual masking.We came to know that when there are many smooth regionsin the picture, the bit rate is increased slightly .In such casesless watermarking information should be embedded to avoidincrease in the bit rate.

Table 1.Results of salesman sequence

Original Nvwm in vwm-pvideo motion frames

vectors[14]

PSNR Y 35.82 35.80 35.81

PSNR U 39.54 39.53 39.52

PSNR V 40.23 40.19 40.20

Bit rate 91504 94816 93310

VII. CONCLUSION

In this paper, a novel video watermarking scheme in motionvectors is proposed. The visual masking information of themotion vector is used to embed the watermarking information.This method impacts the video quality slightly but bit rate iscontrolled to a large extent as we are embedding in p-framesand that to in the motion vectors which are having highimperceptibility.

Original video Watermark embedded



6

REFERENCES

[1] I. J. Cox and M. L. Miller, “Electronic watermarking: the first 50years”. Fourth, IEEE Workshop on Multimedia Signal Processing,2001, pp. 225-230.

[2] Hartung F. and Girod B., Digital watermarking of mpeg-2 codedvideo in the bit stream domain, IEEE International Conference onAcoustics, Speech, and Signal Processing, vo1.4 pp. 2621-2624,1997

[3] Chung T-Y.,P ark K.-S., Oh Y.-N., Shin D.-H. and Park S.-H.,Digital watermarking for copyright protection of ’MPEG2Compressed video: IEEE Transactions 0n Consumer ElectronicsV01.44,N0.3p,p . 895 - 901.1998.

[4] Hsu Chiou-Tung and Wu Ja-Ling, Digital watermarking for video,13th International Conference on Digital Signal ProcessingProceedings, Vol. 1, pp.217 -220, 1997

[5] Kutter M., Jordan E and Ebrahimi T., Proposal of a watermarkingtechnique for hiding/retrieving data in compressed and decompressedvideo, Technical report M228 1 , ISO/IEC document, JTC I/SC29/WG11,1997

[6] [6] Zhang Jim, Li Jiegu and Zhang Ling, video watermark techniquein motion vector, Proceedings of XIV Brazilian Symposium onComputer Graphics and Image Processing pp.179 -182, 2001.

[7] [7]F.Hartung and B.Girod,”Watermarking of Uncompressed andCompressed Video,”Signal Processing,Vol. 66,N0. 3,pp.283-302,1998.

[8] [8]G.C.Langelaar,I.Setwan, and R.L.Lagendijk, “WatermarkingDigital Image and Video Data, “Signal Processing Magazine,Vol.17, No.5.pp.20-46,2000.

[9] [9] F. Hartung and B. Girod, “Watermarking of uncompressed andvariable code, and MClL for motion compensation. compressedvideo,” Signal Processing, vol.6, no.3, pp.283-301, 1998.

[10] [10] A. Alattar, E. Lin, and M. Celik, “Watermarking low bit-rateadvanced simple profile MPEG-4 bitstreams,” 2003 IEEEInternational Conference of Acoustics, Speech, and SignalProcessing(ICASSP2003) vol.3 pp.111-513—111-516, 2003.

[11] [11]D.Ghosh, and K. Ramakrishna, “Watermarking compressedvideo stream over Internet,” The 9th Asia-Pacific Conference onCommunications 2003 (APCC2003), vol.2, pp.711-715, 2003.

[12] [12] s. Sakazawa, Y. Takishima, and M. Wada, “A watermarkingmethod retrievable from MPEG compressed stream,” IEICE Trans.Fundamentals, vol.E85-A, no.11, 2002.

[13] [13] Zhi Li, Xiaowei Chen, “The Imperceptible Video WatermarkingBased on the Model of Entropy,” ICALIP 2008, 480~484.

[14] [14] Wang Pei, ZHENG Zhendong, YING Jun,” A NovelVideo Watermark in motion vector based on H.264 [A], ICALIP

2008[C], 2008:1555~1559.

It is a low power base station communicating in a licensedspectrum, offering improved indoor coverage with increasedperformance, improved voice and broadband services in lowcost with the operators approval.



7

NLM FOR IMPROVEMENT OF SENSINGCAPABILITY IN FEMTOCELLS

Dr S.Srinivasa Rao, Professor & Head , Department of ECE,

Malla Reddy College of Engineering & Technology, Secunderabad

Abstract: This paper presents the working of Network ListenModule (NLM) which is used to improve the sensingcapability of Femtocell device. Using NLM, the Femtocellwill be able to scan the air interface, detect neighboringcells and tune its network and RF parameters accordingly.

Keywords: NLM, Femtocell, DTX, STX, RSSI

1. INTRODUCTION

Femtocells are ‘domestic base stations’ which acts like cellularnetwork access points used to connect mobile devices tomobile network service providers using residential DSL, cablebroadband connections, optical fibers or wireless last miletechnology.

It is a low power base station communicating in a licensedspectrum, offering improved indoor coverage with increasedperformance, improved voice and broadband services in lowcost with the operators approval.

Figure 1. Femtocell house

A femtocell looks like a WiFi access point which containsRNC (Radio Network Controller) and all the core network

elements. It requires a data connection to the DSL or cable tothe Internet, through which it is connected to the mobilesoperator core network. The femtocell works through cellularnetwork provider and enhances connectivity for cellularphones, smart phones and other portable/mobile devicesespecially in thelocations where coverage by cellular systemsusing large cells is weak and discontinuous. Finally, themobile devices are connected to the backbone of the networksupplied by Internet service provider via femtocells.

2. NETWORK LISTEN MODULE

When a customer buys a femtocell, the network operatorprovides the customer with the femtocell device and afemtocell ID. This femtocell ID will be used to register andauthenticate the femtocell in the network after switching on.Moreover, when the customer buys the femtocell, he/she mustprovide some information to the operator. For example, theaddress where the femtocell is going to be installed and thelist of femtocell subscribers (registration data). Furthermore,in order to let the customer update the list of subscribers, theoperator also gives him/her a secure web site. It is to be notedthat, the list of authorized users resides in the core network.

After acquiring the femtocell, the customer only needs toplug the femtocell into a power source and Internet connectionto start using it. The customer cannot be assumed to have theknowledge to install or configure the femtocell, hence theseprocesses need to be automatic. Therefore, after power on,the first thing the femtocell does is to connect to the networkof the operator through the backhaul connection.

The femtocell is then authenticated and registered into thesystem as an operative device by the OSS, using the femtocellID. Afterwards, the femtocell can update its software bydownloading the latest available version from the OSS. Notethat this software update can also be triggered by the OSS atany time after power on. Subsequently, the femtocell verifiesthe functioning of this new software, self-testing theinstallation.



8

Fundamental information such as:

• frequency for DL and UL,

• scrambling code list, or

• radio channel bandwidth

must be provided during the booting procedure by theoperator over the backhaul link.

Network configuration parameters:

• location, routing and service area code information,

• neighbouring list,

• physical cell ID,

• RF parameters (pilot and maximum data power . . .)

can be automatically calculated from information on themacro cell layer provided by the operator (OSS data), andfrom information on the femtocell layer provided. By theusers (registration data). These data arrive at the femtocellthrough the backhaul link.

If the core network does not support this configuration orcannot supply any suggestions, the femtocell will derive theseparameters, using data gathered by monitoring the radiochannel. However, setting up the femtocell parameters from ablind configuration using only sensing techniques will delaythe booting procedure, and might result in an undesirableperformance. Therefore, it is advisable that a defaultconfiguration is provided by the femtocell firmware or throughthe backhaul.

The sensing of the radio environment is done by the networklistening mode, designed to scan the air interface. By decodingthe existing broadcast and control channels, the femtocellsynchronizes its internal oscillator and synchronizes thefemtocell to the external network. The information derivedfrom the initial sensing is also used to detect new neighbouringmacrocells and femtocells. In this way, the defaultconfiguration of the femtocell can be set up or reconfigured[22]. For example, the femtocell can add or remove new

Figure 2. Femtocell start up procedure

neighbouring relationships, select/re-select its physical cellID in order to minimize the collision probability, or tune itshandover parameters in order to facilitate the handoverprocedure towards other cells.

After the femtocell has been self-configured, the life cycle ofa femtocell moves towards a self-optimization loop, since thefemtocell needs dynamically to adapt its parameters to thechanging environment conditions.

Using the network listening mode and other inputs, e.g.broadcast messages, measurement reports, cognitive radio thefemtocell will collect statistics to optimize its performancedynamically (coverage and capacity). For example, in orderto provide an adequate signal quality to its users, and minimizethe impact (interference) on other cells, the femtocell willadapt its power and channel usage, as well as optimizing itsneighboring list and handover parameters according to thegathered information.

The most challenging environment is that of the home becausethe femtocell base station must be installable by the homeowner and the femtocell must be able to zero touch selfconfiguration to allow the femtocell to interoperate with theexisting Radio Access Network while causing the minimuminterference to the existing infrastructure.

From a network operators perspective the main requirementsfor the femtocell is to fit into the network with the minimumlevel of operator involvement in the process while minimizingthe impact of the femtocell on the existing network. In orderto do this the femtocell is required to boot up into a UE /network listen mode so that it can scan the air interface foravailable frequencies, scrambling codes and other networkresources.

A further complication for femtocell deployments is that theyare typically connected to the operators network through anIP connection and further they are located in doors so thefemtocell does not have access to any of the usual facilitiesfor providing timing and synchronization - for example theATM backhaul or GPS 1pps signal. The timing and frequencysynchronization requirements for modern radio networks arevery tight (typically 0.1 ppm) and while there are networktiming protocols such as NTP or IEEE1588 available, theseoften struggle to achieve the required accuracy.

In addition to the requirements of being a zero touch solutionfrom both the operator’s and the user’s perspective and havingto minimize the interference with the macro network and othermtocells; the femtocell also has to provide seamless hand-inand hand-out to the macro network.



9

2.1. Network Detection and Integration

During boot up it is imperative that the femtocell correctlydetects the surrounding network and integrates into it with theminimum interference. The startup procedure for a femtocellcan be summarized by the following four stages:

• Synchronizing the internal oscillator of the femtocell andsynchronize the femtocell to the System Frame Number(SFN) of the external network.

• Search the surrounding radio environment (including 2Gand 3G macro networks) for neighboring macro and femtocells. The search process detects the frequencies used,scrambling codes, CPICH Receive Signal Code Power(RSCP) and UTRA Receive Signal Strength Indicator(RSSI) required to minimize network interference andoptimize the transmit power.

• Decode the neighbor cell information to configure ahandout neighbor list and update the system database.

• Establishing the country code and location to ensure thedevice is being used within the terms of the operator’slicense.

In a UMTS network the cell search process must detect thefollowing information from the network:

1. P-SCH: slot sync

2. S-SCH: frame sync + SCG identification

3. P-CPICH I : PSC identification

Note stage 2 can be avoided in a warm search mode tosignificantly reduce the amount of time required to performthe search - for example when the femtocell is already awareof the Primary Synchronization Codes (PSC) in use on thenetwork.

2.2. Decoding The Synchronization Channel (SCH)

A UMTS network broadcasts the cell and system informationon the Broadcast Channel (BCH). This information is requiredby the femtocell before it can be integrated into the network.The Synchronization Channel (SCH) is transmitted by thebase stations and used by the UEs for cell search. There areactually two sub-channels on the SCH the Primary (P-SCH)and Secondary (S-SCH), each with a 10 ms frame length. The10 ms frames are divided into 15 slots, each of length 2560chips. This information is used in a five stage cell searchprocess, as follows:

• Search for the P-SCH and output the slot headerinformation

• Using the slot header information, S-SCH decode givesthe frame header and scrambling code group information

• The scrambling code group information allows thedetection of the primary scrambling code number

• Calculate the frequency offset of the base station by usingthe primary scrambling code number from 3

• Measure the Received Signal Code Power (RSCP) of thebase station by using the primary scrambling code numberfrom 3.

2.3. Timing And Synchronization

The femtocell is required to detect the time and frequencyfrom the radio network because it is not always possible to beable to achieve this over the network interface. In order to beable to achieve these timing requirements it is necessary touse a very accurate crystal oscillator that is typically voltageand temperature controlled and housed in an oven to controlthe temperature Crystal oscillators that meet these requirementsare typically much too expensive to be used in a consumerproduct so it is common to use cheaper crystals thatneed tobe re-conditioned on a regular basis.

The internal clock is responsible for:

• The accuracy of the absolute timing to ensure framealignment between receiver and transmitter and to avoidIntersymbol Interference (ISI)

• The spectrum accuracy to maintain frequency alignmentbetween the receiver and the emitter.

Clock Accuracy Requirements

The accuracy of a clock is usually measured in parts-per-billion (ppb) or parts-per million (ppm). These units representthe maximum variation obtained over a high number ofoscillations. For example a watch crystal has a typical errorof 20 ppm, giving a maximum error per day equal to 0.00002× 24 × 60 × 60 = 1.7 seconds.

In the 3GPP specifications, the requirements defined for aNodeB ask for a precision of 50 ppb. However in Release 6it has been relaxed to 100 ppb for indoor base stations, andlater in Release 8 is reduced to 200 ppb for Home nodeBwith certain standards. Some typical accuracy requirementsfor femtocells recommended by 3GPP are summarized in Table9.1. Even if it is reasonable for macrocell base stations toafford expensive and accurate oscillators, this is not the casefor FAP, which need to be manufactured at low prices.Therefore cheap and easily implementable solutions are stillrequired in this field.



10

3. OSCILLATORS FOR FEMTOCELLS

Piezoelectricity is the ability of certain materials (like quartz)to create an oscillating electrical potential when mechanicalpressure is applied. The resonance of this material can beused to create a signal oscillating at an accurate frequency.Cheap crystals usually have a precision of about 20 ppm.However, the main drawback of such material is that theoscillation frequency changes with the temperature.Furthermore these changes do not repeat exactly upontemperature variation, i.e. resonators exhibit an hysteresis inthe frequency variation. That is why in femtocells some moreadvanced oscillators must be used, in order to compensate forthe errors due to the variations in temperature.

3.1 Temperature Controlled Crystal Oscillators

A Temperature Controlled Crystal Oscillator (TCXO) is atype of oscillator that compensates for temperature changesto improve stability. In a TCXO, the signal from a temperaturesensor is used to generate a correction voltage that is appliedto a voltage-variable reactance, also called varactor. Thevaractor then produces a frequency change equal and oppositeto the frequency change produced by temperature.

TCXOs are used in many applications, which is why they arethe cheapest accurate oscillator components. Because whenusing a TCXO there are always delays between themeasurement of the temperature change and the generation ofthe frequency correction, the compensated frequency is notperfectly stable.

4. SYNCHRONIZATION VIA SENSING OF THENETWORK

To avoid using the backbone connection as a reference, agood approach could be to listen to neighboring cells, inparticular the surrounding macro cell. Indeed, the conditionfor low price is not requested by the macro cells, which iswhy they are equipped with accurate oscillators, and alsovery often with GPS receivers to synchronize them. This iswhy the timing accuracy is high in macro cells, and an efficientsynchronization solution would be for the FAP to listen to thenearest macro cell to synchronize its clock. If the clock is notaccurately synchronized it is possible that a subscriber, who

enters his home, will not be able to handover from one cellto another. Moreover, if the frequency shift of the femtocellis too high, it could happen that the mobile would not be ableto decode the different channels of the femtocell.

REFERENCES

1. 3GPP, FDD Home Node B (HNB) Radio Frequency (RF) requirements(FDD) TR 25.967, ver 8.0.1

2. 3GPP, Base Station (BS) radio transmission and reception (FDD), TS25.104, ver 8.8.0

3. 3GPP, Base Station (BS) conformance testing (FDD), TS 25.141, ver8.8.0

4. 3GPP, User Equipment (UE) radio transmission and reception (FDD),TS 25.101, ver 7.8.0

5. 3GPP, Physical layer Measurements (FDD), TS 25.215, ver 8.3.0

6. 3GPP, Physical layer procedures (FDD), TS 25.214, ver 8.6.0

7. Parth Amin, Olav Tirkkonen. Department of Communication andNetworking, Aalto University, Finland, Network Listening basedSynchronization techniques for Femtocell Systems.

8. John Edwards, picoChip Designs Ltd, Implementation Of NetworkListen Modem For WCDMA Femtocell.

9. Xiaochuan Xu Lab of Wireless Communication Systems andNetworks(WCSN) Beijing University of Posts andTelecommunications, Implementation of Network Listen for TD-SCDMA Femtocell.

