A UWB Transmit-Only Based Scheme for Multi-tag Support in a. Millimeter Accuracy Localization System. Nathan C. Roweâ, Aly E. Fathyâ, Michael J. Kuhnâ , ...
A UWB Transmit-Only Based Scheme for Multi-tag Support in a Millimeter Accuracy Localization System Nathan C. Rowe∗ , Aly E. Fathy∗ , Michael J. Kuhn† , Mohamed R. Mahfouz† ∗ EECS † MABE
Department, University of Tennessee, Knoxville, TN 37996, USA Department, University of Tennessee, Knoxville, TN 37996, USA TABLE I C OMPARISON OF COMMERCIAL AND RESEARCH UWB LOCALIZATION SYSTEMS .[1], [2], [3], [4], [5], [6]
Abstract— Ultra Wideband (UWB) wireless positioning systems have many advantages for tracking and locating items in indoor environments. Surgical navigation and industrial process control are potential applications for high accuracy UWB localization systems with millimeter accuracy. Several experimental systems have achieved very high accuracy, but have generally neglected significant system features like multi-tag access. This paper outlines a UWB localization system, which addresses both multi-tag performance and localization accuracy using UWB transmitonly digital communication and time difference of arrival (TDOA). The scheme takes advantage of a digital sampling circuit that can be used for both sub-sampling and data reception. Preliminary 1-D experimental results using this system resulted in a total system update rate of 889 Hz and localization accuracy of 3.25mm.
Company/Group Commercial Zebra Decawave Research Low et al. Zetik et al. Meier et al. Kuhn et al. This Work *1-D Result
Multi-tag Throughput (Events/Sec)
Localization Accuracy (mm)
3500 11000
300 100
N/A N/A N/A 744 889
4.8-14 15 0.1-2 3 3.25
I. I NTRODUCTION chronously as transmit-only devices in order to save on cost and power requirements at the tag. This introduces the challenge of collisions inherent in asynchronous transmitonly systems. The scheme utilizes a UWB transmitter and receiver designed for the 3.1-10.6GHz FCC approved UWB band and a high speed digital sampling board. This eliminates the need for an additional 2.4GHz radio link and operates only in the FCC approved UWB band. Currently 1-D results have been obtained for two tags operating simultaneously. Future work will expand the system to 3-D localization with a focus on improved accuracy and better multi-tag performance. The remainder of this paper is organized as follows: Section II describes the theoretical limitations of the system including collisions resulting from asynchronous operation. Section III introduces the localization method, sampling hardware, and multi-tag access scheme. Section IV provides the current experimental results. Finally, Section V gives an overview of future work and conclusions.
There are many potential uses for a high accuracy localization system such as industrial process control, high value asset tracking, and surgical navigation. A number of commercial and experimental systems exist that utilize Ultra wideband (UWB) for this purpose. The known commercial systems support multiple-tags under robust system architectures, but to-date have been limited in localization accuracy of 10-30 cm. Experimental systems have pushed the localization accuracy much lower, even sub-millimeter, but have generally neglected the need for a robust system that can support multiple-tags in near simultaneous operation. In the one exception, a secondary 2.4GHz radio has been used for tag control to implement a simple time division multiple access scheme [1]. A summary of the state of the art in both commercial and experimental UWB localization is provided in Table I along with results of this work. This paper discusses research into a high accuracy localization system with support for multi-tag operation. The scheme presented utilizes common hardware to accomplish both time difference of arrival (TDOA) measurements used in determining location and digital data transmission for tag identification. A subsampling method is utilized for time expansion to improve accuracy of TDOA measurements, but also extends the localization time and therefore limits the multi-tag performance. Because subsampling is used to improve accuracy this system cannot adhere to the 802.15.4a standard. In our scheme tags operate asyn-
978-1-4673-3105-0/13/$31.00 © 2013 IEEE
II. T HEORETICAL L IMITATIONS Common performance metrics that may be considered for a multi-tag localization system include location accuracy, number of supported tags, and tag refresh rate. Accuracy measures are typically expressed as the rootmean-square (RMS) error in position. The number of tags, K, and refresh rate, R, describe the multi-tag performance of the system and typically scale inversely for most systems. It is convenient for comparison to combine these into
7
WiSNet 2013
Multi tag Performance of an Asynchronous System
a single figure of merit, measured in events per second (EPS), by multiplying them as seen in equation (1). Multi tag Performance (EPS)
(EP S) = KR
2000
(1)
A primary limitation of asynchronous transmit-only systems is the possibility of collisions between transmitters. Collisions occur because the tags operate asynchronously with no knowledge of each other or the transmission channel. An equation for the probability of collisions for this system has been adopted from [7] as seen in equation (2) where Pc is the probability of collisions, Tp is the localization time or packet duration, Tf is the frame duration which is the inverse of the transmit rate RT , and K is the number of tags. Pc (K) = 1 − (1 −
2Tp K−1 Tf )
0 < Tp ≤ 0.5Tf
1500
1000
500
0 0
500
1000 1500 Tag Transmit Rate (Hz)
2000
Fig. 1. Theoretical comparison of the multi-tag performance of the two tag system with packet duration of 270µS.
