distributed antenna-based epon-wimax integration and its cost-efficient cell planning

808 IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL. 28, NO. 6, AUGUST 2010

Distributed Antenna-Based EPON-WiMAXIntegration and Its Cost-Efficient Cell Planning

Min-Gon Kim, Gangxiang Shen, and JungYul Choi, Members, IEEE, Bokrae Jung, Student Member, IEEE,Hong-Shik Park, Member, IEEE, and Minho Kang, Senior Member, IEEE

Abstract—Achieving the benefits of high-capacity of opticalnetworks and the mobility feature of wireless networks leads tointegrate EPON and WiMAX for a promising broadband accesssolution. To efficiently put both benefits together, we proposean integration architecture of EPON-WiMAX based upon aDistributed Antenna (DA) environment, where collaborative BaseStations (BSs) concurrently transmit same wireless downlinksignals (specifically for multicast and broadcast services (MBSs))to Mobile Stations (MSs) in overlapped cell coverage areas. Ithelps enlarge the region of the available network coverage areaby increasing Signal to Interference and Noise Ratio (SINR)in overlapped cell coverage areas through cooperation betweenOptical Line Terminal (OLT) and Optical Network Unit (ONU)-BS. We also present an cost-efficient cell planning to optimallycontrol the size of overlapped cell coverage areas for the proposedDA-based integration architecture with a case study undera required region of the available network coverage area inconsideration of the number of ONU-BSs and the distancebetween ONU-BSs. Performance evaluation results show thatthe proposed DA-based integration architecture enhances costefficiency compared to the Traditional Antenna (TA) (non-DA)-based integration architecture with a similar level of spectralefficiency of MSs.

Index Terms—EPON, WiMAX, integrated network, dis-tributed antenna, cell planning, cost analysis.

I. INTRODUCTION

THE MODERN broadband access systems for high band-width and operational efficiency has been considered im-

portantly due to the growing popularity of quad-play services(e.g., video, voice, data, and mobility to a Mobile Station(MS)). As key technologies for worldwide Fiber To The Home(FTTH), Fiber-Wireless (FiWi) access technique variants arebeing taken into consideration by utilizing the merits of bothhigh-capacity of optical communications and mobility featureof wireless communications. Among optical and wireless com-munications advances, the Ethernet Passive Optical Networks

Manuscript received 18 August 2009; revised 23 January 2010. Part of thiswork was first conducted at University of Melbourne when he was there asa visiting research engineer.M.-G. Kim is with the public & original technology research center, Daegu

Gyeongbuk Institute of Science & Technology (DGIST), Daegu, Republic ofKorea (e-mail: [email protected]).G. Shen was with University of Melbourne when he did this work,

and now he is with Ciena Corporation, Baltimore, America (e-mail:[email protected]).J. Choi is with Network R&D Laboratory, KT Corporation, Daejeon,

Republic of Korea (e-mail: [email protected]).B. Jung is with the Department of Information and Communications

Engineering, Korea Advanced Institute of Science and Technology (KAIST),Daejeon, Republic of Korea (e-mail: [email protected]).H.-S. Park and Minho Kang are with the Department of Electrical Engineer-

ing, Korea Advanced Institute of Science and Technology (KAIST), Daejeon,Republic of Korea (e-mail: [email protected], [email protected]).Digital Object Identifier 10.1109/JSAC.2010.100806.

(EPON) [1]-Worldwide Interoperability for Microwave Access(WiMAX) (IEEE 802.16) [2] integration has been regarded asa promising solution for achieving the aforementioned featuresby implementing an integrated box, so called Optical NetworkUnit (ONU)- Base Station (BS) including an EPON ONU andan WiMAX BS. Fundamentally, they are well-matched witheach other on account of their service objectives, multipleaccess mechanisms, and capacity hierarchy (e.g., each ONUpossessing about 65Mbps bandwidth (in case of the opticalsplit ratio = 1:16) is matched with the total capacity offeredby a single BS in WiMAX systems (=70Mbps over a 20MHzchannel)) [3].

Fundamentally, EPON systems [1], a part of the EPON-WiMAX integration, can offer high bandwidth but an weak-ness on the cost aspect due to relatively expensive deploymentcost. Particularly, the total length of the deployed optical fibershas to be taken into account more important than other systemcomponents, because the total deployment cost of EPONsystems is mostly affected by the cost of trenching and layingoptical fibers [4]. On the other hand, WiMAX systems [2],the other part of the EPON-WiMAX integration, can lowerdeployment cost and support mobility under limited wirelesstransmission performance due to spectrum, shadowing, fading,and InterCell Interferences (ICIs). Specifically, ICI is interfer-ence from other wireless signals of other neighbor BSs, whichcould lower the Signal to Interference and Noise Ratio (SINR),and hence it can have a bad effect on wireless downlinktransmission performance in both aspects of the achievabledata rate of an MS and the available size of cell coveragearea.

In order to mitigate ICI in the EPON-WiMAX integrationarchitecture, we adopt a Distributed Antenna (DA) environ-ment, where collaborative BSs concurrently transmit samewireless downlink signals (specifically for Multicast BroadcastServices (MBSs)) to MSs and thus increased SINR in over-lapped cell coverage areas can be achieved. Consequently, itcan enhance the wireless downlink transmission performance(e.g., the available network coverage area, the achievable datarate of MSs, and ICI [5], [13]). In particular, under a lim-ited region of the integrated network, the more collaborativeONU-BSs produce the better wireless downlink transmissionperformance, but it inevitably demands a longer total length ofthe deployed optical fibers and thus a higher total deploymentcost of the network has to be followed [4], [5]. Owingto this tradeoff relationship between the wireless downlinktransmission performance and total deployment cost of theintegrated network, a delicate cell planning is required for costefficient network design and dimensioning [5], [6]. Thus, we

0733-8716/10/$25.00 c© 2010 IEEE

Authorized licensed use limited to: Korea Advanced Institute of Science and Technology. Downloaded on August 10,2010 at 02:20:30 UTC from IEEE Xplore. Restrictions apply.

