- Количество слайдов: 15
Project: IEEE P 802. 15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Outline for Annex D: Location Topics] Date Submitted: [January 17, 2006] Source: [Camillo Gentile] Company: [NIST] Address: [100 Bureau Drive, Gaithersburg, MD, USA] Voice: [+301 -975 -3685] FAX: [+301 -590 -0932], E-Mail: [camillo. [email protected] gov] Re: [802. 15. 4 a CFP] Abstract: [802. 15. 4 a CFP response] Purpose: [Response to WPAN-802. 15. 4 a CFP] Notice: This document has been prepared to assist the IEEE P 802. 15. It is offered as a basis for discussion and is not binding on the contributing individual(s) or organization(s). The material in this document is subject to change in form and content after further study. The contributor(s) reserve(s) the right to add, amend or withdraw material contained herein. Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE and may be made publicly available by P 802. 15.
Status of Informative Annex D: Location Topics • Three presentations this week: -15 minutes today - 60 minutes on Tuesday - 60 minutes on Thursday • Activity since Vancouver: - Nominated editor at last meeting - Solicited participation through reflector right after meeting - Prepared draft for letter ballot • Contributors so far: - Y. Qi, H. -B. Li, M. Umeda, S. Hara and R. Kohno, NICT - V. Brethour, Time Domain - R. Roberts, Harris Corporation - S. Emami, C. A. Corral and G. Rasor, Freescale Semiconductor • Get more direction this week for scope of document
Annex D: Outline 1 Parameter (Range or Direction) Estimation Techniques 1. 1 Range Estimation 1. 1. 1 Time of Arrival (TOA) 1. 1. 2 Time Difference of Arrival (TDOA) 1. 1. 3 Time Sum of Arrival (TSOA) 1. 1. 4 Signal Strength Ranging (SSR) 1. 2 Direction Estimation (AOA) 1. 3 Combined Range and Direction Estimation 2 Location Estimation Algorithms 2. 1 Location Estimation from Range Data 2. 1. 1 Isolating outliers in range data 2. 2 Location Estimation from Direction Data 3 Leading Edge Detection 3. 1 Coherent delay estimation with low sampling rate and feasible ADC implementation 3. 2 A First-arrival Detection Method 4 Error Induced by Timing Crystal Offset 4. 1 The effect on the preamble and channel-sounding segments 4. 2 The effect on ranging accuracy
Positioning from TOA 3 anchors with known positions (at least) are required to retrieve a 2 D-position from 3 TOAs Anchor 2 (x. A 2, y. A 2) Anchor 1 (x. A 1, y. A 1) Specific Positioning Algorithms Measurements Estimated Position Mobile (xm, ym) Anchor 3 (x. A 3, y. A 3)
TOA Message Exchange • Synchronous one-way ranging (OWR) (Doc. 15 -04/0581 r 7): • Asynchronous two-way ranging (TWR) (Doc. 15 -04/0581 r 7): Unknown propagation delay Unknown clock offset Message 1 Message 2 Device A • Device B Asynchronous symmetric doubly-sided two-way ranging (SDS-TWR) (Doc. 1505/0246 r 1): Device B Device A – - “The Impact of the Relative Crystal Drift on The Performance of Symmetric Double-Sided Two- Way Ranging, ” S. Emami, C. A. Corral, and G. Rasor
Positioning from TDOA Anchor 3 (x. A 3, y. A 3) Anchor 2 (x. A 2, y. A 2) 3 anchors with known positions (at least) are required to find a 2 Dposition from a couple of TDOAs Mobile (xm, ym) Specific Positioning Algorithms Measurements Estimated Position Anchor 1 (x. A 1, y. A 1)
TDOA can operate in one of two modes … Mode 1 – The mobile station receives multiple reference pulses from fixed base stations synchronized through a controller The mobile carries the burden of running the hyperbolic location algorithms • Mode 2 – The mobile station transmits a reference pulse which is received by multiple fixed base stations The fixed stations must forward the TDOA information to a controller which runs the hyperbolic location algorithms FIXED MOBILE Key: Sync Pulse Location Pulse Position Report Key: Sync Pulse Location Pulse TDOA backhaul controller Mode 1 - Passive Mode 2 - Active