10. Jie Zhang, Guillaume de la Roche, University of Bedfordshire, UK,FEMTOCELLS: TECHNOLOGIES AND DEPLOYMENT.

11. 3GPP, User Equipment (UE) procedures in idle mode and procedures

for cell reselection in connected mode TS 25.304 ver 8.7.0.

About the authors:

Dr S.Srinivasa Rao received his B.Techdegree in Electronics & ComunicationEngineering from Anna University,M.Tech from JNTU Hyderabad and PhDfrom JNTU Hyderabad. He has more than23 years of teaching experience andpublished more than 20 research papersin reputed National & International

Journals and conferences. He is Professor & Head inDepartment of Electronics Communication Engineering, MallaReddy college of Engineering & Technology, Hyderabad. Hisareas of interest are Wireless Communications, MobileCommunications & Adhoc Networks.



11

A Novel Digital Image Tamper Detection andRecovery by grouping Most Significant Bits

Dr. B. Vijay KumarProfessor, CSE,

Malla Reddy College of Engineering & Technology, Secunderabad

Abstract - Chan proposed an image authentication methodby producing the parity check bits from pixels whose bitshave been rearranged (in 2011) [6]. Due to thisrearrangement, the value of the most-significant bit ofeach tampered pixel can be determined according to itsparity check bits. With the help of the most-significant bitof the pixel, the pixel can be recovered by selecting twopossible (7, 4) Hamming code words. However, if thedistance between two Hamming code words is within acertain range, incorrect selection may occur. Chan’smethod added one additional bit to indicate the correctone. It is trivial that this may degrade the quality of theauthenticated image. This paper groups four most-significant bits into different groups to form a mappingcodebook and the mapping codebook is used to produceauthentication data instead of the (7, 4) Hamming codebook. The experimental results show that the proposedmethod has a greater ability to recover tampered areas.

1. INTRODUCTION

The most important property of digital media is that it iseasily tampered with. Consequently, media security andcopyright protection mechanisms have developed quickly, andmany image authentication schemes have been developed.Among these various image authentication schemes, some havebeen proposed to embed recovered data into the digital images[1]–[4] so that the tampered areas can be detected andrecovered. It is trivial that the fineness of the recovered areasis related to the recovery unit of the different methods.However, the recovery unit of Yang and Shen’s method [4] isjust a block with 4×4 pixels, and that of Lee and Lin’s method[2] is just a block with 2×2 pixels.

In order to improve the fineness of the recovered areas, Chanand Chang’s method [1] produced parity check bits from eachpixel through a hamming code technique [5]. The recoveryunit of Chan and Chang’s method becomes one pixel. AlthoughChan and Chang’s method can reduce the recovery unit toone pixel, there are some drawbacks in their method. Chanand Chang’s method must predict the most-significant bit of

each tampered pixel first. The tampered pixel then can berecovered by referring to the predicted bit and the extractedparity check bits. However, once an incorrect prediction ismade in the recovery procedure, the tampered pixel cannot berecovered successfully. Furthermore, a pixel with an incorrectprediction may affect the prediction accuracy of the followingpixels. On the contrary, in Chan’s method [6], the parity checkbits are produced from pixels whose bits have been rearranged.The value of the most-significant bit of each tampered pixelcan be determined according to its parity check bits. Therefore,the recovery procedure of the proposed method does not needto predict the most-significant bit of the pixel. Although thevalue of the most-significant bit can be known through theparity check bits, the method still has to predict pixels byselecting one from two candidates of (7, 4) Hamming codewords. However, if the distance between two Hamming codewords is within a certain range, there is still the possibility ofmaking incorrect predictions. In the new version of Chan’smethod in [6], one additional bit was added to record thecorrect one from two candidates.

In this paper, we group four most-significant bits into differentgroups to form a mapping codebook. The mapping codebookhas an important property in that the distancebetween twocandidates in each group is always out of the mentioned range.The proposed method will produce authentication data througha mapping codebook instead of a (7, 4) hamming codebook.The experimental results show that the proposed method hasbetter ability to recover the tampered areas with good quality.

The rest of this paper is organized as follows. Chan’s methodis reviewed in Section 2. The method proposed in this paperwill then be presented in Section 3. In Section 4, theexperimental results are offered to demonstrate theeffectiveness of the proposed method. Finally, the conclusionswill be made in Section 5.

2. RELATED WORK

In this section, Chan’s method [6] is reviewed. Chan’s methodcontains three procedures: the embedding procedure, the



12

detection procedure, and the recovery procedure. Theembedding procedure produces parity check bits from therearranged bits of the pixel, and the parity heck bits areembedded into another pixel by using a modulus function [7].In the detection procedure, the extracted parity check bits areused to check whether the pixel has been tampered with.According to the indication of the tampered pixels, thetampered areas are also located in this procedure. The finalprocedure, the recovery procedure, recovers the tamperedareas. The details of Chan’s method are described as follows.

2.1 Embedding Procedure

The embedding procedure contains three steps: Hamming codeproduction, bit rotation, and Torus automorphism [8]. In thefirst step, Hamming codeproduction, parity check bits areproduced from therearranged four most-significant bits of eachpixel. Therelation between rearranged bits and parity checkbits isshown in Fig. 1. The values of the parity check bits (P1,P2,P3) can be decided by achieving the goal that the numberof”1" in each circle should be even. According to theexamplein Fig. 1, the values of the parity check bits P1, P2,and P3 are 1, 0, and 0, respectively.

It should be noticed that the parity check bits aregeneratedfrom the reversed four most-significant bits. Thereason whywe reverse the bits is that by referring to Fig. 2,the mostsignificant bits of two original data bits are thesame if thevalues of their parity check bits are the same.

Fig.1. Way to generate the parity check bits from thereversed data bits

That means whether the value of the original data bits islargerthan 128 or not can be decided according the value oftheparity check bits.In the second step the produced parity checkbits arerotated. The purpose of bit rotation is to rotate theorder ofthe parity check bits to increase security. This stepfirstselects one secret key k1 as a seed to generate a sequenceofrandom numbers, R1, R2, …,RN×N, where N is thepixelnumber of the height and width of the cover image. For

theith pixel, its three parity check bits must be rotatedaccordingto the random number Rias follows.

J’ = (J + Ri) mod3 (1)

whereJ represents the original order of the parity check bits,andJdenotes the new order of the parity check bits. ThevariablesJ and J2 are both in the range from 0 to 2. Aftergoing through(1), the bit at the Jth position will be rotated to the Jth position.In the third step, the rotated parity check bits areembeddedinto the three least-significant bits of anotherpixel indicatedby the Torus auto-morphism. The formula of the Torus auto-morphism is shown as follows.

Where the variable k2 represents the second key. The symbol(xi, yi) denotes the position at where the ith pixel P

i is located.

On the other hand, the position (xi2, yi2)represents the newposition in which the parity check bits ofPi are going to beembedded and the symbol N is the pixelnumber of the heightand the width of the cover image.Once the embedded positionfor each Pi is known, theparity check bits can be embeddedby using a modulus

Function [7]. Finally, the authenticated image can beobtained.

2.2 Detection and Recovery Procedures

The purpose of the detection procedure is to locate thetamperedareas so that the recovery procedure has targetareas to recover.In the detection procedure, the paritycheck bits for the pixelPi at (xi, yi) can be extracted fromthe pixel at (xi’, yi’)according to (2). Meantime, the paritycheck bits of the pixelPi can be produced through the fourmost-significant bits ofthe pixel Pi. Once the extractedparity check bits are not thesame as the produced bits, both positions (xi, yi) and (xi’, yi’)are marked as the tampered pixels. After checking all pixels,morphological operations[9] are performed to eliminate theisolated faulty judgments. Finally, the tampered areas can belocated through the detection procedure.

In the recovery procedure, if the parity check bits of thetampered pixel are not modified, these bits can be usedtopredict the value of the tampered pixel. According to thevalue of the parity check bits, the value of the most-significantbit, d4, can be known by referring to Fig. 2. Moreover, thetwo candidates of the four most-significantbits can also beknown according to Fig. 2. The value of the most-significantbit of the pixel can help us to select thecorrect one from twocandidates so as to recover the tampered pixel.



13

More precisely, in the detection procedure, the tampered areahas been located. Chan’s method uses an indicated matrix Mto indicate the number of the surrounding un tampered pixelsfor each tampered pixels. It is obvious that the tampered pixelswith a larger number at the corresponding positions ofindicated matrix M have a better opportunity to select a correctone from two candidates because they have more un tamperedpixels for references. Therefore, only the pixels with the largestnumber of un tampered pixels will be processed in each round.Because the most-significant bit of the predicted pixel can beknown by referring to the parity check bits, the un tamperedpixels whose most-significant bit is the same as the predictedpixel are gathered to calculate their average value.

In this section, the experimental results are demonstrated. Tobegin with, the test image, Lena, with 512×512 pixels, isshown in Fig. 5. We first show the authenticated images andtheir PSNR values in Fig. 6. Because both Chan’s methodand the proposed method use a modulus function to embeddata instead of least

The candidate that has the minimal distance with the averagevalue is selected to recover this tampered pixel. After that,the recovered pixels are marked as un tampered pixels, andthe same procedures are performed to recover tampered pixelsuntil all tampered pixels are marked as untampered pixels.The recovery procedure is shown in Fig. 3.

3. PROPOSED METHOD

In Chan’s method[6], although the value of the most-significantbit can be known through the parity check bits, the methodstill has to predict pixels by selecting one from two candidates.Most pixels can be predicted correctly by using the methoddescribed in Section 2. However, there is still the possibilityof making incorrect predictions while two candidates are tooclose. In fact, if the distance between two candidates is smallenough, selecting any one of two candidates to recover thepixel is acceptable. The quality of the recovered pixel is stillgood. However, by referring to the last column in Fig. 2, itcan be seen that the distance between two candidates becomeslarger when the values of the parity check bits are 2, 5, 1 and6. Their distances are 3 and 5. In the new version of Chan’smethod in [6], one additional bit is added to record the correctone from two candidates whose parity check bits are 2, 5, 1,and 6. It is trivial that this may degrade the quality of theauthenticated image.

Fig. 2. Original data bits and the parity check bits Fig. 3. Recovery procedure



14

In order to improve the quality, we propose grouping fourmost-significant bits into different groups to form a mappingcodebook. There are two characteristics of the mappingcodebooks. The data bits with same MSB are grouped together.Thus the value of difference between the bits will be 8. Thiswill be useful while retrieving the tampered pixels as themost significant bit can be easily found out. This is shown infigure4.

In the embedding procedure, the proposed method is almostthe same as Chan’s method except for the Hammingcodeproduction step. Chan’s method obtained authentication datafrom four most-significant bits byproducing parity check bits.On the contrary, our proposed method obtains authenticationdata from four most significant bits through the MappingCodebook. The mapping value can be decided by referring tothe fourmost-significant bits and the Mapping Codebook. Therole of the mapping value is treated as the parity bits in Chan’smethod. This means that the mapping value will go throughthe steps of bit rotation and the rotated bits are embedded tothe three least-significant bits of another pixel indicated byTorus automorphism.

4. EXPERIMENTAL RESULTS

Significant bit (LSB) replacement, the peak signal-to-noiseratio (PSNR) values of Chan’s method and the proposedmethod are higher than that of Chan and Chang’s method.

Next, we show the recovery ability for different methods indifferent areas. There are three areas, including a non-detailedarea, a detailed area and a complex area (which includes anon-detailed area and a detailed area), used in our experimentalprocedures. These areas are shown in Figs. 7 (a), (b) and (c).The tampered areas are shown in Figs. 7 (d), (e), and (f). Thesize of the tampered area is 64×64 pixels.

Fig. 4. Mapping Codebook.

Fig. 5. Test image.

Fig. 6. Authenticated images: (a) Chan and Chang’s method(PSNR: 38.57), (b) Chan’s method (PSNR: 40.02), and

(c) proposed method (PSNR: 40.07).

Fig, 7. Experimental areas: (a) to (c): original areas, and (d) to(f): tampered areas.



15

In Fig. 8 we demonstrate the experimental results of the relatedmethods and our proposed method. Although the recoveredareas of Chan and Chang’s method shown in Fig. 8 (a) and(b) have good results, the pixels in the recovered area in Fig.8 (c) are different to the original area. On the other hand,Chan’s method [6] and the proposed method can recover thetampered areas successfully. Comparing the experimentalresults of the proposed method with those of Chan’s method,we can see that the recovered areas, by using our proposedmethod, have better image quality. Because the distance oftwo candidates for some Hamming code words is in thementioned range, it is hardfor Chan’s method to judge whichone is correct. Therefore, the recovered area in Fig. 8 (e) isnot very good. On thecontrary, because the distance of thetwo candidates of each group is out of the mentioned range,the recovered area in Fig. 8 (h) is better than that in Fig. 8

Fig. 8. Results of the recovered areas: (a) to (c): Chan andChang’s method, (d) to (f): Chan’s method, and (g) to (i):

proposed method.

(e). It is trivial that the recovered areas of the proposed methodare better than those of Chan’s method.

5. CONCLUSIONS

In this paper, we group four most-significant bits into differentgroups to form a mapping codebook. The mappingcodebookis used to replace the role of (7, 4) Hamming code book inChan’s method. The mapping codebook hastwo importantproperties. First, only the candidates with the same value ofthe most significant bit can be gathered in the same group.This means the value of the most-significant bit of the predictedpixel can be determined according to its mapping values.Second, the distance between two candidates of each group isout of the range from 3 to 5. This means one additional bitfor some pixels is not necessary in our method. According tothe experimental results, the proposed method has a betterability to recover the tampered areas with good quality.

REFERENCES

[1] C.-S. Chan and C.-C. Chang, “An efficient image authenticationmethod based on hamming code,” PatternRecognition, vol. 40, no.2, pp. 681"690, 2007.

[2] T.-Y. Lee and S.-D.Lin, “Dual watermark for image tamper detectionand recovery,” Pattern Recognition, vol. 41, no. 11, pp. 3497"3506,2008.

[3] S.-S. Wang and S.-L.Tsai, “Automatic image authentication andrecovery using fractal code embedding and image in painting,” PatternRecognition, vol. 41, no. 2, pp. 701"712, 2008.

[4] C.-W. Yang and J.-J.Shen, “Recover the tampered image based onVQ indexing,” Signal Processing, vol. 90, no. 2, pp. 331"343, 2010.

[5] R. W. Hamming, “Error detecting and error correcting codes,” TheBell System Technical Journal, vol. 26, no. 2, pp. 147"160, 1950.

[6] C.-S. Chan, “An image authentication method by applying hammingcode on rearranged bits,” Pattern RecognitionLetter, vol. 32, no. 14,pp. 1679"1690, 2011.

[7] C.-C. Thien and J.-C. Lin, “A simple and high-hiding capacity methodfor hiding digit-by-digit data in images based on modulus function,”Pattern Recognition, vol. 36, no. 12, pp. 2875"2881, 2003.

[8] G. Voyatzis and I. Pitas, “Chaotic mixing of digital images andapplications to watermarking,” in Proc. of EuropeanConf. onMultimedia Applications, Louvain-la-Neuve, Belgium, 1996, pp.687"695.

[9] J. Serra, Image Analysis and Mathematical Morphology, New York:

Academic Press, 1982.



16

Verification of GPIO Core Functions using UniversalVerification Methodology

K. Niranjan Reddy1, U.DhanaLakshmi2, Dr. PVY Jaya Sree3

[email protected], [email protected], [email protected]

Professor & HOD, ECE1, Asst Professor2, Assoc Professor3

Malla Reddy Engineering College for Women, Hyderabad, Telangana State, India1,2

Gitam Institute of Technology, Gitam University, Visakhapatnam, AP, India3

Abstract—The OPB GPIO design provides a general purposeinput/output interface to a 32-bit On-Chip Peripheral Bus(OPB). The GPIO IP core is user-programmable general-purpose I/O controller. That is use is to implement functionsthat are not implemented with the dedicated controllers in asystem and require simple input and/or output softwarecontrolled signals. It is one of the important peripheral thatis listed on any FPGA board. In this project we are atomizingthe operation of the GPIO by writing the code in SYSTEM-VERILOG and simulating it in QUESTA MODELSIM. Themain aim of this project is to verify the output by using GPIOpins depending up on the preference the code. We verify theGPIO modules by using UVM [Universal verificationMethodology]. The functional verification of the RTL designof the GPIO is carried out for the better optimum design.

Index Terms— GPIO,OPB,QUESTA MODELSIM, SystemVerilog, FPGA.

