This document proposes methods for localizing assets in indoor environments using RFID technology. Fixed assets can be localized by reading their tag's stored location data from a distance. Fixed tags also enable user localization through proximity techniques. Moveable assets are localized by matching RSSI patterns between their tags and reference tags on fixed assets, without requiring distance estimates. The approach incorporates environmental dynamics and does not need centralized infrastructure or calibration. This allows users to locate themselves and targets using a handheld RFID reader.

1. Localization of RFID-Equipped Assets During the Operation Phase of
Facilities
Ali Motamedi & Mohammad Mostafa Soltani
Graduate Research Assistants, Concordia University, Canada
Amin Hammad
Professor, Concordia University, Canada
Abstract
Indoor location information has the potential to improve the utilization and maintenance of facilities.
RFID technology has been employed for localization in indoor environments in various research
projects. However, several RFID-base localization systems are inaccurate in indoor environments. In
our previous research, long-range RFID tags are attached to building assets at an early stage of their
lifecycle and the memory on tags is used during the lifecycle. This paper focuses on the localization
opportunities that our proposed RFID-tagged environment can provide. We propose to save current
location data (e.g., coordinates) on tags attached to fixed assets and locate them by reading this data
from a long distance. Additionally, these tags act as reference points for RFID reader localization
techniques to estimate the position of the user. The paper also evaluates an approach to locate
moveable assets (e.g., tools) using the data saved on fixed assets’ tags and an RSS pattern matching
algorithm. This localization method incorporates the dynamics of the environments, is device-
independent and does not require calibration. As a result, a user equipped with an RFID reader is able
to estimate his/her position, as well as obtaining the location information of target assets, without
having access to any central RTLS infrastructure.
Keywords: Radio Frequency Identification, Facilities Management, Localization, Building Information
Modelling
1 Introduction
The localization problem has received considerable attention in the areas of pervasive computing as
many applications need to know where objects are located. Location information can be used by
occupants unfamiliar with a building to navigate and find their destinations; additionally, facilities
management (FM) personnel could be provided with locations of assets. Hence, indoor location
information is especially valuable as it has the potential to improve the utilization and maintenance of
facilities. Furthermore, location information is central to personalized applications in different areas
and it is the basis for the delivery of personalized and location-based services (LBS). It is the basis for
context awareness within the building that involves an automatic recognition of the user’s location
and activity (Papapostolou and Chaouchi, 2011, Li and Becerik-Gerber, 2011).
A Radio Frequency Identification (RFID) tag is a memory storage device for storing a certain
amount of data that can be read wirelessly. RFID technology does not require line-of-sight and the
stored data are dynamic and modifiable. The RFID reader can be a part of other mobile devices such
as cell phones or Personal Digital Assistants (PDAs) (Aimglobal, 2008).