KIM et al.: DISTRIBUTED ANTENNA-BASED EPON-WIMAX INTEGRATION AND ITS COST-EFFICIENT CELL PLANNING 809

develop a cost-efficient cell planning in consideration of (i)the number of ONU-BSs and (ii) the distance between ONU-BSs to optimally control the size of overlapped cell coverageareas for the proposed DA-based integration architecture. It isused to conduct a case study regarding spectral efficiency ofMSs and total deployment cost of the integrated network undervarying the required region of the network coverage area.Performance evaluation results substantiate that the proposedDA-based integration architecture enhances cost efficiencycompared to the TA-based integration architecture with asimilar level of spectral efficiency of MSs.To achieve this end, this paper is structured as follows.

Related works regarding the integration of FiWi access tech-niques and DA environments are described in Section 2.Section 3 introduces the integration architecture based uponthe DA environment on the aspect of network coverage areacompared to the TA environment. The problem definition, andtransmission and cost models for evaluating the performanceof the proposed DA-based integration architecture compared tothe TA-based integration architecture are presented in Section4. Based upon these models, performance evaluation includingcomparison between the DA and TA environments and a casestudy is shown in Section 5. Finally, Section 6 concludes thispaper.

II. RELATED WORKS

Much research effort has been carried out on the integra-tion of FiWi access techniques in viewpoint of the physicallayer [7], the integrated MAC layer pertaining to bandwidthallocation and scheduling [8]–[10], wireless communicationsnetwork perspectives [11]–[13], and architecture and scenariosperspectives [3], [14]. First, in the physical layer, a newmodulation scheme, so called Radio-over-Fiber (RoF) wasintroduced by using a single modulator to provide both wiredand wireless services concurrently [7]. It made it possibleto physically reduce the gap between optical and wirelesssides. However, the RoF scheme is not fully standardizedyet and not widely deployed to date. Second, in the MAClayer, an optimal utility-based bandwidth allocation schemefor video-on-demand (VoD) services [8] and a QoS-awaredynamic bandwidth allocation (DBA) scheme under the con-cept of a Combined ONU-BS (COB) [9] were proposed overan integrated optical and WiMAX networks. In addition, aQoS-aware scheduling scheme in a Hybrid ONU-BS (HOB)architecture was investigated [10]. Bandwidth allocation andscheduling in the MAC layer is well advanced for support-ing the defined QoS types in optical and wireless networkssimultaneously. Third, in wireless communications networkperspectives, different types of FiWi networks focusing onthe placement of mesh points were proposed in WirelessMesh Networks (WMN) [11], [12]. Besides, advanced codingtechniques (e.g., Space Time Coding (STC) [23]) for improv-ing perceived SINRs of MSs for better wireless downlinktransmission performance were adopted for IPTV services[13]. Finally, the architecture issues arisen in the integrationof EPON and WiMAX were insightfully discussed in [3], andseveral scenarios for the integration of FiWi access techniquesin an Independent ONU-BS (IOB) architecture was proposed[14]. As presented in the aforementioned studies, it is obvious

that there have been some considerations for reducing the gapbetween optical and wireless sides but not a consideration fordeployment cost management, one of the most important fac-tors for practical success to deploy broadband access networks.Only recently were there studies that tried to minimize the totaldeployment cost of PON networks [4], [22]. Thus, minimizingthe total deployment cost of the integrated network need to beconsidered importantly.For performance enhancement of the EPON-WiMAX in-

tegration, it is available to apply some research studies onenhancing DA techniques in a wireless environment [13]including two types of transmission: Cooperative Single CellTransmission (CSCT) [15], [16] that ONU-BSs transmit dataonly to User Equipments (UEs) within their cell coverage bycontrolling the transmission according to the transmissionsof other ONU-BSs, and Cooperative Multi-Cell Transmission(CMCT) [17], [18] that data is routed to and transmitted byother ONU-BSs. DA techniques can expand the available sizeof the network coverage area by letting collaborative BSswith DAs concurrently transmit wireless downlink signals toMSs for mitigation of ICI in overlapped cell coverage areas[5], [13]. However, under a limited region of the integratednetwork, the more collaborative ONU-BSs produce the betterwireless downlink transmission performance, but it inevitablydemands a longer total length of the deployed optical fibersand thus a higher total deployment cost of the network hasto be followed [4], [5]. Hence, it is essential to considertotal deployment cost because of the tradeoff relationshipbetween those in the integrated networks. As a consequence,an cost-efficient cell planning, which reduces both the numberof ONU-BSs and the distance between ONU-BSs in theintegrated network for the reduction of the total deployedlength of optical fibers, is imperative to create synergisticeffects on enhancement of cost efficiency [5]. Therefore, inthe following sections, the DA-based integration architectureand how to plan cells with a newly proposed cost efficiencyevaluator will be presented.

III. DISTRIBUTED ANTENNA-BASED INTEGRATIONARCHITECTURE

A. Components and Operations

The major components of an integrated EPON-WiMAXnetwork include an OLT, an optical splitter, optical fibers,and ONU-BSs. The OLT located in the Central Office (CO)is connected with multiple ONU-BSs via an optical splitter.It has powerful computational capability to perform trafficmanagement to and from ONU-BSs, including a classifier todivide packets into their corresponding queues and a schedulerto schedule packets in the queues. The optical fibers from theOLT are connected to the splitter and fans out to multipleoptical drop fibers connected to ONU-BSs for supporting theconnected ONU-BSs (typically, the optical split ratio rangesfrom 1:4, 1:8, 1:16, 1:32, 1:64, to up to 1:128; namely, a 1:Noptical splitter means that up to N ONU-BSs can be connectedto the splitter), as illustrated in Fig. 1. Due to the power budgetin the EPON, the limited number of splitting ratio may beallowed, which directly influence on the maximum number ofONU-BSs in the EPON.



OpticalLine

Terminal(OLT)

ONU-BS0

ONU-BS2

ONU-BS1

ONU-BS3 ONU-BS4

ONU-BS5 ONU-BS6

20km

1:N

Splitter

Cell coverage area

Optical line

Fig. 1. Distributed Antenna (DA)-based integration architecture for broadcast access services.