I. INTRODUCTION

The GPIO module is part of Inicore’s IP module family. Thisgeneral purpose input/output controller provides some uniquefeatures that eases system integration and use. Each GPIO portcan be configured for input, output or bypass mode. All outputdata can be set in one access. Single or multiples bits can be setor cleared independently. Every GPIO port can serve as aninterrupt source and has its own configuration options: • Levelsensitive, single edge triggered or level change • Active high orlow respectively rising edge or falling edge • Individual interruptenable register and status flags The core provides severalsynthesis options to ease the system integration and minimizethe gate count: • Selectable CPU bus width: default options are8/16/32-bit • Selectable number of GPIO ports • CPU read backenable.

II. GPIO(GENERAL PURPOSE I/O)

GPIO is a generic pin on a chip whose behavior (includingwhether it is an input or output pin) can be controlled

(programmed) through software. GPIO pins have no specialpurpose defined, and unused by default. The idea is thatsometimes the system integrator building a full system that usesthe chip might find useful to have a handful of additional digitalcontrol lines, and having these available from the chip can savethe hassle of having to arrange additional circuitry to providethem. For example, the Realtek ALC260 chips (audio codec)have 4 GPIO pins, which go unused by default. Some systemintegrators (Acer laptops) employing the ALC260 use the firstGPIO (GPIO0) to turn on the amplifier used for the laptop’sinternal speakers and external headphone jack.

A. Architecture of GPIO

Fig. 1 Architecture of GP I/O

i. Clocks: The GPIO core has two clock domains. All registersexcept RGPIO_IN are in system clock domain. RGPIO_INregister can be clocked by system clock or by external clockreference.

ii. APB Interface: The host interface is implemented using a 32bit APB compliant slave interface.

iii. GPIO Registers



17

The GPIO IP Core has several software accessible registers.Most registers have the same width as number of general-purpose I/O signals and they can be from 1 – 32 bits. The hostthrough these registers programs type and operation of eachgeneral-purpose I/O signal or three-state outputs, appropriateopen-drain or three-state I/O cells must be used. Part of externalinterface is also ECLK register. It can be used to register inputsbased on external clock reference. General-purpose inputs cangenerate interrupts so that software does not have to be in pollmode all the time when sampling inputs. Switching outputdrivers into open-drain or three-state mode will disable general-purpose outputs. To lower number of pins of the chip, other on-chip peripherals can be multiplexed together with the GPIOpins. For this purpose, auxiliary inputs can be multiplexed ongeneral-purpose outputs.

iv. Hardware Reset:

Following hardware reset all general-purpose I/O signals areset into input mode. Meaning, all output drivers are disabled.All interrupts are masked, so that inputs would not generateany spurious interrupts. Gpio_eclk signal is not used to latchinputs into RGPIO_IN register; instead system clock is usedGeneral-Purpose I/O as Polled Input. To use general-purpose I/O as input only, corresponding bit in RGPIO_OE register mustbe cleared to select input mode. Bit RGPIO_CTRL[INTE] andcorresponding bit in RGPIO_INTE register must be cleared aswell ,to disabled generation of interrupts. Bit RGPIO_IN registerreflects registered value of general-purpose input signal.RGPIO_IN is updated on positive edge of system clock or ifRGPIO_ECLK appropriate bit is set, on gpio_eclk edge. Whichclock edge is selected is defined by value of RGPIO_NECappropriate bit.

v. General-Purpose I/O as Input in Interrupt Mode:

To use general-purpose I/O as input with generation ofinterrupts, corresponding bit in RGPIO_OE register must becleared to select input mode. Corresponding bit inRGPIO_PTRIG register must be set to generate an interrupt onpositive edge event on general-purpose input. To generate aninterrupt on negative edge event, corresponding bit inRGPIO_PTRIG register must be cleared. If we are enablinginterrupts for the first time, we also need to clear interrupt statusregister RGPIO_INTS. Last, RGPIO_CTRL[INTE]bit andcorresponding bit in RGPIO_INTE register must be set to enablegeneration of interrupts. Bit RGPIO_IN register reflectsregistered value of general-purpose input signal. RGPIO_IN isupdated on positive edge of system clock or if RGPIO_ECLKappropriate bit is set, on gpio_eclk edge. Which clock edge isselected, is defined by value of RGPIO_NEC appropriate bit.Which input caused an interrupt is recorded in interrupt statusregister RGPIO_INTS. Inputs that caused an interrupt since last

clearing of RGPIO_INTS have bits set. Interrupt can be de-asserted by writing zero in RGPIO_INTS register and controlregister bit RGPIO_CTRL[INTS]. Another way to de-assertinterrupts is to disable them by clearing control bitRGPIO_CTRL[INTE].

vi .General-Purpose I/O as Output

To enable general-purpose I/O output driver, corresponding bitin RGPIO_OE must reset. Corresponding bit in RGPIO_OUTregister must be set to the value that is required to be driven onoutput driver. Corresponding bit in RGPIO_INTE register mustbe cleared to disable generation of spurious interrupts. Clearingbit in RGPIO_OE register will disable output driver and enablethree-state or open-drain. General-Purpose I/O as Bi-DirectionalI/O. To use general-purpose I/O as bi-directional signal,corresponding bit in RGPIO_OE must be toggled to enable ordisable three-state or open-drain mode of bi-directional driver.Corresponding bit in RGPIO_OUT register must be set to thevalue that is required to be driven on output driver.Corresponding bit in RGPIO_INTE register must be cleared todisable generation of spurious interrupts. If input shouldgenerate interrupts, corresponding bit in RGPIO_INTE registermust be set and if required also corresponding bit inRGPIO_PTRIG should be set. Corresponding bit RGPIO_INregister reflects registered value of general-purpose input signal.RGPIO_IN is updated on positive edge of system clock or ifRGPIO_ECLK bit is set, on gpio_eclk edge. Which clock edgeis selected, is defined by value of RGPIO_NEC bit. If aninterrupt is enabled and pending, it can be de-asserted by writingzero in RGPIO_INTS register and control register bitRGPIO_CTRL[INTS]. Another way to dessert interrupts is todisable them by clearing control bit RGPIO_CTRL[INTE]General-Purpose I/O driven by Auxiliary Input To drivegeneral-purpose output with auxiliary input, corresponding bitin RGPIO_OE must be set to enable output driver.Corresponding bit in RGPIO_AUX must be set to enablemultiplexing of auxiliary input onto general-purpose output.

III. UNIVERSAL VERIFICATION METHODOLOGY

The UVM (Universal Verification Methodology) was introducedin December 2009, by a technical Sub committee of Accellera.UVM uses Open Verification Methodology as its foundation.Accellera released version UVM 1.0 EA on May 17, 2010.UVMClass Library provides the building blocks needed to quicklydevelop well-constructed and reusable verification componentsand test environments. It uses system Verilog as its language.All three of the simulation vendors (Synopsys, Cadence andMentor) support UVM today which was not the case with otherverification methodology. Today, more and more logic is beingintegrated on the single chip so verification of it is a verychallenging task. More than 70 percent of the time is spent on



18

the verification of the chip. So it is a need of an hour to have acommon verification methodology that provide the base classesand framework to construct robust and reusable verificationenvironment. UVM provides that.

In this paper , all the terminology related to UVM is introducedalong with the sample example. In first phase uvm componentsare introduced. In second phase some of the features related toUVM are introduced and in final phase small environment isbuilt using UVM from the scratch.

2. Test Bench architecture

Driver (BFM)

Driver as the name suggest, drive the dut signals. It basicallyreceives the transaction object from the sequencer and convertsit in to the pin level activity. So for example it can generateread or write signal, write address and data to be transferred. Itis the active part of the verification logic.

Sequencer

Sequencer is the component on which the sequences will run.The dut needs to be applied a sequence of transaction to test itsbehavior. So sequence of transaction is generated and it isapplied to driver whenever it demands by the sequencer.

Monitor

A monitor is the passive element of the verification environment.It just sample the dut signal from the interface but does notdrive them. It collect the pin information , package it in form ofa packet and then transfer it to scoreboard or other componentsfor coverage information.

Agent

Agent is basically a container. It contains driver, monitor andsequencer. Driver and sequencer are connected in agent. Agenthas two modes of operation: passive and active. In active modeit drives the signal to the dut. So driver and sequencer areinstantiated in active mode. In passive mode it just sample thedut signals does not drive them. So only monitor is instantiatedin passive mode. Normally there is one agent per interface likeAHB or APB.

Scoreboard

Scoreboard is a verification component that checks the responsefrom the dut against the expected response. So it keeps track ofhow many times the response matched with the expectedresponse and how many time it failed.

Environment

Environment is at the top of the test bench architecture, it willcontain one or more agents depend on design. If more than oneagents are there then it will be connected in this component.Agentsare also connected to other components like scoreboardin this component.

The following subsections describe the components of averification component.

• Data Item (Transaction)

• Driver (BFM)

• Sequencer

• Monitor

• Agent

• Environment

Data Item (Transaction)

Data item are basically the input to the device under test. Allthe transfer done between different verification components inUVM is done through transaction object. Networking packets,instructions for processor are some examples of transactions.From the top level test many data items are generated and appliedto the dut so by intelligently randomizing the data items objectwe can check corner cases and maximize the coverage on thedevice under test.

Fig. 1 Test bench architecture



19

Functional Verification of GPIO Core Using UVM

As verification methodology plays a important phase in thecircuit design. The read operation of the GPIO is carried out inXILINX for RTL design and the verification methodology iscarried out using Questasim 10.0b. The design is carried outusing in HDL and the verification is carried out in UVM. TheGPIO is set up as DUT for the functional verification and thecode coverage is determined using Modelsim is obtained for100%

V. CONCLUSION

In this we have verified the GPIO core based on UVM techniqueusing Questasim simulator and Modelsim. The code coverageis obtained for the RTL design and 100% code coverage isextracted. This methodolgy provides the complete coverage ofthe RTL design so as to acquire the fault free Protocol designof GPIO. So that can be implemented in real time systems. Thiscan be further implemented for the ASIC implementation andSOC Applications.

REFERENCES

[1] D.Gajski et al, “Essential Issues for IP Reuse”, Proceedings of ASP-DAC, pp.37-42, Jan. 2000

[2] C.K.Lennard et al, “Industrially proving the SPIRIT ConsortiumSpecifications for Design Chain Integration”, Proceedings of DATE2006, pp. 1-6, March 2006

[3] K.Cho et al, “Reusable Platform Design Methodology For SOCIntegration And Verification”, Proceedings of ISOCC 2008, pp. I-78-I-81, Nov. 2008

[4] W.Kruijtzer et al, “Industrial IP integration flows based on IP-XACTstandards” proceedings of DATE 2008, pp. 32-37, March 2008

[5] M.Strik et al, “subsystem Exchange in a Concurrent Design ProcessEnvironment” Proceedings of DATE 2008, pp. 953-958, March 2008

[6] GensysIO, http://www.atrenta.com/solutions/gensys-family/gensys-io.htm

[7] SocratesSpinner

IV. RESULTS AND VERIFIACTION

The GP I/O is carried out for the functional verification usingthe UVM technique for both the read and write operation. Thefunctional verification is of the RTL design is of the GPIO isyields the complete code coverage.

Fig 3 simulation showing GPIO Functional Verificationfor read operation

Fig 4 simulationresults for GPIO functional verificationfor write operation

ModelSim Coverage Report

Fig 5 Functional code coverage of GPIO



20

Intelligent Wireless Embedded System for VehicleControl in Transportation System Based on GSM

and GPS TechnologyS.Srivani1, M.Haritha2, K. Sumalatha3

Assoc Professor1, Assoc Professor2, Asst Professor3

[email protected], [email protected], [email protected]

Malla Reddy Engineering College for Women, Hyderabad, Telangana State, India

Abstract— Currently almost of the public having an ownvehicle, theft is happening on parking and sometimes drivinginsecurity places. The safe of vehicles is extremely essentialfor public vehicles. Vehicle security and accident preventionis more challenging. So in order to bring a solution for thisproblem this system can be implemented. Vehicle securityenhancement and accident prevention system can bedeveloped through the application of ignition control (trackingand locking), fuel theft, accident detection and prevention,driver fatigue, pollution control and speed limiting withefficient vehicle management system. The need for this projectis to provide security to the vehicles by engine locking systemwhich prevents the vehicle from unauthorised access. Thistechnique helps to find out the exact location of the accidentand with the help of server an emergency vehicle can be sentto the exact location to reduce the human life loss. It alsodetects the behaviour of the driver through sensors whetherhe/she is drowsy or drunk, so that occurrence of accident canbe prevented. The place of the vehicle identified using GlobalPositioning system (GPS) and Global system mobilecommunication (GSM). This is more secured, reliable andlow cost.Index Terms— Vehicle Tracking, Locking, embeddedSystem, GPS, GSM.

I. INTRODUCTION

The GPS/GSM based System is one of the most importantsystems, which integrate both GSM and GPS technologies. It isnecessary due to the many of applications of both GSM andGPS systems and the wide usage of them by millions of peoplethrough out the world [1]. This system designed for users inland construction and transport business, provides real-timeinformation such as location, speed and expected arrival timeof the user is moving vehicles in a concise and easy-to-readformat. This system may also useful for communication processamong the two points.

Currently GPS vehicle tracking ensures their safety as travelling.

This vehicle tracking system found in clients vehicles as a theftprevention and rescue device. Vehicle owner or Police followthe signal emitted by the tracking system to locate a robbedvehicle in parallel the stolen vehicle engine speed going todecreased and pushed to off. After switch of the engine, motorcannot restart without permission of password. This systeminstalled for the four wheelers, Vehicle tracking usually used innavy operators for navy management functions, routing, sendoff, on board information and security. The applications includemonitoring driving performance of a parent with a teen driver.Vehicle tracking systems accepted in consumer vehicles as atheft prevention and retrieval device. If the theft identified, thesystem sends the SMS to the vehicle owner. After that vehicleowner sends the SMS to the controller, issue the necessarysignals to stop the motor.

In this paper, proposed method is presented in section 2 andrelated technology in section 3. The sensors used in the vehicletracking and locking System are described in section 4 andsection 5 gives the conclusion.

II. PROPOSED METHOD

In this proposed work, a novel method of vehicle tracking andlocking system used to track the theft vehicle by using GPS andGSM technology. This system puts into sleeping mode whilethe vehicle handled by the owner or authorized person otherwisegoes to active mode, the mode of operation changed by in personor remotely. If any interruption occurred in any side of the door,then the IR sensor senses the signals and SMS sends to themicrocontroller. The controller issues the message about theplace of the vehicle to the car owner or authorized person. Whensend SMS to the controller, issues the control signals to theengine motor. Engine motor speeds are gradually decreases andcome to the off place. After that all the doors locked. To openthe door or restart the engine, authorized person needs to enterthe passwords. In this method, tracking of vehicle place easyand doors locked automatically, thereby thief cannot get awayfrom the car.



21

• Accident location and vehicle theft identification involvesvehicle tracking using GPS technology.

• Vehicle security is enhanced by ignition control system.

• Anti-vehicle theft using engine locking system.

• Fuel theft can be prevented by monitoring the fuel level infuel tank.

• Accident identification system provides the location at whichaccident occurs.

• Driver fatigue system provides information about driversheart beat rate, eye blink rate which prevent accident becauseof the drowsiness or drunk and drive by the driver.

• Obstacle detection system helps in stopping the vehicle whenan obstacle is detected using IR sensor and pollutiondetection helps in controlling pollution from vehicle usingCO sensor.

• GSM and GPS is used for tracking the location of vehicleand for providing Short Message Service (SMS).

III. TECHNOLOGY

A. GPS Technology

The Global Positioning System (GPS) is a satellite-basednavigation system consists of a network of 24 satellites locatedinto orbit. The system provides essential information to military,civil and commercial users around the world and which is freelyaccessible to anyone with a GPS receiver. GPS works in anyweather circumstances at anywhere in the world.

A GPS receiver must be locked on to the signal of at least threesatellites to estimate 2D position (latitude and longitude) andtrack movement. With four or more satellites in sight, thereceiver can determine the user’s 3D position (latitude, longitudeand altitude). Once the vehicle position has been determined,the GPS unit can determine other information like, speed,distance to destination, time and other. GPS receiver is used forthis research work to detect the vehicle location and provideinformation to responsible person through GSM technology.