2. RFID technology has been employed for localization in indoor environments in various research
projects (Li and Becerik-Gerber, 2011). RFID tags or readers can be the target for positioning.
However, the main shortcoming of RFID is the interference among its components and some
materials, which makes it sensitive to changes in the environment resulting in inconsistent
performance.
In our previous research, a framework has been proposed in which long-range RFID tags are
attached to building assets at an early stage of their lifecycle and the memory on tags is used to store
various types of data during the lifecycle of buildings’ assets. The stored information on tags is
beneficial for several lifecycle processes and is used by various stakeholders.
This paper focuses on the localization opportunities that our framework can provide. We propose
to save current location data (e.g., coordinates) on tags attached to fixed assets. Consequently, an FM
personnel is able to read a tag from a distance and locates the fixed assets on a floor plan. Fixed tags
with known positions act as reference points for RFID reader localization techniques (e.g. trilateration
and proximity). In this scenario, the user estimates his/her position by scanning the surrounding tags.
Additionally, information such as floor plans, navigational aid and RF fingerprinting database can be
stored in these tags. Furthermore, the paper evaluates an approach to locate moveable assets (e.g.,
tools). In this approach, radio signals sensed from fixed tags (attached to fixed assets) help the user to
estimate the location of the target tag attached to a movable asset based on received signal strength
indicator (RSSI) pattern matching. This method introduces several improvements to available
techniques as it incorporates the dynamics of the environments since the target tag and surrounding
reference tags are affected by the same environmental effects. Additionally, this approach is device-
independent and does not require calibration since it uses RF signal data without converting them into
distance information. As a result, a user equipped with an RFID reader is able to estimate his/her
position, as well as obtaining the location information of target assets, without having access to any
central Real-Time Location System (RTLS) infrastructure. The objectives of the paper are: (1) to
elaborate on a new method to localize fixed assets, (2) to investigate reader localization opportunities
in buildings with RFID-tagged assets, (3) to investigate techniques to localize movable assets using
fixed tags and, (4) to investigate the applicability of the proposed methods using several case studies.
2 Literature review
2.1 RFID localization techniques
Fuchs et al. (2011) categorized RF-based localization methods into four major groups: (1) Lateration
that uses the distances of the target to at least three points with known positions. The estimation of
distance based on RF properties are done using the following techniques: Time of Arrival (ToA),
Time Difference of Arrival (TDoA), interferometry, signal attenuation and, hop-based. (2) Angulation
determines the location of an object from the measured angles to at least two fixed points with known
locations. (3) Fingerprinting uses mapped properties of the environment for position estimation. In
this method a sensor can be located by measuring its current signal strength pattern and comparing it
to a previously surveyed signal map. (4) Connectivity/Proximity uses the analysis of connectivity, i.e.
the number of attainable neighbours. It operates by measuring nearness to a known set of points.
Various indoor RFID localization techniques are designed based on the above-mentioned methods.
Papapostolou and Chaouchi (2011) and Li and Becerik-Gerber (2011) provided thorough surveys and
comparisons between various projects for tag and reader localization.
2.2 RFID-assisted lifecycle management
The use of attached RFID tags for lifecycle management has been proposed in the aerospace industry
for storing unique ID and important lifecycle information on tags attached to aircraft parts for

3. enhancing inspection and repair processes (Harrison, 2007). The framework developed in our
previous research proposed adding structured information taken from the Building Information Model
(BIM) database to RFID tags attached to the building assets (Motamedi and Hammad, 2009). Having
data related to the assets readily available on the tags provides easy data access for stakeholders
regardless of having real-time connection to a central database. In this framework, every asset is a
potential target for tagging. Having tags attached to assets results in a massive tag cloud in the
building. The target assets are tagged during or just after manufacturing and are scanned at several
points in time during the lifecycle. The scan events are for reading the stored data or modifying the
data based on system requirements and the stage at which the scan is happening. The scanned data are
transferred to different software applications and processed to manage the activities related to the
assets (e.g., inspection). Considering the limited memory of the tags, the subset of BIM data has to be
chosen and stored on tags based on the requirements of the tasks. This data is used by different
software applications based on designated access levels (Motamedi et al., 2011).
3 Proposed approach
RTLS technologies are capable of providing real-time location information of assets. However,
providing RTLS infrastructure inside the building is costly and imposes tremendous amount of
technical design and implementation issues such as the scalability of RTLS.
In our proposed approach, it is assumed that a subset of assets is equipped with long-range RFID
tags. The memories of these tags contain information taken from a BIM based on our previous
framework (Motamedi and Hammad, 2009). This research aims to utilize the available mass of RFID
tags in the environment for localization purposes. In our approach, the user who is searching for assets
is equipped with a handheld RFID reader and is able to read the content of the tags from a distance. In
order to identify an appropriate location tracking method, the categorization of assets introduced in
Motamedi and Hammad (2011) is used (i.e., fixed, semi-fixed, movable and temporary).
3.1 Fixed assets localization
Real-time location tracking for fixed assets, that constitute a large portion of available assets, is
unnecessary. In our proposed approach, the memory of tags attached to these assets contains the exact
location data taken from a BIM. Consequently, by accessing the memory of these tags from a
distance, the location of the associated asset will be identified. Having the location data together with
preloaded-floor plan, the personnel are able to find the asset even if it is obstructed or hidden without
having access to any RTLS infrastructure. In this method, the location data on a tag is manually
updated and is not real-time. Hence, this method is not suitable for movable assets.
Attaching long-range tags with large memory to all fixed assets may not become financially
feasible in the near future. In order to benefit from the proposed method, specific long-range tags can
be attached to selected assets to store location information related to a set of assets in the
neighbouring area (location tags). Consequently, when a user tries to locate an asset, the data on the
nearest location tag is read from a distance, which contains location data for all assets in that area
including the target asset. These location tags can also be equipped with large memory chipsets that
contain floor plans and navigational aid information. The location information is updated on location
tags when a fixed asset is installed in that specific area, or when a semi-fixed asset is moved to/from
that specific area.
Figure 1 shows the process flowchart for asset localization: (1) The user scans the area to look for
target RFID tag. (2) The handheld’s reader detects surrounding tags and reads their IDs and data. (3)
The software application queries for the ID/property of target asset amongst detected tags. The queries
properties could be the unique ID of a specific tag or a property of an asset (e.g., condition=require
maintenance, type=boiler, status=faulty). (4, 5) If the target tag is found, the application reads the

