Content uploaded by Stavros Stavrou
Author content
All content in this area was uploaded by Stavros Stavrou on Jul 16, 2015
Content may be subject to copyright.
Accelerating TOA/TDOA packet based localization
methods
Amin Gholoobi
Information and Communication Systems
Faculty of Pure and Applied Sciences
Open University of Cyprus
Email: amin.gholoobi@st.ouc.ac.cy
Stavros Stavrou
Information and Communication Systems
Faculty of Pure and Applied Sciences
Open University of Cyprus
Email: stavros.stavrou@ouc.ac.cy
Abstract— Time-based localization methods have always been an
interesting topic in commercial and educational research
activities. Compared to Receive Signal Strength (RSS) based
methods, time-based localization methods can be especially useful
in dynamic environments where the surroundings can change
from time to time, a process that can heavily influence RSS based
methods. When it comes to time-based localization methods,
these types of methods are dependent heavily on the accuracy of
the hardware clocks. Typical 802.11 time-based methods collect a
large number of packets, which include timing information, and
apply statistical analysis in order to localize a user. This process
tends to be time consuming and in many cases is in the range of
minutes. This paper investigates ‘off the shelf’ ways to accelerate
time-based methods. By accelerating packet capturing and
packet validation it has been shown that the localization process
time of these methods can be shortened without reducing
positioning accuracy.
Keywords-component; TDoA, ToA, localization, indoor,
Beacon, Beacon Interval
I. INTRODUCTION
The fact that time-based methods are not easily affected by
changes in the environment and do not require fingerprinting
databases unlike Received Signal Strength methods, makes
them very attractive for complex dynamic environments such
as Airports, Shopping Malls, Hospitals or other dynamic
environments. On the other hand, practical implementation of
Real Time Location Systems (RTLS) with off the shelf devices
is a challenging task. One of the main challenges is the packet
collection and position evaluation process which takes more
time than RSS based methods [1]. In [2-5] different techniques are
presented that try to optimize the process. In [6] a time-based
RTLS was implemented by using high accuracy clocks. This
process has a high cost on the RTLS deployment since it is
hardware based.
Other researchers have tried to implement and use time-based
methods by creating customized packets and hardware solutions
[7-9] [16-17]. In [10] the method calculates the travel time of
two sets of sounds which have been generated by two mobiles
and estimate their in-between distance with an achieved accuracy
of 30cm. Unfortunately sound-based methods may not work
well in crowded areas where the environment is very noisy.
The proposed method in this paper accelerates packet capturing
and packet validation for Time of Arrival (ToA)/Time-
Difference of Arrival (TDoA) methods. Most of the time-based
methods need collection times from a couple of minutes up to
20–30 minutes [2-4]. Such time durations may restrict practical
usage of dynamic RTLS. In PinPoint [2] it takes about four
minutes to collect and evaluate the packets, while in [4] 15
minutes of collection Ping data was required in order to
achieve the suggested accuracy. As mentioned in [3], collection
and evaluation of data requires a large number of packets,
suggesting a collection process of several minutes. And, last
but not least, using our Beacon method [12] requires 2000 to
2500 valid Beacon packets for the best accuracy. Since the
default beacon generation interval in Access Points (APs) is
100 ms, several minutes are required to collect a significant
number of packets. The Beacon method has been tested in
different scenarios by generating various Beacon Intervals
(BIs) down to 10 ms in order to evaluate the effect of different
intervals on localization accuracy and speed. Faster Beacon
generation does not directly translate to faster and more
accurate localization results. Changing BI affects other
elements such as collision and throughput which are important
for better localization accuracy.
In this paper we present several measurements which are taken
at different BIs at 200, 100, 50, 20 and 10 ms at various
distances up to 30 meters.
II. DESCRIPTION OF THE METHOD
A. Background and observations
Software-based time localization methods are utilizing
different types of packets (RTS-CTS, Ping, Custom packets) [3-
4] [6-7] [13]. A correct packet type and an optimum packet
generation technique, can lead to faster and more accurate
distance estimation in software-based localization. This paper
focuses on presenting the results and achievements based on
the Beacon frame method [12].
In a previous investigation it was demonstrated [12] that in
order to achieve acceptable localization estimations, 2000-2500
valid packets are required. The majority of APs by default
generate Beacon packets every 100 milliseconds, suggesting 3–
4 mins of packet collection time.
In the presented method, broadcasted Beacon packets, captured
and used. After capturing the packets, ToA calculation is the
first step. The calculation with multiple nodes has been
explained. For a single node calculation, i is equal to one (i=1)
where i represents the ith node in the system. This calculation
ri = di + ni , i ∈ LOS
ri = di + bi + ni , i ∈ NLOS (2)
can be divided into two sets, LOS (Line of Sight) and NLOS
(Non Line of Sight) which can be presented as (1)
where B represents a set of available Access Points (APs)
which can be seen by a Mobile Terminal (MT).