Fig.1. Block diagram of Vehicle tracking and locking system basedon GSM and GPS

B. GSM Modem SIM300 V7.03

The GSM modem is a specialized type of modem which acceptsa SIM card operates on a subscriber’s mobile number over anetwork, just like a cellular phone. It is a cell phone withoutdisplay. Modem sim300 is a triband GSM/GPRS engine thatworks on EGSM900MHz, DCS1800MHz and PCS1900MHzfrequencies. GSM Modem is RS232-logic level compatible, i.e.,it takes -3v to -15v as logic high and +3v to +15 as logiclow.MAX232 is used to convert TTL into RS232 logic levelconverter used between the microcontroller and the GSM board.The signal at pin 11 of the microcontroller is sent to the GSMmodem through pin 11 of max232. This signal is received atpin 2 (RX) of the GSM modem. The GSM modem transmitsthe signal from pin3 (TX) to the microcontroller throughMAX232, which is received at pin 10 of IC1 [9].

Fig.2. GPS module



22

IV. SENSORS

A. Eyeblink Sensorm

Vehicle accidents are most common if the driving is inadequate.These happen on most factors if the driver is drowsy or if he isalcoholic. If sensor detects that the driver is unconscious whiledriving in traffic or at the middle of the road, the vehicle isexpected to detect the edge of the road and then stop. If it doesso the vehicle may not disturb the other vehicles.

B. IR Sensor

IR Transmitter is an LED which emits infrared rays. IR Receiveris used to receive the IR rays. Both IR transmitter and receivershould be placed straight line to each other.The transmittedsignal is given to IR transmitter whenever the signal is high, theIR transmitter LED is conducting and it passes the IR rays tothe receiver. The IR receiver is connected with comparator.

C. Gas Sensor

The system has two modules namely the Gas sensing moduleand the Obstacle detection module and they are interfaced withthe microcontroller.

Obstacle Sensing Module

The obstacle sensing module is used to sense such that, accidentsdue to unwanted parking of the vehicles and collision with treesand other objects especially during the night time could beavoided. These obstacles could be detected using variousmethods such as ultrasonic sensors.

Gas sensing module

The gas sensing module is used to sense the presence of toxicgases such as CO, LPG, Alcohol and other toxic gases insidethe vehicle. If critical levels of gases were found, the CO exceeds20ppm and the level of LPG exceeds 10,000ppm and the

Fig.3. GSM module

presence of alcohol is detected then the digital data from thegas sensing module is sent to the microcontroller which displaysthe information about the gas leakage inside the vehicle andproduces an alarm to alert the persons inside the vehicle. It alsosends a text message to the authorized person through GSMmodem connected to the microcontroller such that remedymeasures could be taken by the authorized person and to giveproper medical treatment to them if required.

D. Float Level Sensor

Level sensors detect the level of substances that flow, includingliquids, slurries, granular and materials. Fluids and fluidizedsolids flow to become essentially level in their containers (orother physical boundaries) because of gravity whereas most bulksolids pile at an angle of repose to a peak. The substance to bemeasured can be inside a container or can be in its natural form(e.g., a river or a lake). The level measurement can be eithercontinuous or point values. Continuous level sensors measurelevel within a specified range and determine the exact amountof substance in a certain place, while point-level sensors onlyindicate whether the substance is above or below the sensingpoint. Generally the latter detect levels that are excessively highor low. There are many physical and application variables thataffect the selection of the optimal level monitoring method forindustrial and commercial processes. The selection criteriainclude the physical: phase (liquid, solid or slurry), temperature,pressure or vacuum, chemistry, dielectric constant of medium,density (specific gravity) of medium, agitation (action),acoustical or electrical noise, vibration, mechanical shock, tankor bin size and shape. Also important are the applicationconstraints: price, accuracy, appearance, response rate, ease ofcalibration or programming, physical size and mounting of theinstrument, monitoring or control of continuous or discrete(point) levels.

E. Heart Beat Sensor

The circuit is designed to measure the heart rate. IR transmitterand receiver measure the heart rate. Infrared transmitter is onetype of LED, which emits infrared rays generally called as IRTransmitter. Similarly IR Receiver is used to receive the IRrays transmitted by the IR transmitter. One important point isboth IR transmitter and receiver should be placed straight lineto each other. The IR transmitter and receiver are placed in thepulse rate sensor. When we want to measure the pulse rate, thepulse rate sensor has to be clipped in the finger. The IR receiveris connected to the Vcc through the resistor which acts aspotential divider. The potential divider output is connected toamplifier section. When supply is ON the IR transmitter passesthe rays to the receiver. Depending on the blood flow, the IRrays are interrupted. Due to that IR receiver conduction isinterrupted so variable pulse signals are generated in thepotential divider point which is given to A1 amplifier through



23

the capacitor C1. The coupling capacitor C1 is used to blockthe DC component because the capacitor reactance is dependson the frequency. For DC component the frequency is zero sothe reactance is infinity now capacitor acts as open circuit forDC component.

F. Ultrasonic Sensor

Ultrasonic sensors (also known as transceivers when they bothsend and receive) work on a principle similar to radar or sonarwhich evaluate attributes of a target by interpreting the echoesfrom radio or sound waves respectively. Ultrasonic sensorsgenerate high frequency sound waves and evaluate the echowhich is received back by the sensor. Sensors calculate the timeinterval between sending the signal and receiving the echo todetermine the distance to an object. This technology can beused for measuring wind speed and direction (anemometer),fullness of a tank and speed through air or water. For measuringspeed or direction a device uses multiple detectors and calculatesthe speed from the relative distances to particulates in the air orwater. To measure the amount of liquid in a tank, the sensormeasures the distance to the surface of the fluid. Furtherapplications include: humidifiers, sonar, medicalultrasonography, burglar alarms and non-destructive testing.Systems typically use a transducer which generates sound wavesin the ultrasonic range, above 18,000 hertz, by turning electricalenergy into sound, then upon receiving the echo turn the soundwaves into electrical energy which can be measured anddisplayed. The technology is limited by the shapes of surfacesand the density or consistency of the material.

V. CONCLUSION

In this paper, we have proposed method of vehicle tracking andlocking systems used to track the theft vehicle by using GPSand GSM technology. This system puts into the sleeping modewhen vehicle is handled by the owner or authorized persons;otherwise goes to active mode. The mode of operations changedby persons or remotely. When the theft identified, the responsiblepeople send SMS to the micro controller, then issue the controlsignals to stop the engine motor. After that all the doors locked.To open the doors or to restart the engine authorized personneeds to enter the passwords. In this method, we easily trackthe vehicle place and doors locked.

• Thus in this project we have provided the means of accidentprevention using eye blink sensor, accident sensor whereinthe vehicle is stopped immediately and intimated whereverneeded.

• It involves automated security system that provides highsecurity to driver through the use of GPS and GSMtechnologies.

• It involves obstacle detection with lane detection for efficientvehicle parking management at emergency situation.

• The GPS and GSM have been used for tracking the vehiclesto identify the accident and theft location.

• This system has been incorporated as a single unit insidethe vehicle.

REFERENCES

[1] Saif Al-Sultan, Ali H. Al-Bayatti and Hussien Zedan, “Context AwareDriver Behavior Detection System in Intelligent TransportationSystems” IEEE Vol 15, 2014

[2] S.Sonika, Dr.K.Sathiyasekar, S.Jaishree, (2014), Intelligent AccidentIdentification System using GPS, GSM modem, IJARCCE,Vol 3, Issue2, pp 5487-5489

[3] Pau Muñoz-Benavent, Leopoldo Armesto, Vicent Girbés, J. ErnestoSolanes, Juan Dols, Adolfo Muñoz, and Josep Tornero, “AdvancedDriving Assistance Systems for an Electric Vehicle” AUSMT, Vol 2,No 2, 2013

[4] V.Ramya, B.Palaniappan, K.Karthick (2012), “Embedded Controllerfor Vehicle In-Front Obstacle Detection and Cabin Safety Alert System”,IJCSIT,Vol4,Issue2,pp117-131

[5] Zhang Wen, Jiang Meng, “ Design of Vehicle Positioning System basedon ARM” IEEE Vol 14, No 4, 2011

[6] Chen, H., Chiang, Y. Chang, F., H. Wang, H. (2010). Toward Real-Time Precise Point Positioning: Differential GPS Based on IGS UltraRapid Product,SICE Annual Conference, The Grand Hotel, Taipei,Taiwan August 18-21.

[7] Asaad M. J. Al-Hindawi, IbraheemTalib, “Experimentally Evaluationof GPS/GSM Based System Design”, Journal of Electronic SystemsVolume 2 Number 2 June 2012

[8] KunalMaurya ,Mandeep Singh, Neelu Jain, “Real Time Vehicle TrackingSystem using GSM and GPS Technology- An Anti-theft TrackingSystem,” International Journal of Electronics and Computer Science

Engineering. ISSN 2277-1956/V1N3- 1103-1107



24

Multiple frequency Polyhedron Ring SlotMicrostrip Antenna

Srilakshmi. A, Associate Professor, Vasavi college of Engineering, ECE department, Hyderabad, India

[email protected]. Koteswara Rao, Head and Professor of ECE,

CBIT, Hyderabad, India, [email protected] Rao

Professor of ECE, JNTUH,HyderabadIndia, [email protected]

Abstract: A polyhedron shaped annular ring slot antenna ispresented. A simple 50 ohms Microstrip line is used to excitethe slot. . Parametric Design and Analysis is carried out withpermutations of the feed length from edge to the centre ofannular ring. RT-duriod substrate is used because of its highdirectivity and high power radiation[7]

Keywords Microstrip Antenna, Wideband, Annular Slot

1. INTRODUCTION

Microstrip Patch antennas generally consists of dielectricsubstrate with radiation patch on one side and ground plane onthe other side of the substrate. They are mainly four types ofMicrostrip antennas. Microstip dipole, Microstrip patch,Microstrip slot antennas and Microstrip travelling waveantennas.[4] The Microstrip antennas are widely used in wirelesscommunications because of its small size, light weight, lessfabrication cost, planar shape and integrability with other partsof the system. many of Microstrip antennas are used in differentapplications like Radar systems, Space communication systemsetc. The main disadvantage of Microstrip antennas is lowbandwidth. Normally the bandwidth for single frequencyannular slot is about 10 percent.

However the development of wireless communicationsexperiencing an exponential growth hence increase the needfor wideband Microstrip antennas. As a result, new antennashave to developed to provide larger bandwidth and this, withinsmall dimensions. challenge which arises is that the gain andbandwidth performances of an antenna are directly related toits dimensions. These applications include WWANs, WLANsand WPANs. Usually, broad band characteristics are tough toachieve, because good impedance matching is difficult. In thispaper a polyhedron ring shaped slot antenna is proposed, thereby increasing the surface current path and enabling widebandwidth of the slot antenna. By adjusting the parameters of

the antenna parametric analysis is carried out using HFSSsoftware by changing the feed length. A simple 50 ohmsMicrostrip line is used to excite the slot. The results are explainedbelow.

The annular ring antennas has received attention when it is tobe operated in in fundamental TM

11 mode. Annular ring slot

antenna is smaller than rectangular and circular patch antennaat a given frequency. The annular ring also has broad bad naturewhen operated near the TM12 resonance mode. [3]

A polyhedron ring antenna is designed for TM11

mode at theresonant frequency 2.4 GHz and analyzed for various resultssuch as VSWR, Return loss, input impedance, 2Dimensionaland 3D radiation patterns. Gain vs frequency results are alsoanalyzed.

2. Method of Analysis of Microstrip Antenna.

There are three methods to analyze Microstrip Antennas.Transmission line method, Cavity method and full wave model.Transmission line method is simple to implement, but is lessaccurate. Cavity model is complex to analyze and is moreaccurate than transmission line model. Full wave model is mostcomplex and is very accurate for finite , infinite arrays andstacked structures. There are different full-wave methods tosolve EM structures. Finite difference method, finite elementmethod, Method of moment etc.

In this paper finite element method is used in the analysis. Thequantity associated with EM wave is the poynting vector S=EXH w/m2. [5]This paper is aimed to design and analyzed apolygonal annular ring antenna suitable at ISM band.

The basic shape of the antenna is circular and polyhedral slot isincorporated in the ground plane as shown in the figure 1. Anddesign equations of circular patch antenna are taken toconsideration.



25

Fig1: Geometry of tetrahedral annular ring slot antenna

The fundamental resonant frequency of conventional annularcircular ring slot antenna comprises in the ground plane of adielectric substrate fed by a microstrip conductor can becalculated according to the following[6]

—————— 1

Where Rm is the average radius of inner ring radius and outer

ring radius in mm. f0 is the resonant frequency in GHz. And ε

re

is effective dielectric constant of annular ring patch is givenby[2]

—————— 2

To improve the impedance bandwidth one more annular linkedslot is introduced. Using below equation width of the microstripline width for 50 ohms is calculated[5].

——————3

3. ANALYSIS AND STUDY OF THE PARAMETERS

Polyhedron slot ring parameters are discussed and analyzedusing HFSS software which is a FEM based simulator. Annularslot antennas have relatively broad band characteristics compareto conventional patch antennas.

A single layer with Polyhedron ring shaped slot is taken out ofinfinite ground plane. FR4 substrate was selected to reduce thecost, which has a relative permittivity of 4.4, loss tangent of0.01 and a thickness of 1.56 mm. The ground plane size was50mmX50mm. Radius of inner ring of the slot is Ri=4 mm,Radius of the outer ring slot is taken as Ro=6mm.

The 50 ohms feed line length is fl= 24mm, and width is fw=3mmwas designed for good impedance matching. Radius of the innerslot ring is R

i=4 mm , radius of the outer circle of ring is R

o=6

mm,. The dimensions fl is optimized to get multiple frequencyresonance.

Feed line length is varied form 19mm to 34 mm for optimumresults.

4. RESULTS AND DISCUSSIONS

It was analyzed for the different feed lengths from 19mm to 34mm and observed at fl= 26mm results are good and thecorresponding result is shown in the figure 2.

For faith full radiation return loss should be less than -10dB. Inthe figure 3 return loss is represented. The antenna is resonatingat multiple frequencies 2.4GHz , 5GHz and 9.5 GHz with returnloss less than -10db. Anthe corresponding band widths are200MHz,400MHZ and 600MHz. Figure 10 representing theVSWR characteristics.

Fig 2: Optimization of return losss for various feed lengths.

It was observed that VSWR is less than 2 in the ISM band.

Figure 3: Return loss plot



26

Figure shows that the average gain of antenna along phi= 0 andphi = 90 degrees is 15dBm and 25dBm is the ISM band.

Figure 10. VSWR characteristics for different fl

Figure 4: Gain vs frequency

Fig:9: Impedance characteristics of the antennaFigure 5 shows the distribution of the current along the feed and

ground plane and slot.Figure 7 and 8 are representing the two dimensional and three

dimensional Radiation pattern. Figure 9 representing theimpedance characteristics with respect frequency.

Figure5: Current distribution of the antenna

Figure 6: Radiation pattern at 2.36GHz

Figure 4

Figure 8: 3D Radiation pattern at 2.36GHz.

5. CONCLUSION

A Polyhedron slotted ring antenna is designed and analyzed.Simulation results are presented. The antenna is resonating atmultiple frequencies 2.4GHz, 5GHz and 9.5 GHz with returnloss less than -10db. And the corresponding band widths are200MHz,400MHZ and 600MHz. VSWR<2 is achieved in at2.4 GHZ and 5GHZ and 9.5GHZ.

REFERENCES

1) Xiulong Bao, Max Amman,Microstrip-Fed Dual-Frequency Annular-Slot Antenna Loaded by Split-Ring-Slot,IET Microw. Antennas Propag.,2009, Vol. 3, Iss. 5, pp. 757–764

2) BATCHELOR J.C., LANGLEY R.J.: ‘Microstrip ring antennasoperating at higher order modes for mobile applications’,IEE Proc. H,Microw., Antennas Propag., 1995, 141, (2),pp. 151–155

3) GUO Y.X., LUK K.M., LEE K.F.: ‘L-probe proximity-fed annular ringmicrostrip antennas’, IEEE Trans. Antennas Propag., 2001, 49, (1),pp. 19–21

4) Gonca, C. (2005) Design, Simulation and Tests of Low-cost MicrostripPatch Antenna Arrays for the Wireless Communication Turk J ElectEngin, 13 (1) [5]. Richards, W.F. (1988) Microstrip Antennas. Theory,Application and Design. Van Reinhold Co., New York.