(2)
The theoretical performance of the proposed asynchronous transmit-only system is limited by the probability of collisions. Equation (3) gives the multi-tag performance in (EPS) of the system for two tags with a transmission rate of RT and a probability of collisions Pc evaluated at K = 2. We also introduce a term Pr for the probability of collision recovery. Packet collisions can be recovered with some probability because the low duty cycle of the UWB transmissions makes it very unlikely that actual pulses will overlap in time. For this work the Pr is assumed close to 50% because one or the other tag transmission is typically recovered during each collision. Figure 1 shows the multi-tag performance as calculated for a range of tag transmission rates. The maximum multi-tag performance for 50% probability of recovery is achieved at the maximum transmission rate of 1845Hz with 100% probability of collisions. Very high collision rates will negatively affect system accuracy, and this will become the limiting factor in selecting a transmission rate. (EP S) = 2[RT − Pc (2)RT (1 − Pr )]
No Recovery (Pr=0) Half Recovery (Pr=0.5)
Fig. 2. The transmission sequence for a single localization packet containing the preamble and data.
Each tag transmission sequence, or packet, consists of a preamble period used for localization followed by transmission of the tag id as shown in Figure 2. The transmit rate defines a series of frames which each contain a single packet, but the actual transmission time of the packet is randomized within the frame. A packet is made up of a series of UWB pulses produced at a constant rate of 10MHz. The preamble segment consists of unmodulated UWB pulses. The base station subsamples this preamble to reconstruct the pulse shape which allows the identification of the peak position for use in the localization algorithm and to set the offset between the base station sampling clock and received pulse necessary for data reception. This allows the following pulses to be sampled in real time. The preamble is followed by the tag id transmission using onoff keying. The base station is able to sample each period and recover the digital data by comparing the samples with a predefined threshold. The tag, based on the design in [1], consists of a data source, a 10MHz clock, a 300 ps pulse generator, mixer, and medium power amplifier. The base station utilizes a digital sampling circuit as shown in Figure 3. The sampling circuit, as described in [8], consists of an FPGA, a digital programmable delay chip, a 150MHz clock source, and a fast high bandwidth ADC. The sampling circuit can be used in both a subsampling mode and a real-time sampling
(3)
III. M ULTI -TAG ACCESS V IA UWB OOK C OMMUNICATION The system utilizes subsampling to accurately measure TDOA at multiple base-stations for localization. On-Off Keying of the UWB signal is used for digital communication of 16 bit tag ids. Tags operate in an asynchronous transmit-only mode which results in a probability of random collisions. We utilize time hopping at the packet level to prevent catastrophic collisions, in which two tags could collide repeatedly for a period of time effectively blocking communication for either tag. This section begins by describing the transmission sequence, followed by discussion of the tag and base station hardware.