4. location data from the memory, locate the appropriate floor plan and shows the asset on the floor plan.
(6) In case the target tag is not found in the scanned area, the application starts an exhaustive search
among all detected location tags to find the data related to the target tag. (7,8) In case a location tag is
detected, the reader reads the data and queries for the target tag. (9) In case the target tag information
is found on location tags, the application reads the location data and shows the target tag on the floor
plan. (10) If the target tag’s data is not found on the location tag, the application prompts that the
target assets could not be found and advises the user to move and change his location and rescan. In
case the location tags are placed in the building based on planned criteria, the application can provide
the user with tips about how to perform the move action.
It is proposed to place the location tags based on a predefined guideline known to users to facilitate
finding these tags. The following are recommendations for their placement: (1) Per Room: tags can be
placed at the exterior side of the entrance to each room in a common area (hallway). These tags can
contain information related to the assets that are located inside each associated area. The tags will be
placed at the exterior to provide maximum data accessibility and read range for users who are
navigating in the common areas, (2) Per floor: location tags can be placed at the entrances to the floor
(e.g. near elevator, in the lobby). These tags can contain information about the important assets in the
associated floor. Moreover, the tags can contain floor plans and occupants’ data for each floor.
Consequently, the user will be able to retrieve the data related to each floor as soon as he/she enters
that floor.
Read
Read tags Is target asset Y Show on floor
Start Scan Area location End
description detected? plan
1 2 4 data on tag 5
3
N
Y
Read
Change Location N Any location Y Query for target Is target asset
location
(Move) tag detected? asset detected?
10 7 data on tag 8
6 9
N
Figure 1. Process flowchart for fixed asset localization
3.2 User localization
In addition to locating assets, location data on fixed tags can help users finding their estimated
locations in the building. Additionally, in scenarios where the user aims to find the location of an asset
by retrieving the location information from its RFID tag, as explained in Subsection 3.1, he/she needs
to know his/her own location to be able to find the path to the target asset. There are two major
scenarios for the user to estimate his/her location from surrounding tags: (1) Scanning a visible tag:
The user scans a visible tag and reads the current location data of the tag. Consequently, the user is
provided with his/her current location on the floor plan. (2) Scanning the area: The user scans the area
and reads the location data of surrounding tags to be used for RFID reader localization technique.
Several RFID reader localization techniques are available to locate an RFID reader using reference
tags (Li and Becerik-Gerber, 2011). In our proposed framework, tags attached to fixed assets can act
as reference points for RFID reader localization techniques. Moreover, RFID tags can also store part
of the signal fingerprint database. In this method, tags are not necessarily visible or in a close
proximity.
3.3 Moveable assets localization
In our proposed method, fixed assets are used as reference points to help locating moveable assets.
The similarity of received RSSI between target tags and fixed-assets’ tags is used for localization. The
RSSI received from reference tags and from target tags are logged by a handheld RFID reader at
several locations and the received power for all signals are processed to determine the similarity
between signal strength patterns. Tags that show similar signal patterns are considered to be spatially