5) Balanis.C.A.Antenna Theory Analysis Design, Second Edition, Unitedstates of America, John Wiley &Sons 1997..p73

6) Srilakshmi.A, Koteswararao.N.V, Srinivasarao.D, Recent Advances inIntelligent Computational Systems (RAICS), 2011 IEEE , 2011 ,Page(s): 851 – 855

7) Wikipedia www.http.wikipedia The free on-line encyclopedia.htm



27

Provision of Trustworthy Associations in Peer toPeer Structures

Ramesh Babu, Associate professor, CSE [email protected]

MallaReddy Engineering College for WomenMaisammaguda, Dhulapally Via Hakimpet, Secunderabad -14. Telangana, India

A. Radha Rani, Associate professor, CSE [email protected]

MallaReddy Engineering College for WomenMaisammaguda, Dhulapally Via Hakimpet, Secunderabad -14. Telangana, India

ABSTRACT: The systems of Peer-To-Peer are motivatingtowards a foremost paradigm move in the direction ofgenuinely distributed computing. Abundant works onreputation as well as trust management in online communitieshave come into view in recent times. Classification of peersas moreover trustworthy or untrustworthy is not enough inmost of the cases. Metrics should include accuracy hence peerscan be ranked consistent with trustworthiness. Metric ofreputation is considered on basis of recommendations and itis considered as important when deciding in relation tostrangers as well as new acquaintances. When scheming ofreputation metric, recommendations are assessed on basis oftrust metric of recommendation. In this work we aim atintroducing an architecture known as SORT which is arepresentation of self-organizing Trust intends to reducemalevolent activity in P2P structure by setting up trustassociations between peers in their proximity. Inrepresentation of self-organizing Trust, peers conveyreputation queries merely to peers which had worked togetherin past, which decrease network traffic when compared to themethods of flooding-based approaches. In representation ofself-organizing Trust, rather than considering an exact trustholder’s feedback as valid, public view from each and everyacquaintance is measured as more trustworthy information.SORT’s trust metric allow a peer to consider constancy ofprevious peers on basis of local information.

Keywords: Peer-To-Peer systems, Trustworthiness,Reputation, Trust, Self-organizing trust model, Acquaintance.

1. INTRODUCTION

Within Infrastructure of P2P system, established distinctionamong clients as well as back-end servers is basicallydisappearing. Managing of trust information is dependenttowards structure of P2P system. The system of peer to peerwas generally characterized by several properties such as no

central harmonization, no peer has a global vision of system,global performance come out from local interactions, and peersas well as associations are variable. Managing of trust is asetback of meticulous significance in peer-to-peer setting whereone normally meets unidentified agents [6]. The majority oftrust representations do not consider the method in whichinteractions are rated and believe that a rating mechanism exists.The techniques which are traditional for managing trust are onbasis of reputation; spotlight on semantic properties of trustrepresentation. Those methods do not extent since they dependon a fundamental database or else necessitate for maintainingglobal information at every agent to make available data onprevious interactions. Necessary problem connected tomanaging of reputation-based trust in P2P systems is thatinformation concerning transactions performed among peers isdispersed through-out network with the intention that each peercan put up an estimation of global circumstance in network [8].Architecture for managing trust was shown in fig1 which relieson entire system layers, specifically network, storage as well asmanaging of trust, on peer-to-peer mechanisms [5]. In suchdesign a mechanism which was put into practice at an advancedlevel in a peer-to-peer manner has to consider properties,especially quality of service, concerning mechanisms ofunderlying layers. In this work we aim at introducing anarchitecture known as SORT which is a representation of self-organizing Trust intends to reduce malevolent activity in P2Pstructure by setting up trust associations between peers in theirproximity [1] [4]. In the representation of SORT, peers aresupposed to be strangers for each other at beginning and a peerturns out to be an acquaintance of another peer subsequent toproviding a service. Parameters associated to peer capabilities,peer performance, as well as distribution of resources areapproximated to quite a lot of empirical results which permitsto build practical observations on progression of trustassociations [12].



28

2. AN OVERVIEW TOWARDS METHODOLOGY OFSELF-ORGANIZING TRUST REPRESENTATION:

In the systems of distributed hash table - basis methods eachpeer turns out to be a trust holder by means of accumulatingfeedbacks regarding other peers [15]. Classification of peers asmoreover trustworthy or untrustworthy is not enough in mostof the cases. Metrics should include accuracy hence peers canbe ranked consistent with trustworthiness [3]. Solution on astructured system depends on DHT organization to store uptrust information. Every peer turns out to be a trust holder of anadditional peer, which is supposed to make available reliableglobal trust information on the other hand; a trust holder maybe malevolent and offer inauthentic information. Abundantworks on reputation as well as trust management in onlinecommunities have come into view in recent times [14]. Therepresentation of common reference model that was applied bymost of the approaches was shown in fig2. Representation oftrust on P2P proposals has additional challenges when measuredto e-commerce scheme. Malicious peers include additionalattack opportunities in P2P trust representations due to shortageof a central authority [10]. SORT which is a representation ofself-organizing Trust intends to reduce malevolent activity inP2P structure by setting up trust associations between peers intheir proximity and defines three trust metrics. Metric ofreputation is considered on basis of recommendations and it isconsidered as important when deciding in relation to strangersas well as new acquaintances [9]. Reputation loses its

significance as understanding with an acquaintance enhances.The trust metric of recommendation is significant whenappealing for recommendations. When scheming of reputationmetric, recommendations are assessed on basis of trust metricof recommendation. Service as well as recommendation trustare most important metrics to compute trustworthiness in serviceas well as recommendation circumstance [7]. The metric ofservice trust is used when selection of service provider. A peermight be considered as good service provider however a badrecommender or else vice versa hence representation of self-organizing Trust considers providing services as well asproviding recommendations as altered tasks and describes twocontexts of trust such as service as well as recommendationcontexts [2] [13]. In representation of self-organizing Trust,rather than considering an exact trust holder’s feedback as valid,public view from each and every acquaintance is measured asmore trustworthy information. Rather than consideringworldwide trust information, information concerning local trustis adequate to build decisions as peers build up their personaltrust networks [16]. In representation of self-organizing Trust,peers convey reputation queries merely to peers which hadworked together in past, which decrease network traffic whencompared to the methods of flooding-based approaches [12].Each peer increases its trust network and gets hold of additionalconvincing recommendations from acquaintances.

Fig1: An overview various system levels concerning P2Pcomputing.

Fig 2: An overview of representation of reputationand managing trust.

3. RESULTS:

The systems of Peer-To-Peer are motivating towards a foremostparadigm move in the direction of genuinely distributedcomputing. The majority of trust representations do not considerthe method in which interactions are rated and believe that arating mechanism exists. In representation of self-organizingTrust to assess interactions as well as recommendationsimproved, significance, recentness, as well as parameters of peersatisfaction are measured. Recommender’s trustworthiness as



29

well as assurance concerning recommendation is consideredwhen assessing recommendations. Service as well asrecommendation circumstance is separated which facilitates tocompute trustworthiness in extensive variety of attack situations.Simulation tool of P2P file sharing was put into practice andperformed experiments to recognize impact of SORT inmitigating attacks. Parameters associated to peer capabilities,peer performance, as well as distribution of resources areapproximated to quite a lot of empirical results which permitsto build practical observations on progression of trustassociations. Experiments which are carried out on SORTdemonstrate that superior peers can protect themselves againstmalevolent peers devoid of containing global trust information.SORT’s trust metric allow a peer to consider constancy ofprevious peers on basis of local information. Service as well asrecommendation circumstance allow improved measurement oftrustworthiness in offering services as well as providingrecommendations.

4. CONCLUSION:

Managing of trust information is dependent towards structureof P2P system. The techniques which are traditional formanaging trust are on basis of reputation; spotlight on semanticproperties of trust representation. Representation of trust onP2P proposals has additional challenges when measured to e-commerce scheme. In this work we aim at introducing anarchitecture known as SORT which is a representation of self-organizing Trust intends to reduce malevolent activity in P2Pstructure by setting up trust associations between peers in theirproximity. In representation of self-organizing Trust, peersconvey reputation queries merely to peers which had workedtogether in past, which decrease network traffic when comparedto the methods of flooding-based approaches. Service as wellas recommendation trust are most important metrics to computetrustworthiness in service as well as recommendationcircumstance. In representation of self-organizing Trust toassess interactions as well as recommendations improved,significance, recentness, as well as parameters of peersatisfaction are measured. SORT’s trust metric allow a peer toconsider constancy of previous peers on basis of localinformation. In representation of self-organizing Trust, ratherthan considering an exact trust holder’s feedback as valid, publicview from each and every acquaintance is measured as moretrustworthy information. Service as well as recommendationcircumstance allow improved measurement of trustworthinessin offering services as well as providing recommendations.

REFERENCES:

[1] S. Staab, B. Bhargava, L. Lilien, A. Rosenthal, M. Winslett, M. Sloman,T. Dillon, E. Chang, F.K. Hussain, W. Nejdl, D. Olmedilla, and V.Kashyap, “The Pudding of Trust,” IEEE Intelligent Systems, vol. 19,no. 5, pp. 74-88, 2004.

[2] E. Terzi, Y. Zhong, B. Bhargava, Pankaj, and S. Madria, “An Algorithmfor Building User-Role Profiles in a Trust Environment,” Proc. FourthInt’l Conf. Data Warehousing and Knowledge Discovery (DaWaK), vol.2454, 2002.

[3] S. Saroiu, K. Gummadi, R. Dunn, S.D. Gribble, and H.M. Levy, “AnAnalysis of Internet Content Delivery Systems,” Proc. Fifth USENIXSymp. Operating Systems Design and Implementation (OSDI), 2002.

[4] Ahmet Burak Can, and Bharat Bhargava “SORT: A Self-ORganizingTrust Model for Peer-to-Peer Systems”, 2013

[5] F. Cornelli, E. Damiani, S.D.C. di Vimercati, S. Paraboschi, and P.Samarati, “Implementing a Reputation-Aware Gnutella Servent,” Proc.Networking 2002 Workshops Web Eng. and Peer-to-Peer Computing,2002.

[6] H. Yu, M. Kaminsky, P.B. Gibbons, and A. Flaxman, “Sybilguard:Defending against Sybil Attacks via Social Networks,” ACMSIGCOMM Computer Comm. Rev., vol. 36, no. 4, pp. 267-278, 2006.

[7] M. Ripeanu, I. Foster, and A. Iamnitchi, “Mapping the GnutellaNetwork: Properties of Large-Scale Peer-to-Peer Systems andImplications for System Design,” IEEE Internet Computing, vol. 6, no.1, pp. 50-57, Jan. 2002.

[8] M. Virendra, M. Jadliwala, M. Chandrasekaran, and S. Upadhyaya,“Quantifying Trust in Mobile Ad-Hoc Networks,” Proc. IEEE Int’l Conf.Integration of Knowledge Intensive Multi-Agent Systems (KIMAS),2005.

[9] Y. Wang and J. Vassileva, “Bayesian Network Trust Model in Peer-to-Peer Networks,” Proc. Second Workshop Agents and Peer-to-PeerComputing at the Autonomous Agents and Multi Agent Systems Conf.(AAMAS), 2003.

[10] A. Habib, D. Xu, M. Atallah, B. Bhargava, and J. Chuang, “A Tree-Based Forward Digest Protocol to Verify Data Integrity in DistributedMedia Streaming,” IEEE Trans. Knowledge and Data Eng., vol. 17,no. 7, pp. 1010-1014, July 2005.

[11] I. Stoica, R. Morris, D. Karger, M.F. Kaashoek, and H. Balakrishnan,“Chord: A Scalable Peer-to-Peer Lookup Service for InternetApplications,” ACM SIGCOMM Computer Comm. Rev., vol. 31, no.4, pp. 149-160, 2001.

[12] R. Zhou and K. Hwang, “Powertrust: A Robust and Scalable ReputationSystem for Trusted Peer-to-Peer Computing,” IEEE rans. Parallel andDistributed Systems, vol. 18, no. 4, pp. 460-473, Apr. 2007.

[13] K. Aberer and Z. Despotovic, “Managing Trust in a Peer-2-PeerInformation System,” Proc. 10th Int’l Conf. Information andKnowledgeManagement (CIKM), 2001.

[14] S. Xiao and I. Benbasat, “The Formation of Trust and Distrust inRecommendation Agents in Repeated Interactions: A Process- TracingAnalysis,” Proc. Fifth ACM Conf. Electronic Commerce (EC), 2003.

[15] G. Swamynathan, B.Y. Zhao, and K.C. Almeroth, “Decoupling Serviceand Feedback Trust in a Peer-to-Peer Reputation System,” Proc. Int’lConf. Parallel and Distributed Processing and Applications (ISPA),2005.

[16] Z. Despotovic and K. Aberer, “Trust-Aware Delivery of CompositeGoods,” Proc. First Int’l Conf. Agents and Peer-to-Peer Computing,2002.



30

Features Extraction and Selection Based on RoughSet in Shot Detection

GS Naveen Kumar1, VSK Reddy2, S.Srinivas Kumar3

1Research Scholar, JNTUK, Kakinada, India.2Professor & Principal, MRCET, Hyderabad, India.

3Professor and Director R&D, JNTUK, Kakinada e-mail: [email protected]

Abstract : Rough Set based reduction algorithm is adoptedas a method for feature selection in abrupt shot boundarydetection. First, some information of macro blocks isextracted in P frame of MPEG video sequence. Then theinformation of the motion-activity, the type of macro blocksand the motion-distribution are obtained by analyzingMPEG compressed-domain. Combined these informationwith the difference of pixel and the difference of histogramthe abrupt shot detection can be achieved. The simulationexperimental results show that the detection modelcombined Rough Set with SVM is effective in featuresselection. Some useful features for abrupt shot detection arediscovered.

1. INTRODUCTION

The extensive amount of video coverage today, generatesdifficulties in identifying and selecting desired information.Obviously, traditional retrieval based on text cannot meet thesedemands, so content-based video retrieval has been proposedas a solution to address this problem [1]. Shot-Boundarydetection technique is the foundation of video analysis andcontent-based retrieval which impact its accuracy. A lot ofresearch works have been done in this field recent years, butthere are few research works on video features selectionespecially in shot boundary detection. An assumption often madeis that the content should remain nearly the same from one frameto the next within one camera shot. So, in general, shotboundaries can be detected by employing a difference metricto measure the change between two consecutive frames. A shotboundary is declared if the difference exceeds a certainthreshold. Pixel- or block-based temporal image difference [2]-[3], or difference of gray and color histograms [4]-[5] weresupposed to be the measure metric. Histograms are robust toobject motion. And they are easy to compute. So they have beenwidely used in shot-based video analysis. And several authorsclaimed that this measure could achieve good trade-off betweenaccuracy and speed. Unfortunately there is a problem ofsegmenting the film into a sequence of shots based on differenceof histograms when illumination varies. Video parsing on

MPEG compressed data has been reported by making use ofDCT blocks and motion vectors information [4]. A count ofnon-zero motion vectors was used to detect scene discontinuity.However, this method failed to handle special effects. Meng etal. [6] used the variance of DC coefficients in I and P framesand motion vectors information to characterize scene changes.Sethi and Pate1 [5] used only the DC coefficients of I frame toperform hypothesis testing via luminance histogram. It wasassumed that distance between two I frames was fixed and small.The exact location of abrupt changes cannot be located withthis method. Liu et al. made use of only information in P and Bframes to detect shot boundary changes. The above related worksshow that different researchers took different features as inputfor shot boundary detection.

Based on these features, few of them took feature selection as amodule of their video retrieval system. They did not pay muchattention to finding and selecting important and acoustic featuresin develop shot boundary detection system. Video frequencydata has the property of non-constitutive and the complexity,including information about text, picture, and sound. At themeantime, the features extracted from the video frequency dataare not precise, whereas we use Rough Set in our abrupt shotdetection. Rough Set theory was proposed by Z. Pawlak in 1982as a powerful mathematical analysis tool to process incompletedata and inaccurate knowledge [7]. Without any mathematicaldescription of attributes and features of detected objects inadvance, it determines the knowledge reduction and educesdecision rules via indiscernibility relations and classes directlyfrom given knowledge classification. So far, Combined RoughSet theory with content based video retrieval, we have achieveda series of research results: proposed a partition algorithm forhuge data set based on Rough Set [9]; developed a motioninformation based video pre classification retrieval system usingRough classification [10]-[11]; combined Rough Set with videokey-frame extraction and got an acceptable result[12]. In theaspect of shot-boundary detection, we achieved an algorithmreferenced to [15] and improved its performance [10] whichcould effectively detect abrupt shot, but it was still a bit subject



31

to light and object movements. In this paper, an approach toshot boundary detection using Rough Set is introduced whichcomprehensively utilizes the motion features of P frame inMPEG video sequence, the difference of pixel and the differenceof histogram.