As explained in [14] [15], the ToA can be calculated as follows:
where ri represents (c * ti) in meters (m); c being the speed of
light in free space (c = 3 * 108 m/s) and ti is the signal
propagation time between AP and MT in nanoseconds (ns); di
is the actual distance between the two nodes in meters (m) and
ni represents the error introduced by the processing delays in
(ns) * c, in meters (m). Finally, bi is the value of unknown
NLOS biases in meters (m).
For our project we have reformed (2) since we are interested in
estimating the distance by using ti, c and ni. The formula is as
follows:
where edi is the estimated distance of the ith position in meters
(m); ni (Processing delay in meters) and bi is the value of
unknown NLOS biases in meters (m) which is a constant
calculated by placing the MT in known distances and
calibrating the system in the localization environment.
Calculating bi for different environments needs advance ray
tracing techniques and tools but this is not within the scope of
this study. LOS formula has been used for all conducted
experiments in this paper.
B. Methodology
Since we are using Beacon packets, Beacon Interval (BI)
and Beacon Timestamp, the first step has been to minimize the
packet generation time or BI. By generating beacons faster,
more packets can be captured within the same timeframe. BIs =
200, 100, 50, 20 and 10 were generated and evaluated for
different distances.
Packets were captured by using Wireshark. Two different
timestamps including Beacon timestamp and MAC timestamp
have been used for ToA calculations. For each BI and distance,
300 seconds (5 minutes) of data have been gathered and a
measurement graph has been produced to demonstrate the
achieved localization accuracy.
For each test, the Beacon timestamp and MAC timestamp are
compared in order to obtain the valid packets. The Beacon
timestamp has been used in order to confirm that each packet
has been generated every N millisecond: N being the value of
the Beacon Interval (BI).
Table 1 shows all the Beacon Intervals which have been used
for this test and the equivalent microseconds needed for
filtering and calculation purposes.
BI
Value in µs
200
204800
100
102400
50
51200
20
20480
10
10240
Table 1 - BIs and Microsecond Values
III. EXPERIMENTAL SETUP
Experiments took place at the premises of the School of
Pure and Applied Sciences at the Open University of Cyprus.
The floor is constructed from a mixture of concrete pillars and
walls, including plasterboard walls. Figure 1 shows the floor
plan of the building.
Different APs and MTs have been used to confirm the
consistency of the method. APs which have been used include
the following: Linksys WRT54GL, UBIQUITI BULLET M2
and Cisco E4200. Mobile terminals consisted of a Sony Vaio
VPCYB2M1E and a DELL INSPIRON N5110. APs have been
loaded with DD-WRT V24 SP1 and OpenWRT backfire
firmware.
The Mobile terminal was running Fedora 17 and tshark v1.6.3.
Experiments were carried out at 2.4GHz and 5GHz bands, on
channel 11(2462MHz), channel 13(2472MHz) and channel
36(5180MHz).
A. Localization platform
A localization platform has been developed in order to
speed up the packet collection and evaluation process. The
platform is able to start an automatic capturing session, capture
the packets, filter and evaluate them, and export the result to a
database for further analysis.
Figure 1. Floor plan
LOS := {i | i ∈ B, ith link in LOS}
NLOS := { i | i ∈ B, ith link in NLOS} (1)
, i ∈ LOS
, i ∈ NLOS (3)
B. Measurement set 1
The first set of measurements was conducted in the
corridors of the School of Pure and Applied Sciences at the
Open University of Cyprus (OUC) at a maximum distance of
30 meters and with LOS (Line of Sight) conditions. Table 2
shows the calculation of the ToA0 for capturing time from 300
seconds down to 1 second with BI=100ms. Based on formula 1
in [12] Table 2 will provide all the ToA0 numbers for the
respective TDoA calculations and estimating the distance.
TOA0
(µs)
Capture
time
(Sec)
# of
Captured
Packets
# of
Relevant
Packets
# of Valid
Packets
102401.1877
300
5466
2916
2019
102401.1848
250
4663
2428
1661
102401.1868
200
3772
1947
1344
102401.1801
150
2844
1458
994
102401.1789
100
1925
973
665
102401.1718
50
1023
487
326
102401.1773
25
497
243
141
102401.1758
15
309
146
91
102401.1857
10
211
97
70
102401.1667
5
138
49
36
102401
1
42
10
6
Table 2. Calculated TOA0 for BI=100
Table 3 shows the capturing analysis and estimated distances
for 10 meters with BI=100ms.