5. adjacent. This similarity of pattern stems from the fact that the propagated radio signals are affected
by similar environmental effects for neighbouring tags. Furthermore, this method does not use RSSI
values to estimate the distance between the reader and tags due to the unreliability of this conversion
in indoor environments.
In this method, fixed and target assets are equipped with long range, omni-directional and identical
tags. Tags attached to fixed assets contain their exact location coordinates. Moreover, it is assumed
that the target tags are stationary for the period of localization and the user equipped with a handheld
reader is moving within the facility to collect RSSI values and track assets.
Figure 2 shows the process flow to locate target tags: (1) The user starts searching for target assets
by scanning the area. (2) If the target tags are not in the range of the RFID reader, the user needs to
change his location to be able to detect the tags. (3) As soon as a target tag is detected by the reader,
user starts logging the signal strength from surrounding tags. The user remains stationary during data
logging for the time period of ∆t. (4) The logged data are processed by the data processing module
(Localization Engine). The RSSI values received from each tag are filtered for noise elimination.
Filtered RSSI values are averaged and used for pattern matching (discussed in Subsection 3.3.2). The
pattern matching algorithm identifies a set of reference tags that their behaviours resemble a target tag
and rank them based on calculated similarity indicator values. In order to identify the suitable number
of data logging sessions to localize the target assets (based on the predefined accuracy requirements),
the engine calculates the convergence level. Having adequate data for location estimation, the possible
locations for a target tag are identified. By applying the spatial constraints, some areas from the search
space are removed to narrow down the results. Finally the engine calculates the accuracy of the
estimated location based on the density of reference tags in the target area. (5) After estimating the
location of the target tags, these areas are shown on the floor plan. (6) If the logged data are not
adequate for accurately estimating the location, the localization engine prompts the user to move to a
new location and to continue logging data.
Is target tag
Y Log
Data Data Pattern Convergence
Location Accuracy
Visualization
Spatial
Start Scan Area Data Level on the floor End
detected? Filtering Averaging Matching Calculation Estimation Estimation
(RSSI) Constraints
1 2 3 5 plan
N
6 Change Move
Location (Move) required? Data Processing 4
Y N
Figure 2. Process flowchart for movable assets localization
3.3.1 Pattern matching algorithm
In order to formulate the data processing part of the proposed method, we assume that there are n
available reference tags in the building and Ri [i∈(1, n)] denotes the ith reference tag. A reference tag
is an RFID tag that stores its current coordinates. A target tag is an RFID tag that is the target for
localization and does not store its coordinates. Tj [j∈(1, p)] denotes the jth target tag. Data logging
happens at m different locations/time instances. A data logging session time series, Ls [s∈(1, m)] is a
series of averaged RSSI values recoded where the reader is stationary for the period of data logging
R
(∆t). Each Ls is composed of RSS s k (average of RSSI values for the kth reference tag at the sth data
T
logging session) and RSS s u (average of RSSI values for the uth target tag at the sth data logging
Rk Tu
session). P and P denote the signal patterns received from the kth reference tag and the uth target
tag respectively, for m data logging sessions.
P Rk = { RSS sRk : s∈(1, m)} , PTu = { RSS sTu : s∈(1, m)} Definition (1)
The goal of data processing is to determine which reference tags (Ri) shows similar signal pattern
to the signal pattern received from the target tags (Tj). Additionally, the acceptable value of m should
be identified by the data processing module to stop the process of RSSI logging. The least square

6. difference method is employed to rank the similarity of reference tags to the target tag.  Rk is the
u T
distance indicator value between the k reference tag and the u target tag. β values can be modified
th th
to show the pattern similarity percentage (α) between each target and reference tags after m data
logging sessions.
 n

