rfid based localization
TRANSCRIPT
RFID based localizationA presentation for the course: ELT-46206 Signal Processing for Mobile Positioning
Mehjabin Sultana, Gourab Datta, Jarno Salonen
Definition of the topicTechnology and standardsMethods / Algorithms used
● Multilateration● Bayesian inference● Nearest neighbor● Proximity
Targets of application / demos● 3 demo cases
Discussion on challengesConclusion
Outline
Radio-Frequency IDentification (RFID) is a generic term for technologies that use radio waves for automatic identification of objects and/or people.RFID localization basics: RFID reader and tag(s)
Definition of the topic
Uses algorithms very similar to WLAN localization, main differences:● signal strength as indicator of distance (WLAN: time-of-arrival)● easiness of deployment as RFID tags are portable
Tags can be either active, passive or semi-passive● Active tags with on-board batteries provide a range of <100m● Passive tags are less expensive, but provide a range of only <2m
(LF/HF) or <6m (UHF)● Semi-passive tags use internal battery to enhance the broadcasting
signal strength, range comparable to active tags● Active tags don’t need reader initialization, (semi-)passive ones do
Typical targets of application: healthcare, transport and logistics, Automated Guided Vehicles (AGV)
Definition of the topic (cont.)
Frequencies used by RFID and their target application areas:● LF (inductive coupling): 125 kHz-134 kHz
o ISO 11784/5 and ISO 14223: animal tracking
● HF (inductive coupling): 13,56 MHzo ISO 15693: retail and item-level logistics, pharmaceutical item tracking, patient
tracking in hospitals
● UHF (electromagnetic coupling): 965-928 MHzo ISO 18000-6a and b: tracking of goods with high density of tagso ISO 18000-7: indoor and outdoor Real Time Location Systems
● Microwaves (electromagnetic coupling): 2,4 GHz (unlicensed ISM)o ISO 24730-21 (incl. IEEE 802.11): WiFi-based localization among others in
healthcare (patient tracking), other uses where WiFi interoperability is needed
Technology and standards
•The biggest challenge for RFID localization is how to mathematically model the variation of the RF signals in space. •Friis transmission equation -
•Other methods such as statistical analysis are developed to calibrate the relationship between signal strength and distance as this formula doesn’t do the job done properly.
Methods/algorithms used
•Multilateration estimates the coordinates of the target node from the distances between the target node and the reference nodes with known coordinates.If there are n reference nodes Rk, k = 1,2,...,n with known coordinates (xk, yk), and the distances between the target node, T, with unknown coordinates (x, y) and reference nodes are estimated to be rk, k = 1,2,...,n we can obtain
•r12 = (x − x1)2 +(y − y1)2
•r22 = (x − x2)2 +(y − y2)2,
• rn2 = (x − xn)2 +(y − yn)2•
Multilateration
Typical methods to solve multilateration equations: Normal equations QR-factorization Singular-value Decomposition •Multilateration is a mature algorithm which is easy to code and requires less computation effort.•That’s why, it is being widely used in many localization studies
Multilateration (Cont...)
•Bayesian inference is a method of inference in which Bayes' rule is used to update the probability estimate for a hypothesis as additional evidence is acquired.•The basic principle for localizing a stationary target by Bayesian inference can be represented by Fig.•According to Bayesian Rule, the position of the targetnode can be obtained by the followingrecursive equation,P((x, y)|s1, s2, . . . , sn) = αP(sn|(x, y))×P((x, y)|s1, s2, . . . , sn−1)
Bayesian inference
where α is a normalizing factor to ensure the sum of posterior probability P((x, y)|s1, s2, . . . , sn) to be one, and P(sn|(x, y)) calculates the probability of signal strength given the location of the target node.
•One of the unique advantages of Bayesian inference is that it can update the target’s location from the dynamical data of signal strength. This property makes this method effective in localizing mobile targets, such as the AGVs in manufacturing facilities.
Bayesian inference (Cont...)
•The closer two points, the smaller the difference between the signal strengths of the two points->an object can be localized by its neighbors.•The coordinates of the target point, (x, y), can be obtained by the equation:
•where xi , yi, wi = coordinates, i = 1, 2, . . . , k= weights; and k is the number of nearest neighbors.
••
Nearest neighbor
•This method avoids the signal propagation estimation, and it is suitable for the complex non-isotropic environments and it is not sensitive to the variation of application environments
•Successfully used for localization in some applications such as health care systems
Nearest neighbor (Cont..)
Proximity method uses the approximate communication area to detect whether the target node is in a region or not.ürequires less computation than Bayesian inferenceü positioning accuracy is usually poorer
1.An RFID tag is to be localized by a mobile reader2.The localization area is partitioned into n ×n square cells ,each cell located at the coordinates (i, j ) 0 ≤ i, j ≤ n−11.The communication area of the reader can be approximated by a square composed of 2k ×2k cells.2. If the reader detects the tag at m known positions, (xr , yr ), r = 1, . . . ,m, at each position the communication area is a union of cells (xr − k, xr + k − 1) × (yr − k, yr + k − 1).3.The location of the tag can then be estimated by computing the centroid of this intersection of communication areas.