The ONU-BS acting as a hybrid device, the key componentof the EPON-WiMAX integration, includes the functions of anONU in EPON systems and a BS in WiMAX systems so that itreduces the gap between EPON systems and WiMAX systems.In the literature survey, there are three types of ONU-BSs:(i) Independent ONU-BS (IOB) [3], which directly connectsan ONU to a BS and thus lets the scheduling performedindependently with lower device cost (i.e., the IOB can notguarantee the required Service Level Agreement (SLA) forusers and bandwidth efficiency); (ii) Combined ONU-BS(COB) [9], which logically connects an ONU to a BS witha WiMAX-EPON (WE) bridge, a physical joint controller(JC), and thus lets the scheduling performed centrally (i.e.,the COB can guarantee the required SLA for users andbandwidth efficiency); and (iii) Hybrid ONU-BS (HOB) [3],which physically connects an ONU to a BS with a logicalJC, and thus lets the scheduling performed centrally (i.e., itseffects are same with the COB and lower device cost canbe achieved than the COB). For tht DA-based integrationarchitecture, all types of the ONU-BSs can be adopted,because the features of those are only related to cost, QoSmapping between EPON and WiMAX for guaranteeing SLA,and bandwidth utilization perspectives, but not the wirelessdownlink transmission performance.

In order to implement a DA environment on the integrationarchitecture, a set of cooperations between an OLT and ONU-BSs has to be defined. Basically, BS-user (BU) association(between each MS (user) and a set of ONU-BSs) and rateassignment (selected by the OLT about transmission ratedefining the modulation scheme taken by ONU-BSs) needto be implemented. Moreover, when an ONU-BS recognizesthat an MS is leaving (or subject to a bad channel for someperiod of time) depending on the minimum perceived powerat the MS that an ONU-BS is responsible for, it issues atear-down request to the OLT. Then, for the power breakdown mechanism, the OLT will reconfigure the modulationscheme and power breakdown for the remaining associatedONU-BSs of the MS. To achieve the periodic reconfiguration,the OLT periodically collects a report from each ONU-BS,which contains a list of all the active MSs identified by the

ONU-BS with their relatively long-term SINR levels. TheOLT then performs BU reassociation and rate reassignment,as well as power breakdown. Each ONU-BS dynamicallyadjusts power for the MSs to compensate for the short-termchannel condition fluctuation on account of mobility and anypossible turbulence. This is referred to as dynamic powerallocation, and has to be designed jointly with the schedulingpolicy. Therefore, based on these operations, the integrationarchitecture can support DA techniques.

B. Benefit of a Distributed Antenna (DA) environment onnetwork coverage area

The main benefit of a DA environment is to enable toexpand network coverage area, depicted in Fig. 2. Differingfrom a TA environment, where a single associated ONU-BStransmits a wireless downlink signal to an MS, a DA envi-ronment is based upon a Space Time Coding (STC) technique[23]. It is a method employed to improve the reliability ofwireless downlink transmission in wireless communicationssystems using multiple DAs of collaborative ONU-BSs [13].It transmits multiple and redundant copies of a data streamto the MS between transmission and reception in a goodenough state to allow reliable decoding. Therefore, MSs inoverlapped cell coverage areas of several collaborative ONU-BSs can take much more aggressive modulation and codingratio (i.e., much better SINRs of MSs can be achieved) dueto the concurrently transmitted wireless downlink signals ofcollaborative ONU-BSs. Then, the achievable data rates ofMSs can be significantly increased, although DA techniqueswill definitely consume resources from all of the collaborativeONU-BSs. In addition, the available network coverage areacan be expended under a DA environment compared to a TAenvironment, because some parts out of the network coveragearea could be transferred to a part of the network coveragearea. This mostly comes from the fact that the SINRs of MSsat the edges of the overlapped cell coverage areas increase, asdepicted in Fig. 2. Consequently, for cost efficient deploymentof the integrated network under a required region of networkcoverage area, these positive effects of a DA environment onthe wireless downlink transmission performance, specifically



ONU-BS 1 ONU-BS 2

MS0 at Spot 0

MS2 at Spot 2

Expanded network coverage areasSpace Time Code (STC) 0, 1, and 2 are orthogonal

STC2STC1

ONU-BS 0

MS1 at Spot 1

STC0

Fig. 2. Downstream data transmission operation and effects of a DAenvironment on network coverage area.

the expansion of the available network coverage area, can beapplicable to the DA-based integration architecture.The detailed examples of the wireless downlink transmis-

sion operation regarding the achievable data rate of an MSunder DA and TA environments are as follows. Basically,the wireless downlink transmission operation under a DAenvironment in non-overlapped cell coverage areas is thesame as that under a Traditional Antenna (TA) environment;namely, the key difference between DA and TA environmentscomes from the transmission in overlapped cell coverage areas.According to the location of an MS, the transmission operationcan be categorized into three: 1) Spot 0, where MS0 is in anon-overlapped cell coverage area; 2) Spot 1, where MS1 isin an area overlapped by the cell coverage areas of two ONU-BSs; and 3) Spot 2, where MS2 is in an area overlapped bythe cell coverage areas of three ONU-BSs as depicted in Fig.2.1) MS0 at Spot 0 in a non-overlapped cell coverage area

supported by a single ONU-BS: Since there is no differencebetween the wireless donwlink transmission operations underDA and TA environments at Spot 0 due to the fact thatthere is only one associated ONU-BS (i.e., ONU-BS0 isassociated), the achievable data rate of MS0 at Spot 0 underboth environments (DRDA

0 and DRTA0 ) is given by:

DRDA0 = DRTA

0 = B · log2(1 +P0

N0). (1)

where B is the bandwidth of the channel in Hertz (Hz).2) MS1 at Spot 1 in an area overlapped by the cell coverage

areas of two ONU-BSs: Since ICI due to power leakage fromone cell to another in an overlapped cell coverage area undera TA environment can have a negative impact on the SINRsof MSs, such as the MS1 at Spot 1. Thus, the achievable datarate of MS1 at Spot 1 (DRTA

1 ) under a TA environment isgiven by:

DRTA1 = B ·(log2(1+

P0

P1 + N0)+log2(1+

P1

P0 + N0)). (2)

On the opposite, under a DA environment, the DAs of col-laborative ONU-BSs joins the wireless downlink transmissionto MS1, and thus the achievable data rate of MS1 at Spot 1(DRDA

1 ) is achieved by:

DRDA1 = B · log2(1 +

P0 + P1

N0). (3)

3) MS2 at Spot 2 in an area overlapped by the cell coverageareas of three ONU-BSs: Similar to the case at Spot 1, otherwireless signals are regarded as a part of ICI. Then, theachievable data rate of MS2 at Spot 2 under a TA environment(DRTA

2 ) is given by:

DRTA2 = B · (log2(1 +

P0

P1 + P2 + N0) + (4)

log2(1 +P1

P0 + P2 + N0) + log2(1 +

P2

P0 + P1 + N0)).