 RSS  
m
Equation (1)  Ruk  1   Rk /   100 Equation (2)
Tu 2 T T Tu
 
Tu Rk
 RSS u
Ri
 
Rk s s
s 1 i 1
  R  
Tj
i j 1,.. p ; i 1,..n
Definition (2)
The matrix of β is constructed using the calculated values from Equation 1. β values in the uth row
of the matrix indicate the distance indicators for each reference tag to the uth target tag. The least β
value in each row shows the reference tag that is assumably closer to the associated target tag.
3.3.2 Development of a simulation environment
A simulation environment is developed in Matlab (MathWorks, 2012) in order to analyze the
applicability of our approach for various possible scenarios and setups. Furthermore, new
mathematical and procedural techniques (e.g., data filtering, other pattern matching techniques and
localization modules) can be added and tested. The simulation platform provides a flexible
environment to define and place multiple reference and target tags. Moreover, the floor plans of a
building can be imported to the simulated environment to help defining realistic routes in a complex
building that FM personnel may use in order to find assets. The generation of RSSI values in the
simulation uses Monte Carlo approach based on our field test results (Subsection 4.1).
4 Case studies
4.1 Testing RFID characteristics
In order to realize the proposed method for locating moveable assets, the characteristics of an
available RFID system are analyzed. Several tests are conducted to test the readability range and the
effect of various environment factors on our RFID tags. Long range active tags with the operating
frequency of 915 MHz have been used for the test. The first test has been performed at Concordia
Stinger Dome to examine the readability range and signal attenuation of tags in an obstacle-free
environment. An RFID tag was placed on a tripod and RSSI values were collected at various distances
from the tag. Figure 3(a) shows the decrease of RSSI values by increasing the distance. Available
tag’s antenna are omnidirectional (1/4-wave monopole with 2/3 vertical element and 1/3 horizontal
element). it is observed that the gain is higher in front on the same long axis of the tag. Figure 3(b)
shows that the standard deviations of RSSI values slightly increase as the distance between the tag
and the reader increases.
(a) Average values (b) Standard deviations
Figure 3. RSSI vs. distance relationship

7. 4.2 Case study for fixed assets localization
The case study for estimating the location of fixed assets and location-tags is presented in detail in
Motamedi and Hammad (2011). In that case study, active RFID tags with large memory are attached
to fire extinguisher cabinets. The developed mobile application detects tags using the handheld RFID
reader and shows the location of fixed assets on a preloaded floor plan.
4.3 Case studies for moveable assets localization
4.3.1 Obstacle-free environment (with line-of-sight)
This test is performed to test the applicability of the proposed method for tracking moveable assets in
a multi-tag indoor environment. The test has been conducted in an obstacle-free environment where
all tags were placed inside one room. The tags have been placed on the ground in a grid of 5 m  7.5
m. A target tag is placed randomly in the room with the distance of 70 cm from the closest reference
tag (R9, R12) and data were collected using a handheld reader at six data logging points forming a U
shaped route (black line in Figure 4(a)). The calculated β values of the target tag for all reference tags
are presented in Table 1. Figure 4(b) shows the same setup in the simulation environment. The RSSI
values were generated using our signal propagation model (Subsection 4.1) and are compared with the
actual measured data. In the Figure 4, the diameter of red circles around reference tags are inversely
proportional to their β values. The results show that R12 has the least β value in both field test and
simulation environment which demonstrates that our method is feasible. As can be seen in Table 1
and Figure 4, the simulated β values are systematically less than those of the test values. This can be
explained by the fact that the environmental factors of the space used in the test are different from the
ones of the test explained in subsection 4..
(a) Test results (b) Simulated results
Figure 4. Obstacle-free test
Table 1. β and α (β/α) values for field test and one instance of the simulation (obstacle-free environment)
R1 R2 R3 R4 R5 R6 R7 R8 R9 R10 R11 R12
Field Test 13.5/93 13.2/ 93 106.4/50 15.4/92 13.7/93 12.17/94 8.7/95 6.2/97 4.3/97 8.1/96 8.3/96 4.1/ 98
Simulation 11.1/88 11.3/88 10.5/89 10.2/89 10.3/89 8.6/91 8.2/91 6.1/93 4.1/95 7.6/92 4.4/95 3.8/ 96
4.3.2 Environment with obstacles (without line-of-sight)
In this test, five reference tags and two target tags were placed in different adjacent rooms of the
building. Figure 5 shows the setup for the test. The user with an RFID handheld took a route in the
corridors and collected data in eight data logging sessions. The data are collected for 30 seconds (∆t)
at each data logging point. The building materials are concrete, metal and wood and the rooms
contain several assets. The target tags were placed with the distance of approximately 1.5 meter from
the closest reference tag in the same room. Table 2 shows the calculated β and α values of the two
target tags for each of the reference tags. The results show that the proposed approach is capable of
detecting the closest tag in a cluttered environment where the signal strength cannot be converted to