Capture time
(Sec)
# of
Relevant
Packets
# of
Valid
Packets
Est dis
(m)
σ
(m)
300
2918
2008
8.1
1.9
250
2432
1854
7.8
2.2
200
1945
1501
7.0
3
150
1460
883
5
5
100
973
582
4.2
5.8
50
485
283
4.6
5.4
25
241
146
19.8
9.8
15
143
81
25
15
10
96
49
34
24
5
48
17
45
35
1
10
5
60.0
50
Table 3. Collected data, analysis and estimated distance
for 10m for BI=100
In Table 3 it can be seen that as the number of valid packets
start to drop below 1000, the estimated distance becomes
unreliable. This demonstrates that at least 1000 valid packets
are required for acceptable distance estimation. Table 4 results
are next presented for BI=20ms.
TOA0
(µs)
Capture
time
(Sec)
# of
Captured
Packets
# of
Relevant
Packets
# of Valid
Packets
20480.24302
300
17534
14409
8847
20480.24308
250
14692
12013
7368
20480.24439
200
11689
9614
6056
20480.24424
150
8806
7215
4561
20480.24187
100
5891
4811
3043
20480.23677
50
2966
2404
1474
20480.23207
25
1497
1197
1007
20480.2427
15
911
722
445
20480.24573
10
624
484
293
20480.22517
5
337
244
151
20480.23529
1
62
48
34
Table 4. Calculated TOA0 for BI=20
Table 5 shows the data, analysis and estimated distance for 10
meters with BI=20ms.
Capture time
(Sec)
# of
Relevant
Packets
# of
Valid
Packets
Est dis
(m)
σ
(m)
300
14469
8528
13.2
3.2
250
12060
7445
13.5
3.2
200
9652
6227
13
3
150
7243
5048
14.1
4.1
100
4830
3521
13.3
3.3
50
2417
1763
14.5
4.5
25
1209
1015
13.8
3.8
15
724
516
19.8
9.8
10
484
374
32
22
5
242
200
2
8
1
47
41
66
56
Table 5. Collected data, analysis and estimated distance
for 10m for BI=20
Figures 2 and 3 show the number of valid packets for different
collection times for BI=100ms and BI=20ms. By looking at
these two figures, it can clearly be seen that when BI=20 the
collection time and positioning estimation is reliably
accelerated.
Figure 2. Calculated TOA0 in different time for BI=100
The results recorded in Figure 3 show a collection of more
than 1000 packets was achieved in 25 Seconds. The outcome
is almost 8 times faster than using the normal BI at 100 ms,
which is typically the default BI for commercial APs.
C. BI Selection and performance measurement
This section explains Beacon Interval (BI) selection methods
and why it is important to have a correct interval for this type
of localization technique.
Two of the most important functionalities of Beacon packets
are:
Synchronization of WLAN (Wireless Local Area
Network)
Advertise the SSID (Service Set IDentifier) of the
network
Wireless channels have bandwidth limitations. A Beacon
packet uses and occupies from the available bandwidth of the
channel. Setting the BI according to the available bandwidth,
plus the number of clients and also exercising care for the PSM
(power-saving mode) leads to a greater performance in wireless
local area networks (WLAN).
Reducing the BI without considering other options, introduces
overhead to the network. As a consequence it can trigger drop-
off in throughput and Quality of Service (QOS). By affecting
throughput and QOS negatively, more delays are introduced
which change the value of ni dramatically and which may result
in a less accurate estimation. On the other hand increasing the
BI by too much, forces the clients holding traffic for Beacon to
wait longer which can create periods of “dead air”.
Evaluating different Beacon intervals and selecting the
appropriate one, helped to reduce collisions which leaves more
valid packets in a shorter period of time.
In Figure 4, it can be seen that as BI changes, it affects the
throughput and directly impacts on the number of valid
packets.
IV. CONCLUSION
This work suggests that by reducing the BI, packet time-based
localization methods can be significantly accelerated.
By using the Beacon packets and adjusting the Beacon interval,
more valid packets can be generated and captured in much less
time. Additionally, it has been validated that the localization
system can localize almost 8 times faster when compared to
existing methods.
ACKNOWLEDGMENT
Research work (PENEK/0311/27) is co-funded by the
European Regional Development Fund (ERDF) and the
Republic of Cyprus.
REFERENCES
[1] Hui Liu; Darabi, H.; Banerjee, P.; Jing Liu, "Survey of Wireless Indoor
Positioning Techniques and Systems," Systems, Man, and Cybernetics,
Part C: Applications and Reviews, IEEE Transactions on , vol.37, no.6,
pp.1067,1080, Nov. 2007.
[2] Mah, M.; Gupta, N.; Agrawala, A., "PinPoint Time Difference of
Arrival for Unsynchronized 802.11 Wireless Cards," INFOCOM, 2010
Proceedings IEEE , vol., no., pp.1,5, 14-19 March 2010.