8
Fig. 3. Digital sampling circuit used for both localization and tag id transmission. Fig. 4.
mode. In the subsampling mode a sample resolution of less than 10 ps can be achieved based on the minimum programmable delay. In the real-time sampling mode, a fixed delay is set such that the sampling clock is synchronized with the center of the received pulses. This fixed delay is calculated based on the peak of the pulse shape reconstructed during subsampling. This allows every period of the transmitted signal to be recovered resulting in tag id transmission at 10 Mbit/s.
Experimental setup for testing 1-D system operation.
TABLE II E XPERIMENTAL RESULTS COMPARING THIS WORK WITH THE 2.4GH Z BASED MULTI - TAG SYSTEM DESCRIBED IN [1]. Kuhn et al.
IV. E XPERIMENTAL R ESULTS Experimental results have been achieved for 1-D operation using the setup shown in Figure 4. The results demonstrate simultaneous operation of two tags with successful transmission and reception of the tag ids. Two 16 bit tag ids were selected at random to be 12483 and 45655. Accuracy data was recorded for only one of the tags while the other tag was left transmitting in a static location. The tags were both operated at a transmission rate of 522Hz which was selected due to a noticeable deterioration of accuracy at higher rates. One minute experiments were conducted with the tag at 10 different locations between 2 base stations. The first 5 points are used for calibration. An Optotrak 3020 system with localization accuracy of 0.3mm was used as a location reference for accuracy measurements. The combined average throughput for both tags was 889.1 events per second, and the average localization error after filtering with a 160 sample averaging window was 3.25mm. Table II compares this system with a previous 2.4GHz based multi-tag system described in [1], operated in a similar 1-D experiment.
This Work
Filter
Accuracy
Throughput
Accuracy
Throughput
Window
(mm)
Total (Hz)
(mm)
Total (Hz)
80
3.86
583.9
4.75
889.1
160
3.72
587.6
3.25
889.1
ACKNOWLEDGMENT This work was supported in part by NSF grant ECCS1002318. R EFERENCES [1] M. Kuhn, M. Mahfouz, J. Turnmire, Y. Wang, and A. Fathy, “A multi-tag access scheme for indoor uwb localization systems used in medical environments,” in Biomedical Wireless Technologies, Networks, and Sensing Systems (BioWireleSS), 2011 IEEE Topical Conference on, jan. 2011, pp. 75 –78. [2] Zebra. (2011, November) Dart uwb hub and sensors. [Online]. Available: http://www.zebra.com/id/zebra/na/en/ documentlibrary/product brochures/dart sensors.html [3] DecaWave, “Scensor–precision location ultra low power transceiver,” DecaWave, Advanced Product Information D0801004DS7, 2011. [4] Z. Low, J. Cheong, C. Law, W. Ng, and Y. Lee, “Pulse detection algorithm for line-of-sight (los) uwb ranging applications,” Antennas and Wireless Propagation Letters, IEEE, vol. 4, pp. 63 – 67, 2005. [5] R. Zetik, J. Sachs, and R. Thoma, “Uwb localization - active and passive approach [ultra wideband radar],” in Instrumentation and Measurement Technology Conference, 2004. IMTC 04. Proceedings of the 21st IEEE, vol. 2, may 2004, pp. 1005 – 1009 Vol.2. [6] C. Meier, A. Terzis, and S. Lindenmeier, “A robust 3d high precision radio location system,” in Microwave Symposium, 2007. IEEE/MTT-S International, june 2007, pp. 397 –400. [7] C. Y. Jung, J. W. Chong, Y. J. Hong, B. C. Jung, and D. K. Sung, “Orthogonal time hopping multiple access for uwb impulse radio communications,” in Asia-Pacific Conference on Communications, October 2005. [8] Q. Liu, Y. Wang, and A. Fathy, “A compact integrated 100 gs/s sampling module for uwb see through wall radar with fast refresh rate for dynamic real time imaging,” in Radio and Wireless Symposium (RWS), 2012 IEEE, jan. 2012, pp. 59 –62.
V. C ONCLUSION The system described here compares well with previous results from a multi-tag system based on 2.4GHz communication. The scheme provides the advantage of operating using a single UWB transmitter and receiver to reduce system cost, complexity, and size while also providing for a more robust platform based on a single channel for localization and communication.
9