8. distance using signal propagation formulas. In this scenario, the pattern matching between target and
reference tags is used for identifying the closest reference tags to the target tag.
Table 2. β/α (%) values for
environment with obstacle
Target 1 Target 2
R1 15.13/89 15.14/78
R2 5.60/96 14.22/79
R3 30.05/79 11.72/83
R4 47.32/67 13.92/79
R5 49.68/66 14.06/79
Figure 5. Test with obstacles (a) Tag 1 (c) Tag 2
5 Conclusions and future work
This paper investigated several localization methods based on our previously proposed framework. It
discussed how various types of assets can be localized in an RFID-equipped building. The paper
included the different scenarios to assist users (e.g. FM personnel or occupants) estimate their
locations as well as the location of assets they are looking for. Our proposed method for tracking
movable assets has several advantages over similar techniques. It uses available RFID tags in the
building (used for lifecycle management) for the localization purpose. The method does not require
any fixed infrastructure of RFID readers and operates using handheld units. The accuracy of the
localization is related to the density of reference tags and can be increased by adding tags to more
assets. The main advantage of the proposed system is that it can adapt to the changes in the
environment. The case studies showed promising results for location estimation of assets.
Future work of this research includes: (1) Developing methods to improve the accuracy of movable
asset localization; (2) Providing users with navigational aid including floor plans and routes that are
extracted from a BIM and saved on tags; (3) Improving the simulation environment to include the
effect of obstacles and noise on signal propagation; and (4) Integrating BIM with the simulation.
References
AIMGLOBAL, 2008. What is RFID. Available online: < http://www.aimglobal.org/technologies/rfid/what_is_rfid.asp>.
FUCHS, C., ASCHENBRUCK, N., MARTINI, P. and WIENEKE, M., 2011. Indoor tracking for mission critical scenarios:
A survey. Pervasive and Mobile Computing, 7 (1), pp.1-15.
HARRISON, M., 2007. Guidelines for lifecycle ID & data management. AEROID-CAM-014.
LI, N. and BECERIK-GERBER, B., 2011. Performance-based evaluation of RFID-based indoor location sensing solutions
for the built environment. Advanced Engineering Informatics, 25(3), pp. 535-546.
MATHWORKS, 2012. Matlab, Available online: <http://www.mathworks.com/products/matlab/>.
MOTAMEDI, A. and HAMMAD, A., 2009. Lifecycle management of facilities components using radio frequency
identification and building information model. Journal of IT in Construction, 14, pp. 238-262.
MOTAMEDI, A. and HAMMAD, A., 2011. Location management of RFID-equipped building components. International
Symposium on Automation and Robotics in Construction (ISARC), Seoul, Korea.
MOTAMEDI, A., SAINI, R., HAMMAD, A. and ZHU, B., 2011. Role-based access to facilities lifecycle information on
RFID tags. Advanced Engineering Informatics, 25 (3), pp. 559-568.
PAPAPOSTOLOU, A. and CHAOUCHI, H., 2011. RFID-assisted indoor localization and the impact of interference on its
performance. Journal of Network and Computer Applications, 34(3), pp. 902-913.