Unlikely, the achievable data rate of MS2 at Spot 2 undera DA environment (DRDA

2 ) is given by:

DRDA2 = B · log2(1 +

P0 + P1 + P2

N0). (5)

IV. MATHEMATICAL MODELS FOR CELL PLANNING

Cell planning in the integrated network is for achievingtarget performance gains under limited environments includingthe optimal placement and number of network devices, andfiber layout (the number of ONU-BSs (n) and the distancebetween ONU-BSs (d)). Particularly, under a limited regionof the integrated network, the more collaborative ONU-BSsproduce the better wireless downlink transmission perfor-mance, but it inevitably demands a longer total length of thedeployed optical fibers and thus a higher total deploymentcost of the integrated network has to be followed [4], [5].Therefore, in order to evaluate the total deployment cost andthe total achievable data rate supported by the integratednetwork, the following models (e.g., IV.A. cost model forthe integration architecture and IV.B. transmission model forWiMAX systems) will be presented, and then a cost-efficientcell planning method will be shown with a new cost efficiencyevaluator.

A. Cost Model for the Integration Architecture

The total deployment cost for the integrated network in-cluding the OLT, the splitter, ONU-BSs, and optical fibersis given by the following factors: (a) the cost of the OLT,(COLT ) including uplink line card costs, downlink line cardcosts, fixed costs for a router, and site rentals; (b) the cost oftrenching and laying fibers between the OLT and the splitter(per kilometer) (CA

inst) including ducts and tubes; (c) the costof the optical splitter (Cs); (d) the cost of trenching and layingfibers between the splitter and ONU-BSs (CB

inst) includingducts and tubes; and (e) the cost of ONU-BSs (CN ) includingsite rentals. Since there is no relationship between the wireless



downlink transmission performance and the locations of theOLT and the splitter, optimizing the locations are not includedin a part of the cost model, and thus the location of splitteris voluntarily set to the center of in the integrated network.With the aforementioned considerations and assumptions, amathematical model is presented as follows [4].• Parameters:◦ x0, y0: Position coordinates of the OLT.◦ xs, ys: Position coordinates of the splitter.◦ xi, yi: Position coordinates of the ith ONU-BS.◦ α: The cost of the OLT including uplink line card costs,

downlink line card costs, router costs, and site rentals. Forsimplicity, the cost of every line card is identical.◦ β: The cost of each splitter port. For simplicity, the cost

of every outlet port is identical.◦ γ: The cost of an ONU-BS including site rentals.◦ δ: The cost of trenching and laying fibers (per kilometer).

For simplicity, fibers are deployed under roads and thus siterentals for fibers are not included.◦ θ: The cost of fiber cable including ducts and tubes (per

kilometer).• Variables:◦ DCO (=

√(x0 − xs)2 + (y0 − ys)2): The distance from

the OLT to the splitter.◦ Di (=

√(xs − xi)2 + (ys − yi)2): The distance from the

splitter to the ith ONU-BS.Depending on the number of ONU-BSs and the distance

between ONU-BSs, the total deployment cost of the integratednetwork (CT ) is given by:

CT (n, d) = COLT + CAinst + Cs + CB

inst + CN (6)

= α + (δ + θ)DCO +n∑i

β +n∑i

((δ + θ) · Di + γ),

where n is the number of ONU-BSs, and d is the distancebetween ONU-BSs. Therefore, one of the performance metricswith respect to the number of ONU-BSs and the distancebetween ONU-BSs, the total deployment cost of the integratednetwork can be obtained.

B. Transmission Model for WiMAX Systems

A radio propagation model for WiMAX systems, basedupon a large-scale propagation model by path-loss power law,is adopted to analyze DA and TA performances. Thus, theaverage signal power received from the ith ONU-BS to thelocation of an MS (x, y) (PA

i (x, y)) is obtained as [25]:

PAi (x, y) = K · ( di

d0)e, (7)

where di is the distance from the ith ONU-BS (xi, yi) tothe location of an MS (=

√(x − xi)2 + (y − yi)2), d0 is the

reference distance, K is an arbitrary constant (which dependson the transmitted power, the transmitter, receiver antennagain, and the frequency), and e is the path-loss exponent (=2in free space).Under a DA environment, two or more collaborative ONU-

BSs may communicate with a single MS simultaneously, and

thus the SINR of an MS under a DA environment at thelocation (x, y) (γDA(x, y)) is given by:

γDA(x, y) =∑

i PAi (x, y)N0

, (8)

where N0 is the channel noise power.In contrast, the SINR of an MS from the ith associated

ONU-BS under a TA environment at the location (x, y)(γTA(x, y)) is given by:

γTAi (x, y) =

PAi (x, y)∑

j(j �=i) PAj (x, y) + N0

. (9)

Under a DA environment, there is only one chance to senddata from all of the collaborative ONU-BSs to an MS in anoverlapped cell coverage area. Therefore, through Shannon’sresult, the achievable data rate at the location of an MS (x, y)under a DA environment (DRDA(x, y)) is expressed as:

DRDA(x, y) = B · log2(1 + γDA(x, y)), (10)

where B is the bandwidth of the channel in Hz.Differing from a DA environment, all of the ONU-BSs

can transmit wireless signals to their associated MSs in anoverlapped cell coverage area. So, the achievable data rateat the location of an MS (x, y) under a TA environment(DRTA(x, y)) is given by:

DRTA(x, y) =∑

i

B · log2(1 + γTA(x, y)). (11)

Then, for evaluation of the wireless downlink transmissionperformance, the total achievable data rate supported by theintegrated network is given by:

DRT (n, d) =∫ ∞

−∞

∫ ∞

−∞DR(x, y)dxdy, (12)

where n is the number of ONU-BSs, d is the distancebetween ONU-BSs, and DR(x, y) is either DRDA(x, y) orDRTA(x, y) according to each transmission environment. Asa result, one of the performance metrics with respect to thenumber of ONU-BSs and the distance between ONU-BSs, thetotal achievable data rate supported by the integrated network,can be obtained.

C. Cost Efficiency Evaluator

For measuring cost efficiency of the integrated network(i.e., the total achievable data rate supported by the integratednetwork over the total deployment cost of the integratednetwork under predefined n and d), we introduce a costefficiency evaluator (Z(n, d)):

Z(n, d) =DRT (n, d)CT (n, d)

, (13)

where n is the number of ONU-BSs (decided by the requiredregion of the network coverage area), and d is the distancebetween ONU-BSs (decided by applying this evaluator).Now, we can optimally conduct a cost-efficient cell planning

in consideration of cost efficiency of the integrated network.