Since there are uncertainty and inconsistency in the classificationof shot-boundary detection, the rough set method is adopted toconstruct a better shot type. In our former work, we used RoughSet as an approach for shot-boundary detection .We found thatexperiment result was not stable when Rough Set was taken asfeature selection and classifier. Usually, a higher recognitionrate can be achieved when Rough set is taken as feature selectionand SVM as classifier. In this paper, we use the method of RoughSet and SVM for shot-boundary detection from video sequenceimages separately. That is, Rough Set is taken as a module offeature selection and SVM is taken as classifier in shot-boundarydetection. Through the experiment we find some useful videofeatures for shot-boundary detection. After the overallintroduction, the rest of this paper is organized as follows. Insection 2, some related theory and work are introduced. Insection 3, the framework of a shot-boundary detection system(CBVRSYSTEM) is proposed. In section 4, Simulationexperiments are done by means of CBVRSYSTEM. Finally,conclusion and future works are discussed in section 5.

II. ROUGH SET THEORY: BASIC CONCEPT

Rough set concept was introduced by Polish logician, ProfessorZ.Pawlak in early 1980s [2, 3]. It is an extension of the settheory for the study of intelligent system characterized byinexact, uncertain or vague information and can serve as a newmathematical tool to soft computing [12]. General elementsengage in rough sets theory can be described as follows:

2.1 Information systems

Let, an information system is a set of objects represented in adata table, the rows are considered as objects for analysisand the columns represent a measureable attributes for eachobject. Formally, an information system can be seen as a system,IS = (U, A) where U is finite set of objects, U={x

1, x

2, x

3,…,

xn}; and A is a finite set of attributes (features, variables),

the attributes in A are further classified into

disjoint condition attributes C and decision attributes D,such that A=C D and C D = ∅2.2 Indiscernibility relation

Indiscernibility relation is the relation between two objects ormore, where all the values are identical in relation to asubset of considered attributes. The indiscernibility relationis defined as, R (B) = {(x,y) ∈ U x U : for all ∈B, a(x)= a(y)}where, a A and B⊆A ;

2.3 Lower and upper approximation

Approximations are fundamental concepts of rough set theory,it is can be defined as upper bounds and lower bounds, AS

= ( U, R(C) ) where, C be a set of condition attributes andR(C) be an indiscernibility relation on U. [x]

B denotes the

equivalence class of B containing x, for any element x of U;Based on singleton x, for a given

B⊆A and X⊆U, the lower approximation ( BX) of the set Xin IS and the upper approximation of the set X in IS (BX)are defined as follows:

BX = {x∈U: [x]B ⊆ X}. (1)

= {x ∈ U: [x]B X ≠ φ}. (2)

For a given B⊆A and X⊆U, the boundary of X in IS can bedefined as,

BND(X) = BX- BX (3)

BND(X) consists of objects that do not certainly belong to Xon the basis of A.

3.4 Attribute reduction

Reduction means, the set of remaining attributes is the minimalset, and set which presents in all subsets call cores, in otherwords, removing repetitive or overlapping data. The mainpurpose of reduction is to determine the attributes whichcan represent data in a database and dependencies betweenattributes

3.5 Decision rule

Decision rule created by combining rule reducts attributes. Eachrows of reducttable verify a decision rule, which specifies thedecision that must be taken when condition are indicated bycondition attributes are fulfilled. Decision rules frequentlypresented as implication called “if…then…” rules.

Data used in rough set theory are often presented as a tablewhich is initialized as decision table as illustrated in figure

1. In the table, columns correspond to attributes and rows ofthe decision table correspond to objects. Entries in the tableare attribute values. The decision attributes can have somevalues though quite often it is binary [14].

⊂

⊂

⊂

III Shot-Boundary Detection Systems

Shot boundary detection pre-classification model is shown inFig.2, which is composed of such modules as feature extraction,

Fig.1 : Decision Table



32

As for MPEG, each video sequence is composed of a series ofGroup of Picture (GOP). While P can best represent movementsof the whole GOP. Here, we will use the P frame information incompression region. The macro block types are denoted as thefield macroblock_type in MPEG stream. However, the field isjust used in video decoding, but hardly in video retrieval.Furthermore, they can be divided into three types, which arelow change (L), middle change (M) and high change (H). Sothe change ranges of frames are shown in the fieldmacroblock_typein a certain degree. Via the field

macroblock_type, ratios of macro block types are defined asfollows:

RateH = Hcount / Tount , (3)

RateM = MCount / TCount , (4)

RateL = LCount / TCount , (5)

where Tcount is the whole count of a P frame, and the otherXcount are the count of the macro block type X. Motion activityis a concept related with the activity of a video sequence or amovement pace. In this paper, we refer to literature [13] to definethe motion activity as follow:

In the coordinate system for a P frame, x(i, j) and y(i, j) are thecoordinates of every macro block in x and y directionsrespectively. We define: the energy of the macro block. x(i, j) ishorizontal motion vector of macro block and y(i, j) is the verticalone. The average motion activity for every frame calculated asfollows:

Ratemav

denotes the proportion of the motion activity in the shottotal activity in the P frame, while CmvShot denotes the averagemotion activity of the shot where contains the P frame, and n intwo formula is the total number of P frames in the shot. Thesethree attributes related with motion activity indicate the totalactivities of a video fragment and a few P frame activity in eachshot. In the approach of which based on motion activity intensity,the shot-boundary detection must be selected depend on theseattributes. Motion spatial distribution must be paid attention towhich describes the distribution of the activity in an image. Twomeasurements, Rog and Weight, are introduced to describe thespecial distribution of each image. Referring to the literature[14], we define:

feature selection, classification model. Our system is developedbased on MPEG-2 encoder [19]. So, video sequences based onMPEG-2 are taken into consideration as input.

3.1 Feature Extraction

3.1.1 Motion Information from Compressed MPEG Stream

Fig.2: Shot Boundary detection pre classification model.

We still need to consider the condition of motion activity of theshot with P frame belongs to. We define:

3.1.2 Pixel and Color Histogram

Pixel and color histogram are two important clues for thedescription of video content. For a video frame, the colorinformation summarizes the appearance of the embedded objectswell, not considering the location of the pixel. So, it is possiblethe histograms of two frames are similar, but the contents arecompletely different. The pixel information summarizes thedistribution of the embedded objects well. But, it is easilysubjected to the voice and motion. In this paper, we combinetwo features merit. They can complement each other nicely.

Fig 3: Pixel difference distribution

Fig 4: A series of video sequence



33

where X, Y is the width and height of the picture;

Lk(x,y)and L

k+1(x,y) is the value of the frame K and frame K+1

respectively. The shot-boundary will appear obviously changedbased on pixel. Fig.3 and Fig.4 show the pixel difference of avideo sequence.

As the three color components of the HSV color space areindependent, we first calculate the histogram differences ofconsecutive video frames on each color component respectively.Fig.5 shows the distribution of the HSV histogram between shot-boundary. And then, we use the maximum histogram differenceas the feature. The HSV histogram difference HD

i is calculated

as follows:

decreased. Thus, the complexity and cost for the followingclassification procedure are reduced. It could improve theefficiency of the whole system. In our system, features selectionmodule is based on the Rough Set theory. The attribute reductionalgorithm adopted to select features is introduced in section 2.

3.3 CLASSIFICATION MODULE

SVM are taken as the classifier in our shot-boundary detectionsystem. There are two cases of classification problems: linearseparable and linear non-separable. In practice it is necessaryto employ the non-separable approach. In statistical learningtheory [17] [18], we bound the difference between the expectedrisk R(α), and the empirical risk Remp(α), when both trainingand tests sets are assumed to be generated from the sameunderlying probability distribution P(x, y) . The empirical riskis calculated by:

The difference of pixel is easy to get. D(k,k +1) is defined asthe difference between frame K and frame K+1:

Hi , S

i and V

i are the histogram differences of H, S, and V

component, respectively. hiH, h

iS, and h

iV normalized histograms

of H, S and V, respectively.

3.2 FEATURE SELECTION

Feature selection module is the key part of our shot-boundarydetection system. The purpose of this module is to selectvaluable features for shot boundary detection from all features.In this module, features for shot-boundary detection are

Fig 5: The distribution of HSV histogram

where l is the size of the training set, α is the model parameterf(x

i , α)and is the classifier output for a trainingvector x having

a corresponding label yi∈{−1,1}.The risk for an unseen test

vector x is

IV EXPERIMENTS AND ANALYSIS

In our experiments, the RIDAS system is adopted as a datamining platform, which is developed by Institute of ComputerScience and Technology, Chongqing University of Posts andTelecommuni-cations in China [16]. The system integratesalmost 30 classical algorithms regarding rough set theory. Theexperiment data includes 938 frames which are randomlydivided by 5 and chosen 376 as training data. As aboveintroduction, we choose 11 features as the set of shot-boundarycondition attributes:

From Table2, we can see that the results are not stable usingRough Set as classifier. Using these features, we form thedecision table as shown in Table1. Let A=CD denotes attributesset of P frame, C is the set of total condition attributes of Pframe as listed above, and D is the decision attribute of the Pframe, D = { type }. Then with the reduction algorithm of [8], 5features are selected:

In these 5 features selected by Rough-Set-based attributereduction algorithm, there are 4 compressed features and 1 non-compressed feature. These 5 features are taken as the input ofthe classifier of SVM. At the mean time, we define: if the frame



34

is a hard cut frame, type amount 1; if the frame is a gradualtransition frame, type amount 2; otherwise type amount 0. Inthis way, all the frames are classified into 3 types: hard cut frame,gradual transition frame and common frame. Then, we useRough Set as feature selection and SVM as classifier. The resultis shown in table 3. From Table2, Table3 and Table4, we canconclude that Rough Set can remove some unimportantcondition attributes, while SVM has a good classification result.At the meanwhile, the video features we extract and select caneffectively distinguish shot-boundary. The proposed method istested on three digital videos.

The contents of the experimental data are advertisement, moviefragments and dace fragments. The results are shown in Table5.We calculate Precision and Recall by:

Precision= (Correct boundaries)/

(Correct boundaries + Error boundaries)

Recall= (Correct boundaries)/

(Correct boundaries + Missed boundaries)

A comparative experiment using color algorithm, pixelalgorithm and the algorithm of [15] is done as is shown in Table6.In order to validate the advantage of our method based on RoughSet+SVM in abrupt shot detection, we go on to conduct acontrastive experiment The test video, a fragment of movie lastsfor 80 seconds and contains 2684 frames.

The two adjacent frames are considered as different shot-boundary. In fact,they belong to the same shot. On the contrary,we may see that our method behaves well in overcoming thosedefects.

VCONCLUSIONS AND FUTURE WORK

In this paper, based on Rough Set theory, an approach of shot-boundary detection is proposed. Based on Rough Set reduction,some important features for shot-boundary detection arediscovered according to our simulation experiment results.Depending on these features, an average recognition rate of92.55% is achieved using SVM + Rough Set. That is, the videofeatures we select can effectively use to distinguish shot-boundary especially in abrupt shot-boundary detection. In thefuture, we are going to study further on the gradual transitiondetection, application to manage the video data, and explore itsapplications, such as the system of video retrieval. Effectivereduction algorithm for video features selection for videoretrieval system will also studied.

REFERENCES

[1] H. Zhang, J.Y.A.Wang, and Y.Altunbasak, An integrated system forcontent-based video retrieval and browsing, Pattern Recognition,30:634-648, 1997.

[2] K. Otsccji, Y. Tonomura and Y. Ohba, Video browsing using brightnessdata, Proceedings of the SPIE on Visual Communications and ImageProcessing, 1606: 980-989, 1991.

[3] A. Nagasaka and Y. Tanaka, An automatic video indexing and videosearch for object appearances, Proceeding of the 2nd WorkingConference on Visual Database Systems, pp. 119-133, 1991.

[4] H. Zhang, A.Kankanhalli and S. Smoliar, Automatic partitioning offull motion video, Multimedia System, l1:10-28, 1993.

[5] I. K. Sethi and N. Patel, A Statistical Approach to Scene ChangeDetection, Proceeding of the SPIE on Storage and Retrieval for Imageand Video Databases, 2420: 329-338, 1995.

[6] H. J. Zhang, et al, Video parsing, retrieval and browsing: An integratedand Content-Based Solution, In: Proc. of ACM Multimedia 95 SanFrancisco, pp. 15-24, 1995.

[7] Z. Pawlak, Rough Sets, International Journal of Computer andInformation Sciences, 11:314-356, 1982.



35

[8] S.H.Nguyen, and A.Skowron, Quantization of Real Value Attributes-Rough Set and Boolean Reasoning Approach, Proc of the second JointConference on Information Sciences, pp. 34-37, 1995.

[9] Z.R. Qin, Y. Wu and G.Y. Wang, A Partition Algorithm for Huge DataSets Based on Rough Set, Mode Identification and ArtificialIntelligence, pp. 249-256, 2006.

[10] Z.Yuan, Y. Wu, G.Y. Wang and J.B. Li, Motion-Information-BasedVideo Retrieval System Using Rough Pre-classification, Lecture Notesin Computer Science, Book Transactions on Rough Sets V, 4100: 306-333, 2006.

[11] Z.Yuan, Y. Wu, G.Y. Wang and J.B. Li, A Global-Motion AnalysisMethod via Rough Set Based Video Pre-classification, 10thInternational Conference, RSFDGRC, pp. 323-332, 2005.

[12] T. Wang, Y. Wu and L. Chen, an Approach to Video Key-frameExtraction Based on Rough Set, Multimedia and UbiquitousEngineering, pp. 590-596, 2007.

[13] T. L. Yu and S. J. Zhang. Video Retrieval Based on the Global MotionInformation, http://cs.cqupt.edu.cn/videoretrieval

[14] S. Jeannin, A. Divakaran., MPEG-7 visual motion descriptors, IEEETransactions on Circuits and Systems for Video Technology, 11: 720-724, 2001.

[15] H. B. Lu, S. J. Zhang, An efficient algorithm for detecting abrupt scenechange, Journal of Image and Graphics, 10:805-810, 1999,

[16] G. Y. Wang, Z. Zheng, Y. Zhang, RIDAS-A Rough Set Based IntelligentData Analysis System, Proceedings of the First Int. Conf. on MachineLearning and Cybernetics, pp. 646-649, 2002.

[17] C.J.C.Burges, A tutorial on support vector machines for patternrecognition, Data Mining, and Knowledge Discovery, 2(2): 121-167,1998.

[18] A.B.Hur, D.Hom, H.T.Seigelmann, V.Vapnik, A Support Vector Methodfor Clustering, Advances in Neural Information Processing Systems,

13: 367-373, 2001



36

Mobile Element Routing, Data Gathering andEnergy Efficient Data Transmissionin Wireless

Sensor NetworksS. Ravi Kumar1, D. Srinivasa Rao2

1(Department of Electronics and Communication Engineering, JNTUH, Hyderabad Email: [email protected])

2(Department of Electronics and Communication Engineering, JNTUH, Hyderabad)

ABSTRACT : Recent research shows that significant energysaving can be achieved in wireless sensor networks with amobile base station that collects data from sensor nodes viashort-range communications. We consider the problem ofgathering data from a sensor network using mobile elements.The system is limited to single receive antennas the non-optimization of encoding/decoding order. This project is todevelop the Wireless Distributive System Management withhigh reliability, mobility and routing. We propose analgorithmic solution that to provide the energy efficient datapath planning for the mobile system and we go for uppersampling in the encoding processing.The choice ofimplementing algorithm depends upon the power allocation,nodal analysis, data gathering and node localization.Thesystem to multiple receives antennas for the nonoptimization of encoding/decoding order. Also the existingsystem is an approach to achieve lower data rate withsufficient performance (38Mbps). We should increase DataRate of several Mb/sec (58Mbps). This can be achieved bylinear processing. By balancing the system, the speed ofthe MIMO system is optimum.

Keyword: Wireless Sensor Network, Routing, DataGathering, Energy efficient.