[3] Christian Hoene and Jörg Willmann. Four-way TOA and software based
trilateration of IEEE 802.11 devices. In IEEE PIMRC, Cannes,
September 2008.
[4] A. Günther and Christian Hoene. Measuring round trip times to
determine the distance between WLAN nodes. In Networking 2005,
Waterloo, Canada, May 2005.
[5] McCrady, D.D.; Doyle, L.; Forstrom, H.; Dempsey, T.; Martorana, M.; ,
"Mobile ranging using low-accuracy clocks," Microwave Theory and
Techniques, IEEE Transactions on , vol.48, no.6, pp.951–958, June
2000.
[6] Moustafa Youssef, Adel Youssef, Chuck Rieger, Udaya Shankar and
Ashok Agrawala. 2006. PinPoint: An Asynchronous Time-Based
Location Determination System. In Proceedings of the 4th international
conference on Mobile systems, applications and services (MobiSys '06).
ACM, New York, NY, USA, 165-176.
[7] Alfonso Bahillo, Patricia Fernández, Javier Prieto, Santiago Mazuelas,
Rubén M. Lorenzo and Evaristo J. Abril. 2011. Distance Estimation
based on 802.11 RTS/CTS Mechanism for Indoor Localization,
Advances in Vehicular Networking Technologies, Dr Miguel Almeida
(Ed.).
Figure 3. Calculated TOA0 for BI=20
Figure 4. Valid packets and Throughput,
Test time = 100 Seconds
[8] A. Bahillo, J. Prieto, S. Mazuelas, R.M. Lorenzo, J. Blas and P.
Fernandez. "IEEE 802.11 Distance Estimation Based on RTS/CTS Two-
Frame Exchange Mechanism," Vehicular Technology Conference, 2009.
VTC Spring 2009. IEEE 69th, vol., no., pp. 1,5, 26–29 April 2009.
[9] S. Wibowo, M. Klepal and D. Pesch. Time of flight ranging using off-
the-shelf wifi tag. In: Proceedings of PoCA 2009.
[10] Chunyi Peng, Guobin Shen, Yongguang Zhang, Yanlin Li and Kun Tan.
"BeepBeep: A High Accuracy Acoustic Ranging System Using COTS
Mobile Devices," ACM SenSys 2007, Nov. 2007, Sydney, Australia.
[11] K. Pahlavan, X. Li and J. Makela. "Indoor geolocation science and
technology," IEEE Commun. Mag., vol. 40, no. 2, pp. 112–118, Feb.
2002.
[12] Amin Gholoobi and Stavros Stavrou. "A hybrid TDoA-ToA localization
method," 20th International Conference on Telecommunications (ICT),
May 2013.
[13] M. Ciurana, F. Barcelo and F. Izquierdo. "A ranging system with IEEE
802.11 data frames," in Proc. IEEE RWS '07, Jan. 2007.
[14] Neal Patwari, Joshua N. Ash, Spyros Kyperountas, Alfred O. Hero III,
Randolph L. Moses and Neiyer S. Correal. "Locating the nodes:
cooperative localization in wireless sensor networks," IEEE Signal
Processing Magazine, vol. 22, no. 4, pp. 54–69, July 2005.
[15] Y. Qi, H. Kobayashi, and H. Suda. "Analysis of wireless geolocation in
a non-line-of-sight environment," IEEE Trans. Wireless Commun., vol.
5, no. 3, pp. 672–681, Mar. 2006.
[16] Noha El Gemayel, Sebastian Koslowski, Friedrich Jondral and Joachim
Tschan. A low cost TDOA localization system: Setup, challenges and
results. WPNC. 1–4, IEEE, Year 2013.
[17] S.B. Wibowo, M. Klepal and D. Pesch. Time of Flight Ranging Using
Off-the-Shelf IEEE 802.11 Wi-Fi Tags. In Proceedings of the
International Conference on Positioning and Context-Awareness
(PoCA’09), Antwerp, Belgium, 28 May 2009.
[18] Atheros, "ath5k driver." http://linuxwireless.org/en/users/Drivers/ath5k,
viewed 21-03-2013.
[19] Atheros, "ath9k driver." http://linuxwireless.org/en/users/Drivers/ath9k,
viewed 23-03-2013.
[20] Suman Nath, Zachary Anderson, and Srinivasan Seshan. Choosing
beacon periods to improve response times for wireless HTTP clients, in
MobiWac '04: Proceedings of the second international workshop on
Mobility management & wireless access protocols, ACM, New York,
NY, USA, 2004.
[21] R. Want, A. Hopper, V. Falc˜ao and J. Gibbons. "The active badge
location system," ACM Trans. Inf. Syst., vol. 10, pp. 91–102, Jan. 1992.