(e) 19 ONU-BSs:Octagon

d

(a) 4 ONU-BSs:Diamond

d

(b) 6 ONU-BSs:Triangle

d

(d) 9 ONU-BSs:Diamond

d

(c) 7 ONU-BSs:Octagon

d

Cell coverage area

The location of the splitter

Fig. 3. Basic examples of network topologies (d: the distance betweenONU-BSs).

V. PERFORMANCE EVALUATION

A. Assumptions

To analyze the performance of a DA environment regardingthe wireless downlink transmission performance and totaldeployment cost compared to that of a TA environment, it isassumed that the size of the area to be evaluated is divided bydiscrete rectangular units (each unit of area (U )). Accordingto a predefined number of ONU-BSs, the shape of the networktopology is set variously in consideration of several cases ofwireless downlink transmission and cost, as shown in Fig. 3.To illustrate, in cases where the number of ONU-BSs is 3,6, or 10, the network topology is organized by a triangularshape. Also, in cases where the number of ONU-BSs is 4,9, or 16, the network topology is organized in a diamondshape. Lastly, in cases where the number of ONU-BSs is 7or 19, an octagonal shape is adopted as the network topology.Based upon the system parameters in Table V-A, only wirelessdownlink transmission performance is put into considerationbecause there is no difference of wireless uplink transmissionperformances between under DA and TA environments. Ineach network topology, the ONU-BS providing the strongestRSS under a TA environment is selected as the associatedONU-BS without concerning about handovers of MSs.

B. Evaluation Metrics for Wireless Downlink TransmissionPerformance

To express enhancement degree of the wireless downlinktransmission performance under both DA and TA environ-ments, two evaluation metrics are defined as follows:(i) The total achievable data rate supported by an

ONU-BS (DRA); this can be obtained from dividing thetotal achievable data rate supported by the integrated network(DRT (n, d) from Equ. (12)) by the number of ONU-BSs.Namely, this metric is for showing the total achievable datarate supported by an ONU-BS in the integrated network undera predefined condition.

TABLE ISYSTEM PARAMETERS

Symbol Description Value Unit

PTx Transmitter output power 64 wattsPLF S Path loss 34.5+35log10DL dB

FS Shadow fading None -GTx,A Tx antenna gain 10 dBiGRx,A Rx antenna gain 0 dBiNTx Number of Tx antennas 1 -NRx Number of Rx antennas 1 -SRx Receiver sensitivity -90 dBmBW Channel bandwidth 107 HzDS OFDM symbol duration 102.9 µsTS Useful symbol time 91.4 µsTG Guard time (12.5%) 11.4 µs

DL:UL - 28:9 -PN0 Noise power -104 dBmAtot Total evaluated area 100km·100km -U Basic unit of area 100m·100m -

AMIN Minimum position coordinates in Atot 0 -AMAX Maximum position coordinates in Atot 1000 -SINRc Required SINR for coverage 8 dB

DRA(n, d) =DRT (n, d)

n, (14)

where n is the number of ONU-BSs and d is the distancebetween ONU-BSs.(ii) The total available size of a cell coverage area

supported by an ONU-BS (CCA); the total available size ofnetwork coverage area (CCT ) is given by counting the numberof units of an area whose SINR is greater than SINRc (thisthreshold value decides that each U can be recognized as apart of network coverage area) in the integrated network.

CCT =AMAX∑

x=AMIN

AMAX∑y=AMIN

(CCT + 1), for γ(x, y) ≥ SINRc,

(15)where γ(x, y) is γDA(x, y) or γTA(x, y) according to eachtransmission environment, and the initial value of CCT is setto 0.Then, CCA(n, d) is obtained from dividing CCT by n

within it. In other words, this metric is for presentingCCA(n, d) in the integrated network under a predefinedcondition.

CCA(n, d) =CCT (n, d)

n, (16)

where n is the number of ONU-BSs and d is the distancebetween ONU-BSs.

C. Wireless Downlink Transmission Performance and CostEfficiency

The total available size of a cell coverage area supportedby an ONU-BS (CCA) and the total achievable data ratesupported by an ONU-BS (DRA) are shown in Figs. 4and 5, respectively. As expected, in the overall range of dexcept the distances of more than 6 km, the CCA undera DA environment is much greater than that under a TAenvironment. This implies that some areas out at the edge ofa overlapped cell coverage area, whose SINR are lower than



1 2 3 4 5 6 7 8 9 100

200

400

600

800

1000

1200

1400

1600

1800

2000

Distance between ONU−BSs (km)

CC

A (

U)

DA (n=6)DA (n=10)DA (n=15)TA (n=6)TA (n=10)TA (n=15)

(a) Triangle network topology

1 2 3 4 5 6 7 8 9 100

200

400

600

800

1000

1200

1400

1600

1800

2000


CC

A (

U)


(b) Diamond network topology

1 2 3 4 5 6 7 8 9 100

200

400

600

800

1000

1200

1400

1600

1800

2000


CC

A (

U)

DA (n=7)DA (n=19)TA (n=7)TA (n=19)

(c) Octagon network topology

Fig. 4. Comparison of DA and TA environments regarding the total availablesize of a cell coverage area supported by an ONU-BS (CCA) (n: the numberof ONU-BSs).

SINRc, could be transferred to a part of a network coveragearea, as presented in Fig. 2. In addition, the DRA under a DAenvironment is better than theDRA under an TA environment.This is mostly due to the fact that neighbor ONU-BSs cantake a much more aggressive modulation and coding ratio,although doing so will definitely consume resources from allof the collaborative ONU-BSs. As a result of the observationof the positive effects on CCA andDRA, the best performancecan be achieved under situations where the distance between

1 2 3 4 5 6 7 8 9 100

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

11


DR

A (

bp

s/ce

ll)


(a) Triangle network topology

1 2 3 4 5 6 7 8 9 100

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

11


DR

A (

bp

s/ce

ll)


(b) Diamond network topology

1 2 3 4 5 6 7 8 9 100

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2x 10

11


DR

A (

bp

s/ce

ll)

DA (n=7)DA (n=19)TA (n=7)TA (n=19)

(c) Octagon network topology

Fig. 5. Comparison of DA and TA environments regarding the totalachievable data rate supported by an ONU-BS (DRA) (n: the number ofONU-BSs).