I. INTRODUCTION

A Wireless Distribution System (WDS) is a system that enablesthe wireless interconnection of access points in an IEEE 802.11allows a wireless network to be expanded using multiple accesspoints without the need for a wired backbone to link them, as istraditionally required. The notable advantage of WDS over othersolutions is that it preserves the MAC addresses of client framesacross links between access points. An access point can be eithera main, relay, or remote base station. A main base station istypically connected to the wired Ethernet. A relay base stationrelays data between remote base stations, wireless clients orother relay stations to either a main or another relay base station.A remote base station accepts connections from wireless clientsand passes them on to relay or main stations. Connections

between “clients” are made using MAC addresses rather thanby specifying IP assignments.All base stations in a WirelessDistribution System must be configured to use the same radiochannel, method of encryption (none, WEP, or WPA) and thesame encryption keys. They may be configured to differentservice set identifiers. WDS also requires that every base stationbe configured to forward to others in the system.WDS may alsobe referred to as repeater mode because it appears to bridgeand accept wireless clients at the same time (unlike traditionalbridging). However, with this method, throughput is halved forall clients connected wirelessly.

WDS can be used to provide two modes of wireless AP-to-APconnectivity:

· Wireless Bridging in which WDS APs communicate onlywith each other and don’t allow Wireless clients orStations(STA) to access them.

· Wireless Repeating in which APs communicate with eachother and with wireless STAs.

Figure 1.1: Relay Path

In the recent past, the popularity of wireless sensor networks(WSNs) has been manifested by their deployment in many real-life applications (e.g., habitat study and ecology monitoring.With potentially a large number of sensor nodes scattered in aregion of interest, one of the challenging problems in WSNs ishow to efficiently aggregate the data sampled at each node to abase station, which has the computational power to store andprocess all the collected data. Note that, sensor nodes aregenerally battery powered and it is hard (if not impossible) toreplace those batteries after their deployment. Therefore,



37

developing energy efficient data gathering schemes is ultimatelyimportant to reduce the energy consumption on individual sensornodes, and thus extending the lifetime of WSNs. In conventionalWSN deployments, the data aggregation is normally achievedthrough multi-hop data forwarding schemes.

Figure 1.2: An example of data collection in a 500 X 500 m2

sensing field. The BS moves at 0.5 m/s. It takes the BS about 20minutes to visit all rendezvous points located within 100 m from thecenter of field. It takes more than 2hours to visit 100 source nodes

randomly distributed in the field.

In these schemes, for the sensor nodes that are far away andcannot reach the base station in a single hop, their data will berelayed by their neighbors that are closer to the base station.However, the major shortcoming of such schemes is that theenergy for the sensor nodes that are close to the base stationwill be quickly depleted due to their high data transmissionactivities, thus limiting the lifetime of WSNs. To address thisproblem, the ability of base stations has been exploited, wherethe base station moves around in the field to collect data fromsensor nodes. For the cases where the base station is notmoveable, energy efficient data gathering schemes that exploitsmobile elements, which can move around the deployed fieldand convey the data from each sensor node to the base station,have been studied. The main challenge in these schemes is howto control the mobility of the mobile elements for efficient datagathering while satisfying various constraints (e.g., before bufferis full on each sensor node). More recently, considering theconstraint that the mobile element may not be reachable fromevery sensor node, the hybrid approaches that combine the ideaof multi hop data forwarding and mobile elements have beenexplored. Here, the data is first aggregated locally using multi-hop schemes to some rendezvous points. Then, the mobileelement visits only these rendezvous points to pick the dataup.Note that, in most of the existing studies involving mobileelements, only a single path is calculated for each mobileelement and the same path is followed repeatedly during datagathering. However, such solutions with a single path for themobile element may still lead today’s in WSNs, especially forthe cases where the mobile element needs to collect data directlyfrom every sensor node but it cannot visit the location of allsensor nodes (due to, for example, energy budget of the mobileelement or time limitations).

The sensor nodes that are far away from the path will need totransmit their data to the mobile element at higher power levelsand thus use up their energy budget more quickly. For WSNsthat rely on their entire sensor nodes for normal operations,such uneven energy depletion will lead to limited lifetime ofWSNs. Different from the existing single-path solutions, in ourpreliminary study, we have proposed the idea of exploitingmultiple paths for the mobile element in WSNs to extend theWSN’s lifetime.

Wireless Sensor Networks (WSN) having large number ofsensor nodes which will cover the earth in time to come. SensorNetworks covering vast areas are already in the wild and areinstrumental in ways not possible using preexisting technology.The most important resource on these nodes is the energy supplyand in almost all the cases a battery in the node is responsiblefor supplying energy for the entire lifetime of the node. Thedeployment of nodes could be at inaccessible locations andhence once the battery has drained the node is unusable for thenetwork. Increasing the lifetime of the node becomes a veryimportant factor to bring the technology mainstream. The biggestadvantage in case of WSN is also their biggest limitation, theunconnected nature of the network also limits the life time ofthe network. Energy whole problem limits the total life time bydisconnecting the network from the base node. For many WSNimplementations it is practical to have a mobile base node, wherea mobile base node traverses through the nodes on land or inair to collect data. Instead of communicating to each and everynode the energy consumption can be decreased by using dynamicrouting protocols developed for Ad hoc networking. Maintaina constant path for all the traversals of the base node is equallybad since even this will result in the formation of energy holesin the network. Generating a different path for each traversalwith node energy as constraint will provide an equilibrium inenergy consumption hence giving to the WSN. For the purposeof simulation the network is maintained within the followingconstrains.

1. Sensor nodes are distributed over a large area and may ormay not have a layout in distribution.

2. The distance between two nodes is not greater than thecommunication range of the nodes and does not interferewith the formation of the network.

3. Individual nodes do not communicate to the base unit insteadin a neighborhood a single node acts as an agent between thenodes and the base unit.

4. A mobile base unit is responsible for retrieval of data.Number of sensor nodes is limited and known, for largernumber of nodes the heuristics will have to be updated.

We formally define the mobile element scheduling problem andpresent an integer linear programming (ILP) formulation that



38

can be used to compute optimal solutions for relatively smallinstances. We then discuss two heuristics: one that is based onobtaining a solution of the Travelling Salesman Problem (TSP)and partitioning the resulting tour into smaller ones that meetall the constraints; and another that builds the mobile elementtours in a greedy fashion (tour packing), based on a certain “cost”function for each node. Finally, we evaluate our heuristicalgorithms by comparing their solutions to the optimal ones forsmall instances, as well as by demonstrating that they outperforma heuristic commonly used for the solution of a related vehiclerouting problem with time windows.Recent years have seen thedeployments of wireless sensor networks (WSNs) in data-intensive applications including emergency response, structuralhealth monitoring (SHM), etc. WSNs in these applications oftenproduce high-bandwidth sensor data that need to be collectedunder stringent delay constraints. For instance, SHM sensorsmust sample at higher than 100 Hz and stream the accumulateddata to the base station (BS) every few minutes when the healthof a structure needs to be inspected. On the other hand, sensorsin such applications must operate on limited power supplieslike batteries for extended lifetime up to years. Therefore, afundamental challenge for these WSNs is to support high-bandwidth data collection with minimum network energyconsumption. Several recent works have exploited the use ofWSNs in data collection. In this approach, a small number ofmobile devices referred to as mobile elements (MEs) roam aboutsensing fields and collect data from sensors. As a result,significant network energy saving can be achieved by reducingor completely avoiding costly wireless transmissions. On theother hand, the energy consumption of MEs is less constrainedas they can replenish their energy supplies because of themobility. However, the primary disadvantage of this approachis the increased latency. For instance, the typical speed of severalpractical ME systems (e.g., NIMs and Packbot is about 0.1-1m/s. It is applied in the Gaussian broadcast channels. For agiven transmit power constraint, those points on the boundaryof the capacity region can be regarded as the set of optimaloperational points. It provides high BER but QOS is notachieved in this type of channels.To achieve high transmit powerand overcome transmit power constraint, we should balancetransmit power by balancing the capacity of the channel bydesigning the proper transceiver of WSN. Optimum ratebalancing should be achieved in the downlink of the wirelesscellular system.

II. IMPLEMENTATION

2.1 Hand off Mechanism In Mobile WiMAX

For implementing a mobile network, a handoff mechanism mustbe defined to maintain uninterrupted user communicationsession during his/her movement from one location to another.Handoff mechanism handles subscriber station (SS) switching

from one Base Station (BS) to another. Different handofftechniques have been developed. In general, they can be dividedinto soft handoff and hard handoff.

2.1.1 Soft Handoff

Figure 2.1: Soft Handoff

Soft handoff is used in voice-centric cellular networks such asGSM or CDMA. It uses a make-before-break approach whereasa connection to the next BS is established before a SS leaves anongoing connection to a BS. This technique is suitable to handlevoice and other latency-sensitive services such as Internetmultiplayer game and video conference. When used fordelivering data traffic (such as web browsing and e-mail), softhandoff will result in lower spectral efficiency because this typeof traffic is bursty and does not require continues handover fromone BS to another.

2.1.2 Hard Handoff

Mobile WiMAX has been designed from the outset as abroadband technology capable of delivering triple play services(voice, data, and video). However, a typical Mobile WiMAXnetwork is supposedly dominated by delay-tolerant data traffic.Voice in Mobile WiMAX is packetized (what is called VoIP)and treated as other types of IP packets except it is prioritized.Hard handoff (HHO) is therefore used in Mobile WiMAX. Inhard handoff, a connection with a BS is ended first before a SSswitches to another BS. This is known as a break-before-makeapproach.

Figure 2.2: Hard Handoff

Hard handoff is more bandwidth-efficient than soft handoff,but it causes longer delay. A network-optimized hard handoffmechanism was developed for Mobile WiMAX to keep ahandoff delay under 50 ms. A SS maintains a connection to asingle BS at any given time.

2.1.3 Proposed Energy Efficient Algorithm

Energy saving is a paramount concern in wireless sensornetworks (WSNs). A strategy for energy saving is to cleverly



39

manage the duty cycle of sensors, by dynamically activatingdifferent sets of sensors while non-active nodes are kept in apower save mode. We propose a simple and efficient approachfor selecting active nodes in WSNs. Our primary goal is tomaximize residual energy and application relevance of selectednodes to extend the network lifetime while meeting application-specific QoS requirements. We formalize the problem of nodeselection as a knapsack problem and adopt a greedy heuristicfor solving it. An environmental monitoring application is chosento derive some specific requirements. Analyses and simulationswere performed and the impact of various parameters on theprocess of node selection was investigated. Results show thatour approach outperforms a naïve scheme for node selection,achieving large energy savings while preserving QOSrequirements.

2.2Relay-Assisted Communications System

In the relay-assisted communications system, multiple relaysform a virtual array, and cooperate with one another to work, asshown in Figure 1. The relay-assisted broadband wirelesscommunications networking enables diversified access modes,which is its major difference from legacy wireless accesssystems. Mobile terminals may access wireless networks eitherthrough relay stations directly or by cooperative relaying. Asan effective technology for improving network coverage quality,radio relaying is a high-cost-performance solution to broadbandwireless access at high frequency bands. Generally, the relay-assisted communications system has the following advantagesas a new-type networking technology:

1. Multiple relays can use the same time slots and frequencysimultaneously, which saves radio resources

2. Space diversity and space multiplexing can be used betweenrelays to improve system capacity for transmission

3. It is unnecessary for the system to greatly change theexisting backbone architecture, which will enable thesmooth evolution of live communications networks.

2.2.1 Coverage

Owing to large-scale fading, the data transmission efficiency ofBS-based Cell- structured communications systems gets worsealong with the distance increase between users and BSs.Therefore, users in legacy cellular networks cannot really enjoyhigh data rates at the edge of the networks. Moreover, a BSalways fails to cover every inch of its cells, because signaltransmission is always influenced by geographic conditions,such as buildings in cities and underground environments. Ifrelay stations are deployed in the areas with both weak BScoverage and the cell edges, dead spots will be effectivelyreduced and cell coverage will be expanded through signalrelaying during communications. Compared to legacy cellular

systems, the system with relay stations has wider coverage andbetter communications quality, based on low networking cost.

2.2.3 Transmission

This new wireless network integrating relay into the cellularnetwork may send data to users through BS and relay stationssimultaneously. Accordingly, capacity gains can be obtainedthrough multiplexing or space diversity. Although the datatransmission from BS to user via relay is a two-hopcommunications link, in which the relay requires certainfrequency resources, data transmission to different users canuse different relay stations. This may greatly compensate thecapacity loss caused by the two-hop communications. Whenbuildings and other geographic barriers hinder a transmissionpath from BS to user and lead to large-scale shadow fading, thecapacity loss can even become a gain. With different relayingmodels and different message feedback modes, the number ofrelays in the relay-assisted communications network hasdifferent impacts on system capacity. If both BS and mobileterminal in the system are equipped with M antennas, and ifboth relay and mobile terminal have known channel messages,the system capacity will have a linear increase with M, and alogarithmic increase with the numberof relays.

2.3 Architecture of Relay Stations

Two main points must be understood about RS. First issue iswhether the BS knows about the RS. This means that if the BSneeds to know nothing about RS, then the integration of RSinto the service area is much simpler. No change to the BS isnecessary and there is no special signaling between RS and BS.The RS is a pure helper for the BS. In this situation, RS causesno burden for the BS. Some of the earlier cellular systems suchas GSM, used this kind of RS. They were simply called repeaters.The second point is the characteristic of the RS. Two kinds arepopular: Amplify-and-Forward (A&F), or Decode-and-Forward(D&F). Each has different use and D&F equipment are generallymore expensive than A&F.

2.3.1 Relays

Intelligent relays are an effective technology to achieveimportant deployment tools to provide cost effective methodsof delivering high data rate and avoid coverage holes indeployments areas. In addition, upgrading the networks in orderto support higher data rates is equivalent to an increase of signal-to-interference plus noise ratio (SINR) at the receivers’ front-end. Also, through deployment the network providers have toavoid coverage area holes. A traditional solution to increasethe receiver’s SNR is to deploy additional BSs or repeaters toserve the coverage area holes with required data rates. In mostof the cases, the cost of the BS is relatively high and arrangingbackhauls quickly might be a challenge in serving coverageholes. By now industry has used RF repeaters; however repeater



40

has the problem of amplifying the interference and has nointelligence of signal control and processing. In order to achievea more cost effective solution, relay stations (RS) capable ofdecoding and forwarding the signals from source to destinationthrough radio interface would help operators to achieve higherSINR in cost effective manner. Relay stations do not need awire-line backhaul; the deployment cost of RSs is expected tobe much lower than the cost of BSs. The system performancecould be further improved by the intelligent resource schedulingand cooperative transmission in systems employing intelligentrelays.

2.3.2 Relays Vs Repeaters

They serve to expand the broadcast range of a television orradio station beyond the primary signal’s coverage area, or toimprove service in a part of the main coverage area whichreceives a poor signal due to geographic constraints DeployingRS can improve IEEE 802.16m network in different dimensions.The following figures illustrate the different benefits that canbe achieved by deploying RS within an IEEE802.16m networkand the relay-related connections in IEEE 802.16m. 16m BS(ABS) is a BS capable of acting as a 16m BS as well as a 16eBS. Multi hop relay BS (MRBS) is a 16e BS with 16j RS supportfunctionality. 16m MS (AMS) a MS capable of acting as a 16mMS as well as a 16e MS. Yardstick MS (YMS) is a 16e MS.Advanced RS (ARS) is a 16m RS and RS is a 16j RS.Interconnections between the entities shown in solid lines aresupported by using various protocols such as 16e, 16j, and 16m.There is no protocol specified to interconnections shown indashed lines. Uni-directional zones (e.g. DL Transmit Zone)can exploit scheduling benefits and bi-directional zones (e.g.Network Coding Transmit Zone) can exploit throughput benefitsby using network coding. Relaying is performed using a decodeand forward paradigm, and ARS operates in time-divisiontransmit and receive (TTR) mode. ARSs may operate intransparent or non-transparent mode. Cooperative relaying is atechnique whereby either the ABS and one or more ARSs, ormultiple ARSs cooperatively transmit or receive data to/fromone subordinate station or multiple subordinate stations.Cooperative relaying may also enable multiple transmitting/receiving stations to partner in sharing their antennas to createa virtual antenna array.

2.4 Cooperative Relays

An interesting set of structures for relay stations is calledCooperative Relay. The signal from a base station is picked upby several relay stations, decoded, and forwarded to the mobilestation through different radio paths. This scheme has theadvantage that if one path is poor, another path is likely to makeup for it. The improvement of Bit Error Rate (BER) at the mobilestation is called Cooperative Diversity Gain.Wireless expertsare working on three types of cooperative relays.

1. Same-Signal Cooperative Relay

2. Space-Time Coded Cooperative Relay

3. Hybrid Cooperative Relay

2.4.1 Same Signal Cooperative Relay

The same-signal cooperative relay is the simplest scheme. Here,multiple relay stations pick-up the same signal from the basestation and forward it to the mobile station.