Proximity
Can be used in :•Construction site•Localization in Wireless sensor network
Proximity (Cont..)
Three demo cases in which RFID localization is used in different ways
1. Indoor localization of a person using active RFID tags2. Indoor localization system using passive RFID technology3. Practical use case of using active RFID localization in a restaurant
Targets of application
Video Demonstration:
Case 1: Indoor localization of a person with active RFID tags (1)
Description of the Video:● The Left section of the video shows Probability Density
Function (PDF) of the position of the person.
● The Right section of the video shows True trajectory (green) and estimated trajectory (black) of the person.
Case 1: Indoor localization of a person with active RFID tags (2)
How the system works:● An RF reader is carried by the mobile user.
● A number of active RFID tags are distributed at known positions which regularly emit RF signals with an identification code.
● Upon reception of a signal, the range of the user to the corresponding tag is estimated indirectly from the received signal strength (RSSI), using a previously obtained statistical model. A computationally efficient Bayesian localization method (particle filter) is used to process the measurements and produce an estimation of the user’s position.
Case 1: Indoor localization of a person with active RFID tags (3)
Analysis of the system:● The best approximation was obtained when the route passed entirely
through the common hall. Although measurements from tags in the other rooms are considered nonetheless.
● When the route goes through several rooms, the positioning error often increases. This is caused by the change between rooms and also by a relatively lower tag density in the corridor.
● However, even with lower precision, the trajectory of the user between rooms was traced without problems.
Ref: http://lopsi.weebly.com/publications.html
Case 1: Indoor localization of a person with active RFID tags (4)
Video Demonstration:
Case 2: Indoor location system based on passive RFID technology (1)
Description of the system:● The video shows tracking of autonomous robots within a closed
environment based on passive RFID technology.
How the system works:● Robot Lego Mindstorm NXT carrying a PDA that supports Wi-Fi
connection.
Case 2: Indoor location system based on passive RFID technology (2)
How the system works (contd.):● Sensing surfaces, divided into a grid of small squared surfaces or locations
units with passive RFID tags, where each tag represents a location unit with unique identifier.
● PDA hp iPAQ 2940x with a passive RFID reader (HF-13.56 MHz) which reads tags from the sensing surface and retrieves the location unit identification code.
● Localization manager receives the IDs of the entity location from the PDA, maps the physical ID to a virtual map position and shows the entity locations to
the final user in the map presentation screen.
Case 2: Indoor location system based on passive RFID technology (3)
Practical Feasibility Analysis:● One of the lacks of the system is the capability of being wearable, because
readers must be near the sensing surface. However, the release of RFID reader chips by INTEL (R1000) and Samsung may set a tendency of RFID to be present in everyday applications.
● The cost of maintenance and infrastructure is low because there is no necessity to use any special device apart from the PDA and RFID reader. On the other hand, setup and installation costs are almost irrelevant considering low cost of passive RFID tags.
● As passive RFID does not require external power supply, the cost of power is negligible.
Ref: Youtube Link ; Article
Case 2: Indoor location system based on passive RFID technology (4)
Scenario:● A tag given to the customer is used to locate the customer later in the
restaurant in order to deliver the portion (steak, burger, etc.)Problem:● In a restaurant for 350+ customers, the customer is not easy to find
Solution:● Add an active rfid transponder to the tag, equip the restaurant with
readers and a location system for finding the customer● Method: proximity (signal strength/reader)
Result:● Reduce the time to find the customer to less than half
Case 3: RFID localization in practice (1)
Case 3: RFID localization in practice (2)
Case 3: RFID localization in practice (3)
Major challenge with active tags is battery lifetime.● Lifetime of up to 5 years can be reached, depending on use
Challenge with passive tags is the small reading distance● Price of individual tags is low
Other issues● Interference (multiple tag responses, multiple reader queries, reader signal power, non-
conductive materials such as metal or glass)● Multipath propagation● Require more reliability, availability and precision
o Therefore hybrid technology system will be the optimum solution.● Since no single positioning technology provides all requirements, some researchers are
already working on the combination system like Wireless LAN and RFID● Privacy (when the number of RFID tags around us increases)
Discussion on challenges
RFID localization basics: RFID reader + tags + application = localization● Active tags are self-powered and have a significantly good read range● Passive tags are low-cost and have a shorter (<6m) read range● Semi-passive tags have a power source for enhancing signal strength
Methods and algorithms● Multilateration, Bayesian inference● Nearest neighbor, Proximity
Usage mostly indoor ● Localization of patients, assets, etc.
Challenges● Tag issues such as battery lifetime, short read range● Requires WiFi/LAN to be more efficient
Conclusions