ONU-BSs is similar with the diameter of the cell coveragearea of an ONU-BS (e.g., about 5 km).

Fig. 6 presents the total deployment cost of the integratednetwork according to the number of ONU-BSs (n) and thedistance between ONU-BSs (d), which is achieved throughEqu. (6). To begin with, the following cost factors are assumedfor the evaluation: the cost of the OLT including site rentals



1 2 3 4 5 6 7 8 9 100

1

2

3

4

5

6

7x 10

6


To

tal d

eplo

ymen

t co

st (

$)

n=19n=16n=15n=10n=9n=7n=6n=4

Fig. 6. Total deployment cost of the integrated network ($) (n: the numberof ONU-BSs).

(α) is $28,000, the cost of each splitter port (β) is $100, thecost of an ONU-BS including site rentals (γ) is $40,000, thecost of trenching and laying fiber is (δ) $16,000/km, andthe cost of fiber including duct and tube (θ) is $4,000/km[26], [27]. When d increases, the cost also increases due toincreasing the cost of trenching and laying fibers. Hence, itis revealed that the total deployment cost is mainly affectedby the total length of deployed optical fibers in the integratednetwork.Fig. 7 presents cost efficiency evaluator (Z), which eval-

uates the cost efficiency of the integrated network in con-sideration of the total achievable data rate supported by theintegrated network over the total deployment cost of theintegrated network, under DA and TA environments. Since thetotal achievable data rate supported by the integrated networkdepending on d just increases until reaching a certain degree(e.g., when d is about 5km, as presented in Fig. 5) and the totaldeployment cost linearly increases due to the increase of thetotal length of deployed optical fibers, the maximum point ofZ is mostly d = between 2 to 4 km under a DA environment.On the other hand, under a TA environment, the maximumpoint of Z is d = 5, because interferences from other wirelessdownlink signals decrease significantly. However, from thispoint, the Z decreases due to the fact that there is increasedtotal deployment cost of the integrated network and no changein the total achievable data rate supported by the integratednetwork.

D. Case Study under the Required Size of Network CoverageArea

Based on Z , a case study (Fig. 8) that finds the optimalsolution under the required size of network coverage area isconducted to examine the DA-based integration architectureabout the average spectral efficiency of MSs for wirelessdownlink transmission and the total deployment cost of theintegrated network. In particular, the cost efficiency improve-ment by initiating cooperative transmission among DAs ofONU-BSs is investigated. Fundamentally, this solution is

1 2 3 4 5 6 7 8 9 100

0.5

1

1.5

2

2.5x 10

6


Z (

bp

s/$)

n=4n=6n=7n=9n=10n=15n=16n=19

Maximum point

(a) A DA environment

1 2 3 4 5 6 7 8 9 100

0.5

1

1.5

2

2.5x 10

6


Z (

bp

s/$)

n=4n=6n=7n=9n=10n=15n=16n=19

Maximum point

(b) A TA environment

Fig. 7. Comparison of DA and TA environments regarding cost efficiency ofthe integrated network (Z) (maximum point: the highest cost efficiency undereach n).

found with the following consideration; since the diameter ofa single ONU-BS is about 4.5km under the system parametersshown in Table I, the distance between ONU-BSs is limited to4km so that every cell coverage area is continuously connectedwith cell coverage areas of other ONU-BSs. In addition, thisis conducted with varying the required region of a networkcoverage area from 5000 to 20000U . In the overall range ofthe required region, the total deployment cost under a DAenvironment are positively enhanced compared to that undera TA environment while keeping reasonable spectral efficiencydue to the effect of the increased SINR based upon the ONU-BS cooperation. To exemplify, when the required region ofnetwork coverage area is set to 5000U , the distance betweenONU-BSs under a DA environment is 4 and smaller than thatof a TA environment. Specifically, when the required regionof network coverage area is larger, the number of ONU-BSsor the distance between ONU-BSs under a DA environment ismore decreased than that under a TA environment. This meansthat increasing the overlapped cell coverage areas causesincreasing of the positive effects of a DA environment.

VI. CONCLUSION

In this paper, we proposed a Distributed Antenna (DA)-based EPON-WiMAX integration architecture with a cost-efficient cell planning for enhancement of cost efficiency of the



5000 10000 15000 200000

1

2

3

4

5

6

7

8

A required region of network coverage area (U)

Sp

ectr

al e

ffic

ien

cy (

bp

s/H

z)

DATA

(a) The average spectral efficiency of MSs for wireless downlink transmis-sion

5000 10000 15000 200000

0.5

1

1.5

2

2.5

3

3.5

4x 10

6

A required region of network coverage area (U)

To

tal d

eplo

ymen

t co

st (

$)

DATA

(4,4)

(4,3)

(9,4)

(10,4)

(15,4)

(10,4)

(15,4)

(19,4)

(n,d)=(the number of ONU−BSs, the distance between ONU−BSs)

(b) Total deployment cost of the integrated network

Fig. 8. Comparison of DA and TA environments regarding spectral efficiencyand the total deployment cost perspectives under a required region of networkcoverage area.

integrated network. Fundamentally, the proposed DA-basedintegration architecture can enlarge the available networkcoverage area and provision the achievable data rate thana TA environment, and thus it can support more MobileStations (MSs). Then, a DA environment could generate morerevenue without increasing operating expenditure (OPEX)due to similar conditions including power consumption, formfactor, etc. Besides, a cost-efficient cell planning with a casestudy presented the enhanced total deployment cost gainof the integrated network with a similar level of spectralefficiency of Mobile Stations (MSs) for wireless downlinktransmission under a required region of network coveragearea, as a part of applying the proposed DA-based integrationarchitecture. Therefore, the DA-based integrated network isexpected to be one of the promising solutions for modernbroadband access networks supplying quad-play services. Inaddition, the study for the DA-based integration architecturecan make the synergy effect between optical and wirelesssides to each other in consideration of two key issues: thewireless downlink transmission performance for multicast andbroadcast services (MBSs) in wireless networks and totaldeployment cost efficiency in optical networks.