2.4.2 Space Time Coded Cooperative Relay

A slightly different scheme is to use Space-Time Block Coding(STBC) at the base station to enable the relay station to pick-updifferent signals. Here, the base station transmits two copies ofits signal by using two antennas. The relay station can then makethe most of the various received versions of the signal to improvethe BER at the mobile station.

2.4.3 Hybrid Cooperative Relay

It is the combination of previous two cooperative relays. It isthe most complex and provides the highest gain.

2.5 Individual Wireless Sensor Node Architecture

Depending on the sensors to be deployed, the signal conditioningblock can be re-programmed or replaced. This allows for a widevariety of different sensors to be used with the wireless sensingnode. Similarly, the radio link may be swapped out as requiredfor a given applications wireless range requirement and the needfor bidirectional communications. The use of flash memoryallows the remote nodes to acquire data on command from abase station, or by an event sensed by one or more inputs to thenode. Furthermore, the embedded firmware can be upgradedthrough the wireless network in the field.The microprocessorhas a number of functions including:

1) Managing data collection from the Sensors.

2) Performing power management functions.

3) Interfacing the sensor data to the physical Radio layer.

4) Managing the radio network protocol.

A key feature of any wireless sensing node is to minimize thepower consumed by the system. Generally, the radio subsystemrequires the largest amount of power. Therefore, it isadvantageous to send data over the radio network only whenrequired. This sensor event-driven data collection modelrequires an algorithm to be loaded into the node to determinewhen to send data based on the sensed event. Additionally, it isimportant to minimize the power consumed by the sensor itself.Therefore, the hardware should be designed to allow themicroprocessor to judiciously control power to the radio, sensorand sensor signal conditioner.



41

III. INDENTATIONS AND EQUATIONS

MIMO–TWO TRANSMITTING ANTENNAS AND TWORECEIVING ANTENNAS (Tx=2, Rx=2)

Figure3.1: MIMO (Tx =2, Rx=2)

The received signal in the first time slot is,

Assuming that the channel remains constant for the second timeslot, the received signal is in the second time slot is,

Where,

Y11 and Y

22 are the received information at time slot 1 on receive

antenna 1, 2 respectively,

Y12 and Y

21 are the received information at time slot 2 on receive

antenna 1, 2 respectively,

hijis the channel from ithreceive antenna to jthtransmit

antenna,X1,X

2are the transmitted symbols, n

11 and n

21 are the

noise at time slot 1 on receive antenna 1, 2 respectively and n1

2

and n2

2are the noise at time slot 2 on receive antenna 1, 2respectively.

Combining the equations at time slot 1 and 2,

Also,

To solve for x1 x

2, we know that we need to find the inverse

of .

We know, for a general m x n matrix, the pseudo inverse isdefined as,

.

The Term,

Since this is a diagonal matrix, the inverse is just the inverse ofthe diagonal elements, i.e

The estimate of the transmitted symbol is,

Multi BS MIMO

Multi-BS MIMO techniques are supported for improving sectorthroughput and cell-edge throughput through multi-BScollaborative precoding, network coordinated beam forming,or inter-cell interference nulling. Both open-loop and closed-loop multi-BS MIMO techniques can be considered. For closed-loop multi-BS MIMO, CSI feedback via codebook basedfeedback or sounding channel will be used. The feedbackinformation may be shared by neighboring BSs via networkinterface. This places significant obligation in low latencybackhauls. COMP - Coordinated multi-point (CoMP) is a newclass of transmission schemes for interference reduction in the16m technology. Enabling features such as networksynchronization, cell- and user-specific pilots, feedback of multicell channel state information and synchronous data exchangebetween the base stations can be used for interference mitigationand for possible macro diversity gain. The collaborative MIMO(Co-MIMO) and the closed-loop macro diversity (CL-MD)techniques are examples of the possible options. For downlinkCo-MIMO, multiple BSs perform joint MIMO transmission tomultiple MSs located in different cells. Each BS performs multi-user precoding towards multiple MSs, and each MS is benefitedfrom Co-MIMO by receiving multiple streams from multipleBSs. For downlink CL-MD, each group of antennas of one BSperforms narrow-band or wide-band single-user precoding withup to two streams independently, and multiple BSs.

IV. SIMULATION RESULTS

MATLAB IMPLEMENTATION

● Generate random binary sequence of +1’s and -1’s.

● Group them into pair of two symbols.

● Code it per the antenna, multiply the symbols with thechannel and then add white Gaussian noise.

● Equalize the received symbols.



42

● Perform hard decision decoding and count the biterrors.

● Repeat for multiple values of Eb/N

0and plot the

simulation and theoretical results.

● To select the Energy efficient nodes SVD algorithm

(H= U.S.V*) is proposed.N-min( Tn , Rn ) vectors are

Rx Eigen Modes excited by the transmitter.

● Each node should have Kalman filter which is a linearoptimal filtering approach which helps in node mobilitymanagement and node tracking.

● A Kalman filter combines available measurement data,plus prior knowledge about the system and measuringdevices, to produce an estimate of the desired variablesin such a manner that the error is minimizedstatistically.

● The Capacity of each nodes is calculated usingShannon Formula (C=log2 (1+(S/N)).

It tell about the performance analysis of this proposed algorithmensures its effectiveness thereby providing Efficient Mobility,Low cost, higher bandwidth, maximum distance coverage, LowBit error rate, High Signal to Noise ratio, Reduced ISI, HighData Rate simulated in MATLAB. These are the process donein that MATLAB Simulation Work and their results are furnishedbelow.

0 5 10 15 20 210

-5

10-4

10-3

10-2

10-1

SNR

BE

R

BER Vs SNR for 2Tx and 2Rx

T1 to R1

T1 to R2T2 to R1

T2 to R2

0 2 4 6 8 10 12 14 16 18 200

2

4

6

8

10

12SNR Vs Capacity

SNR in dB

Cap

acity

of

the

Nod

e

Optimization

Node1

Node2Node3

Node4

8 10 12 14 16 18 20 22

2

4

6

8

10

12

14

Data sent through the Nodes - Throughput

nodes

effic

ienc

y

8 10 12 14 16 18 20 22

2

4

6

8

10

12

14

Shortest Path Efficiency

nodes

effic

ienc

y

8 10 12 14 16 18 20 22

2

4

6

8

10

12

14

Energy efficient and shortest path Efficiency

nodes

effic

ienc

y

0 2 4 6 8 10 12 14 1620

30

40

50

60

70

80

90

100Energy Efficient Nodes

Nodes

Dis

tanc

e

0 50 100 150 200 250 300 350 40013.34

13.36

13.38

13.4

13.42

13.44

13.46

13.48

13.5

13.52

angle in degrees

dist

ance

variation in distance with angle

0 5 10 150

10

20

30

40

50

60

70

80Distance between Nodes

Number of Nodes

Dis

tanc

e

0 50 100 150 200 250 300 350 400 450 5000

2

4

6

8

10

12

14

16Time Calculation

Time in msec

Num

ber

of N

odes

0 50 100 150

0

50

100

150

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

Coverage Area of Nodes

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

1 23

4

5

6

7

8

9

10

11

12

13

14

15

16

V. CONCLUSION

Wireless Sensor Networks composed of low cost, low-power,multifunctional sensor nodes. Applications of such WSN aremedical treatment, environmental monitoring, outer-spaceexploration, emergency response, etc. senor nodes distributedover the field may act as information SOURCES. There istypically one or more SINKnodes for whom the measured dataare destined to, which is located within or outside the sensingfield. In order to improve energy efficiency, another efficientway is to maximize throughput without consuming much morepower of nodes. MIMO (multiple-input and multiple-output),which is a multiple-antenna technique, is regarded as one of themost promising solutions for improving spectrum efficiency andincreasing capacity of wireless systems. If we apply MIMOtechnique to the sensor network, it will enable more than onesensor nodes to send dissimilar information to the sink nodesimultaneously with energy efficiency. The observationsmotivate us to address the energy-efficient data gatheringproblem by simultaneously introducing mobility and MIMOcapability to the sink nodes. The algorithm motivates us toaddress the energy efficiency and the scalability of the datagathering scheme can be greatly improved.

A mobile sink can be a robot or a vehicle equipped withadvanced transceivers, sufficient power and large memory. Herethe mobile sink has multiple antennas. It can concurrentlyreceive data from multiple sensor nodes, which will dramaticallyreduce the gathering time by reducing the energy. The networkcooperation between BS and multiple relays is proposed forimproving communication capacity. In order to fulfill thecooperation, distributed schemes are necessary for legacyphysical-layer-based multi-user MIMO technologies (such asdirty-paper coding, linear precoding and decoding, multi-userdetection and STC) to implement cooperative data transmissionamong various nodes in the network. Accordingly, MAC-layer-oriented cooperation strategies seem extremely important. Therelay-assisted communications system can use higher degreeof freedom to improve resource allocation and optimize networkperformance. However, it also brings many non-convexoptimization problems, which cannot be solved by traditionaloptimization algorithms. In fact, the heuristic interactive



43

optimization and greedy search algorithms can achieve a goodcompromise between performance and complexity ofcomputation.

It is also very important to synchronize BSs, RSs and multipleusers in the relay-assisted communications system. Moreover,robust distributed STC designed in the condition of reducinginter-relay message exchanges is an effective technology formaking full use of multi-relay space gains.Relay-based multi-hop transmission system has attracted much attention.

Future Enhancements

In the near future, sensor devices will be produced in largequantities at a very low cost and densely deployed to improverobustness and reliability. This project may be extended byintroducing the concept of multiple-point-to-multiple-point,which enables free communications between any two nodesin network to fulfill quicker, convenient and economical datatransmission which involves the automatic mapping andimproved power saving.

REFERENCES

[1] Multi-sector crisis management consortium (mscmc).

[2] S. R. NingXu, Krishna Kant Chintalapudi Deepak Ganesan, andR. G.Alan Broad, and Deborah Estrin, “A wireless sensor networkfor structuralmonitoring,” in Proceedings of the 2nd internationalconference onEmbedded networked sensor systems (SenSys), pp.13-24, Aug 2004.

[4] A. Mainwaring, J. Polastre, R. Szewczyk, D. Culler, andJ.Anderson,“Wireless Sensor Networks for Habitat Monitoring,”in Proceedings ofthe 1st ACM international workshop on Wirelesssensor networks andapplications(WSNA), pp. 88 - 97, Sep 2002.

[5] A. Somasundara, A. Ramamurthy, and M. B. Srivastava, “Mobileelementscheduling for efficient data collection in wireless sensornetworks withdynamic deadlines,” in Proceedings of the 25thIEEE International Real-Time Systems Symposium (RTSS),pp. 296-305, Dec 2005.

[6] M. Solomon, “Algorithms for the vehicle routing and schedulingproblemswith time window constraints,” Operations Research,vol. 35, pp. 254-265, 1987.

[7] P. Juang, H. Oki, Y. Wang, M. Martonosi, L. Peh, andD. Rubenstein,“Energy-Efficient Computing for WildlifeTracking: Design Tradeoffs andEarly Experiences with Zebranet,”in Proceedings of the 10th InternationalConference onArchitectural Support for Programming Languages andOperatingSystems, pp. 96-107, Oct 2002.

[8] Y. Gu, D. Bozdag, E. Ekici, F. Ozguner, and C.-G. Lee,”Partitioning basedmobile element scheduling in wireless sensornetworks,” in Proceedingsof the Second Annual IEEECommunications Society Conference onSensor and Ad HocCommunications and Networks (SECON), pp. 386-

395, Sep 2005.

Author’s Profile

Mr.S. RAVI KUMAR, pursuing PhD inJNTUH under the guidance of Dr. D.SrinivasaRao HOD ECE JNTUH. Post Graduated inElectronics and communication Engineering(M.Tech) From Andhra University,Visakhapatnam in Mar- 2005 and Graduated inElectronics and communication Engineering(B.Tech) from JNTU, in 2002. He is working as An AssociateProfessor in Department of Electronics and CommunicationEngineering in St.Martin’s Engineering College, R.R Dist,AP and India. He has 8+ years of Teaching Experience. HisResearch area in PhD is QOS Energy Efficient Routing inWireless Sensor Networks.

Subscription Charges

India (2 Years)

Individual : Rs 1500/-

Universities/Colleges/Institutions : Rs 2500/-

All Other Countries (2 Years)

Individual : US $100

Universities/Colleges/Institutions : US $200

Subscriptions and further Enquiries

K.Niranjan ReddyManaging Editor,IJRSCSDProf. & HOD, ECE Dept.Malla Reddy Engineering College for WomenMaisammaguda,[email protected]

Authors should submit only papers that have been carefully proof read and polished. Before submission please makesure that your paper is prepared using the IEEE paper template. This will ensure fast processing and publication.Acceptance or rejection notification will be sent to all authors.

Submission of the manuscript represents that the manuscript has not been published previously and is not consideredfor publication elsewhere. Authors would be required to sign a CTA form (Copyright Transfer Agreement) once themanuscript is accepted which would be sent to the corresponding author’s email. The corresponding author can downloadthe form and after getting authors and co-authors signature it can be sent as an attachment file after scanning [email protected].

Publication Charges:There is no charge for the processing of paper but author(s) of each accepted paper is required to pay thepublication charge which is very nominal fees. One Hard copy will be given to main author. AdditionalCopy can also be obtained at an Extra charge.

Category Indian Authors International Authors

Main Author (1) Rs 2000/- US $100

Co- Author (maximum 2) Rs 1500/- US $50

Guide Free Free

MALLAREDDY ENGINEERING COLLEGE FOR WOMEN Permanently affiliated to JNTUH, Approved by AICTE

Accredited by NAAC with ‘A’ Grade, ISO 9001:2008 Certified Institution Maisammaguda, Dhulapally, Secunderabad-500 100

Vision● Visualizing a great future for the intelligentsia by imparting state-of the art Technologies in the field of

Engineering and Technology for the bright future and prosperity of the students.● To offer world class training to the promising Engineers

Mission● To nurture the high level of Decency, Dignity and Discipline in women in attaining high intellectual abilities.● To produce employable students at National and International levels by effective training programs.● To create pleasant academic environment for generating high level learning attitudes

Special Achievements and Innovative Practices● Accredited by NAAC with ‘A’ Grade, UGC, Govt. of India.● Permanently Affiliated to JNTUH● ISO 9001:2008 Certified Institution.● Received Engineering Educators Award from University of Bradford, U.K● MOU with TECH MAHINDRA.● MOU with ZENSAR Technologies Ltd. for Internships.● More than 80% students are placed in MNCs like L&T Technology Services, Tech Mahindra, NTT

Data, IBM ,HCL Semantic Space, Global Edge Technologies, Infosys, CSC, Unisys, ITC InfoTech,Virtusa, Emphasis, In Rhythm Solutions, Genpact, iGate, Celigo, Amazon and ADP.

● Microsoft Innovation Centre - 1st Women’s campus in India.● Cisco International Certification Authorized Centre.● Oracle Academic Partner for JAVA Certification.● Business English Certification (BEC) Centre, University of Cambridge, U.K● Research and Development Cell for effective research.● Active Industry Institute Partnership Cell.● Entrepreneurship Development Cell to enhance Entrepreneurial Skills.● Centre for Development of Communication Skills (CDCS) for Soft Skills Enhancement.● Centre for Human Excellence in association with Ramakrishna Matt, Hyderabad for personality development.● Global Education and career Counseling Centre to promote International Education.● Finishing School Training programmes for Final Year Students for technology enhancement.● TCS Academic Interface Programme Centre.● Life member of ISTE.● Member of CSI, DELNET & HMA.● IEEE, IETE, ISTE & CSI Student Chapters.● Employability Enhancement Training Programmes for II, III & IV yr students in association with Time Institute.● Regular Conduction of Mock Placements Drives for I, II, III and IV years.● Project Based Training for I, II, III and IV B.Tech Students.● National & International Conferences, Workshops and Refresher Courses.● 80% Faculty ratified by JNT University, Hyderabad.● Inspirational Lectures by persons from Reputed Universities and Industries.● A View Online Video Lectures-Recognized Centre by JNTUH.● NPTEL Video Lectures● Course Files and Lab Manuals for Effective Instruction.● LCD Projectors in every classroom to facilitate effective Teaching-Learning.● State of the Art Laboratories.● Duplex Central Library.● Mentoring and Counseling.

Dr.Y.Madhavee LathaB.Tech, M.Tech, Ph.D.

Principal & Editor-in-Chief IJRSCSD

research in signal processing, computing · 2015-05-16 · research in signal processing, computing...

Documents