ACKNOWLEDGMENT

This work was supported in part by the Development Man-made Disaster Prevention Technology grant funded by the Ko-rea Government (NEMA; National Emergency ManagementAgency) (No. Nema−09−MD−06), the MKE (Ministry ofKnowledge Economy), Korea, under the ITRC (InformationTechnology Research Center) support program supervisedby the NIPA (National IT Promotion Agency)(NIPA-2010-(C1090-1011-0013)), and the IT R&D program of MKEKEIT [2009-F-057-01, Large-scale wireless-PON convergencetechnology utilizing network coding].

REFERENCES

[1] IEEE 802.3ah Task Force: http://www.ieee802.org/3/efm/. [Online on Jan.23, 2008]

[2] IEEE Std 802.16e-D12, ”Air Interface for Fixed Broadband WirelessAccess Systems,” Feb. 2005.

[3] G. Shen, R. S. Tucker, and C. Chae, ”Fixed Mobile ConvergenceArchitecture for Broadband Access: Integration of EPON and WiMAX,”IEEE Commun. Mag., Aug. 2007.

[4] J. Li and G. Shen, ”Cost Minimization Planning for Greenfield PassiveOptical Networks,” in Proc. OFC’08, 2008.

[5] I. D. Garcia, K. Sakaguchi, and K. Araki, ”Cell Planning for CooperativeTransmission,” in Proc. IEEE WCNC’08, Mar, 2008.

[6] D. Amzallag, M. Livschitz, J. S. Naor, and D. Raz, ”Cell Planning of 4GCellular Networks: Algorithmic Techniques and Results,” in Proc. IEEEIntl. Conf. on 3G Mobile Commun. Technol., Nov. 2005.

[7] Z. Jia, J. Yu, A. Chowdhury, G. Ellinas, and G. K. Chang, ”SimultaneousGeneration of Independent Wired and Wireless Services Using a SingleModulator in Millimeter-Wave-Band Radio-Over-Fiber Systems,” IEEEPhotonics Technol. Lett., vol. 19, no. 20, pp. 1691-1693, Oct. 2007.

[8] P. Lin, C. Qiao, T. Wang, and J. Hu, ”Optimal Utility-Based Band-width Allocation over Integrated Optical and WiMAX Networks,” Proc.OFC’06, OThM2, 2006.

[9] K. Yang, S. Ou, K. Guild, and H.-H Chen, ”Convergence of EthernetPON and IEEE 802.16 Broadband Access Networks and its QoS-AwareDynamic Bandwidth Allocation Scheme,” IEEE J. Sel. Areas Commun.,vol. 27, no. 2, pp. 101-116, Feb. 2009.

[10] Y. Luo, S. Yin, T. Wang, Y. Suemura, S. Nakamura, N. Ansari,and M. Cvijetic, ”QoS-aware Scheduling over Hybrid Optical WirelessNetworks,” in Proc. OFC’07, NthB1, 2007.

[11] S. Sarkar, S. Dixit, and B. Mukherjee, ”Hybrid Wireless-Optical AccessNetwork (WOBAN): A Review of Relevant Challenges,” J. Lightw.Technol., vol. 25, no. 11, pp. 3329-3340, Nov. 2007.

[12] W.-T. Shaw, S.-W. Wong, N. Cheng, K. Balasubramanian, X. Zhu, M.Maier, and L. G. Kazovsky, ”Hybrid Architecture and Integrated Routingin a Scalable Optical Wireless Access Network,” J. Lightw. Technol., vol.25, no. 11, pp. 3443-3451, Nov. 2007.

[13] J. She and P.-H. Ho, ”Cooperative Coded Video Multicast for IPTVServices under EPON-WiMAX Integration,” IEEE Commun. Mag., vol.46, no. 8, pp. 104-110, Aug. 2008.

[14] Y. Luo, T. Wang, S. Weinstein, and M. Cvijetic, ”Integrating Optical andWireless Services in the Access Networks,” in Proc. OFC’06, NthG1,2006.

[15] Y. Xiang, J. Luo, and C. Gorg, ”Performance Impact of User Groupingon Fractional Power Loading in the OFDMA Downlink,” in Proc. IEEEPIMRC’07, 2007.

[16] X. Kai, T. Xiaofeng, W. Ying, and Z. Ping, ”Inter-cell Packet Schedulingin OFDMA Wireless Network,” in Proc. IEEE VTC’07, pp. 3115-3119,Spring, 2007.

[17] H. Zhang and H. Dai, ”Cochannel Interference Mitigation and Coop-erative Processing in Downlink Multicell Multiuser MINO Networks,”EURASIP J. Wireless Commun. Netw., vol. 2, pp. 222-235, 2004.

[18] O. Somekh, O. Simeone, Y. Bar-Ness, and A. Haimovich, ”DistributedMulti-Cell Zero-Forcing Beamforming in Cellular Downlink Channels,”in Proc. IEEE GLOBECOM’06, 2006.

[19] H. Viswanathan, S. Venkatesan, and H. Huang, ”Downlink CapacityEvaluation of Cellular Networks with Known-Interference Cancellation,”IEEE J. Sel. Areas Commun., vol. 21, no. 5, pp. 802-811, Jun. 2003.

[20] L. Shao and S. Roy, ”Downlink Multicell MIMO-OFDM: An Architec-ture for Next Generation Wireless Networks,” in Proc. IEEE WCNC’05,2005.



[21] Z. Shen, J.G. Andrews, and B.L. Evans, ”Optimal Power Allocation inMultiuser OFDM Systems,” in Proc. IEEE GLOBECOM’03, vol. 1, pp.337-341, Dec. 2003.

[22] M. Hajduczenia, B. Lakic, H. J. A. Da Silva, and P. P. Monteiro,”Optimized Passive Optical Network Deployment,” J. Optical Netw., vol.6, no. 9, pp. 1079-1104, 2007.

[23] V. Tarokh, N. Seshadri, and A. R. Calderbank, ”Space-Time Codes forHigh Data Rate Wireless Communication: Performance Criterion andCode Construction,” IEEE Trans. Inf. Theory, vol. 44, no. 2, pp. 744-764, Mar. 1998.

[24] J. Garcia-Fragoso and G.M. Galvan-Tejada, ”Cell Planning based onthe WiMax Standard for Home Access: A Practical Case,” in Proc.ICEEE’05, Sep. 2005.

[25] R. Prasad, Universal Wireless Personal Communications, Artech House,1998.

[26] R. Esmailzadeh, Broadband Wireless Communications Business: AnIntroduction to the Costs and Benefits of New Technologies, John Wileyand Sons Ltd., 2006.

[27] C. Lin, Broadband Optical Access Networks and Fiber-to-the-Home:Systems Technologies and Deployment Strategies, John Wiley and SonsLtd., 2006.

Min-Gon Kim is a research engineer with Publicand Original Technology Research center, DaeguGyeongbuk Institute of Science and Technology(DGIST). Before joining DGIST, he received hisBS degree from Ajou University, Korea, in 2004,and his Ph.D degree (Integrated MS and Ph.D pro-gram) from Korea Advanced Institute of Science andTechnology (KAIST), Korea, in 2009, respectively.As a part of his research experience, he worked forresearch development laboratory at Miritek in 2005,where he researched implementation of testbed for

optical networks. Besides, he visited the Center for the Ultra-BroadbandInformation Network (CUBIN), University of Melbourne, Australia, foroverseas exchange program in 2008 for integration of optical and wirelessaccess networks. He was nominated Marquis Who’s Who in the World 2009and 2010. His research interests are management and applications of wirelessnetworks including practical applications of WPAN, mobility management ofWLAN, and management enhancement and extension WMAN.

Gangxiang Shen (S’98-M’99) is a Lead Engineerwith Ciena US. Before he joined Ciena, he wasan Australian Postdoctoral Fellow (APD) and ARCResearch Fellow with ARC Special Research Centerfor Ultra-Broadband Information Networks, Depart-ment of Electrical Engineering, University of Mel-bourne, Australia. He received his Ph.D. from theDepartment of Electrical and Computer Engineering,University of Alberta, Canada, in January 2006.He received his M.Sc. from Nanyang Technolog-ical University in Singapore and his B.Eng. from

Zhejiang University in P. R. China. His research interests are in networksurvivability, optical networks, and wireless networks. He has authored andco-authored more than 40 peer-reviewed technical papers. He is also arecipient of the Izaak Walton Killam Memorial Award of University ofAlberta, the Canadian NSERC Industrial R&D Fellowship, etc. His personalURL is http://www.gangxiang-shen.com/.”

JungYul Choi received his B.S degree from InhaUniversity, South Korea, in 2000, and his M.S andPh.D degrees from Information and Communica-tions University (currently merged to Korea Ad-vanced Institute of Science and Technology), SouthKorea, in 2002 and 2006, respectively. He has beenworking for Network R&D Laboratory, KT Cor-poration (formally, Korea Telecom) since 2006. Hehas authored around 20 reviewed technical journalpapers, and holds around ten patents in areas oftelecommunications networks. He was nominated

Marquis Who’s Who in the World 2009, and International Engineer of theYear for 2010 from IBC. He has been a reviewer of technical journal paperssuch as IEEE Infocom, IEEE Communications Letters, IEICE Transaction onCommunications, IEICE Transaction on Information and Systems, ElservierJournal of Visual Communication and Image Representation, and ETRIJournal. His research interests are in Next-Generation Networks, FutureNetworks, wired/wireless convergence, and network economics.

Bokrae Jung received the B.S. degree in Informa-tion and Communications Engineering from Yeung-nam University, Gyeongsan, Korea, in 2003, andthe M.S. degree in School of Engineering in Ko-rea Advanced Institute of Science and Technology(KAIST), Daejeon, Korea, in 2006, where He is cur-rently working toward Ph.D. degree in Departmentof Information and Communications Engineering.He worked in WDM-PON Technology Team atETRI, Korea from April to December, 2005. Hespent six months in the Center for Ultra-Broadband

Information Networks (CUBIN), the University of Melbourne, Australia, foran overseas exchange program from 2007 to 2008, where he studied opticaland wireless converged networks.

Hong-Shik Park received the B.S. degree fromSeoul National University, Seoul, S. Korea in1977, and the M.S. and Ph.D. degrees from Ko-rea Advanced Institute of Science and Technology(KAIST), Daejeon, S. Korea all in Electrical En-gineering in 1986 and 1995, respectively. In 1977he joined Electronics and Telecommunications Re-search Institute (ETRI) and had been engaged indevelopment of TDX digital switching system fam-ily including TDX-1, TDX-1A, TDX-1B, TDX-10,and ATM switching systems. In 1998 he moved to

Information and Communications Univ., Daejeon, Korea as a faculty. Cur-rently he is a professor of the Dept. of Electrical and Electronics Engineering,KAIST, Daejeon, S. Korea. From 2004 he is a director of BcN EngineeringResearch Center sponsored by KEIT, Korea. His research interests are networkarchitecture and protocols, traffic engineering, and performance analysis oftelecommunication systems. He is a member of the IEEE, IEEK and KICS,S. Korea.

Minho Kang received the BSEE, MSEE, and Ph.D.degrees from Seoul National University, Universityof Missouri-Rolla, and University of Texas at Austinin 1969, 1973, and 1977, respectively. From 1977to 1978, he was with AT&T Bell Laboratories,Holmdel, NJ. Moreover, from 1978 and 1989, hewas a Department Head and a Vice President atElectronics and Telecommunications Research In-stitute (ETRI). Also, he served as the Electricaland Electronics Research Coordinator at the KoreanMinistry of Science and Technology from 1985 and

1988. After that, he was an Executive Vice President at Korea Telecom (KT) incharge of R&D, quality assurance, and overseas business development groupsfrom 1990 to 1998. In 1999, he joined the Information and CommunicationsUniversity (ICU) as a professor and served as Dean of Academic and StudentAffairs. He was the Director of the OIRC from 2000 to 2009. Now, hejoins a professor and served as a vice-president of the college of informationscience & technology, Korea Advanced Institute of Science and Technology(KAIST), Daejeon, Republic of Korea. He was awarded the Order of Merit-DongBaekJang by Korean Government and Grand Technology Medal by21st Century Management Club in 1983 and 1991, respectively for thecontribution of optical communications technology development. In 2007 and2008, He was awarded the COIN Award 2007 and ICU Best Research Award,respectively. He served as the Study Group Chairman at the Asia Pacific Tele-community of Bangkok during 1996-1999, is a member of National Academyof Engineering in Korea, and is a senior member of IEEE. He is an author ofBroadband Telecommunications Technology, published in 1993 and revisedin 1996 by Artech House. He is also an associate editor of IEEE OpticalCommunication and Networks Magazine.


distributed antenna-based epon-wimax integration and its cost-efficient cell planning

Documents