Ranging based on RSS • In past 5 years, considerable effort towards the design of COTS location systems: - “A Probabilistic Approach to WLAN User Location Estimation, ” T. Roos, P. Myllmymaki, H. Tirri, P. Miskangas, and J. Sievanan, Intl. Journal of Wireless Information Networks 2002 - “Robotics-Based Location Sensing using Wireless Ethernet, ” A. M. Ladd, K. E. Bekris, A. Rudys, G. Marceau, L. E. Kavraki, and D. S. Wallach, Mobi. Com 2002 - “Robust Location using System Dynamics and Motion Constraints, ” ICC 2004 - “Practical Robust Localization over Large-Scale 802. 11 Wireless Networks, ” A. Haeberlen, E. Flannery, A. M. Ladd, A. Rudys, D. S. Wallach, L. E. Kavraki, Mobi. Com 2004 • Best systems offer between 1 -2 m location precision => Not bad, but requires a-priori training • Would anyone buy a TG 4 a-compliant radio to build an RSS-based location system? - TOA-based mechanism more accurate and requires no training - RSS measurements would not even improve system
Any interest in seeing results on… • …Near-Field Electromagnetic Ranging (NFER)? - based on proprietary Q-track radios - for 3 -10 GHz, offers sub-meter range • … Angle-of-Arrival (AOA)? - conventionally narrowband - does not work well indoors • … Combined TOA and AOA? - promising as a single-device location system - testing at NIST for firefighters caught under rubble
Network Location Algorithms 1. Ad hoc (local) approaches: • “Ad Hoc Positioning System (APS), ” Globecom 2001 • “Locationing in Distributed Ad-Hoc Wireless Sensor Networks, ” ICASSP 2001 2. Global minimization approaches: • “The Bits and Flops of the N-hop Multilateration Primitive for Node Localization Problems, ” Mobi. Com 2002 • “Localization from Mere Connectivity, ” Mobi. Hoc 2003 3. Convex optimization approaches: • “Semidefinite Programming for Ad Hoc Wireless Sensor Network Localization, ” IPSN 2004 • “Sensor Location through Linear Programming with Triangle Inequality Constraints, ” ICC 2005
Coherent delay estimation with low sampling rate and feasible ADC implementation (Doc. 15 -05/0524 r 1) easy to implement LPF BPF Matched to Gaussian pulse LPF LO Difficult to implement codecorrelator ADC Spreading codecorrelator output ADC π/2 • Conventional approach to leading-edge detection: Correlate known and received signals in analog domain and then convert to digital domain • Efficient approach: Convert received signal to digital domain and then correlate with known digital signal
Simplified Maximum Likelihood Estimation h(tn) correlation function h(tm+1) h(tm+Z-1) h(tm) tm+1 tm+2 tm+Z • Use samples near the peak to fit a Taylor series expansion of the autocorrelation function tn
A First-arrival Detection Method (Doc. 15 -05/0406 r 0) h(t) Removed t • The SMLE is used to detect peaks in the received multi-path profile, each with time and magnitude • Iteratively remove the largest peak in the profile until the largest is the first
Effect of timing crystal offset on the channel-sounding segment (Doc. 15 -05/0335 r 1) • PRF=500 MHz => 5 ns pulse width • Channel-sounding segment length = 1024 pulses • Pulse-event repetition rate = 430 ns Takes 0. 44 ms to integrate the segment! • For X=10 ppm, the worst-case drift places the integration points 8. 8 ns apart 5 ns • For X=2. 5 ppm, the worst-case drift place the Integration points 2. 2 ns apart • Rather, tracking the channel-sounding segment allows a cheap radio (large X)! • But how about the preamble which can’t be tracked?
Effect of timing crystal offset on the preamble • PRF=500 MHz => 5 ns pulse width • Preamble length = 128 pulses • Pulse-event repetition rate = 430 ns Takes 0. 055 ms to integrate the preamble! • For X=20 ppm, the worst-case drift places the integration points 2. 2 ns apart (acceptable!) Þ It’s the preamble that constraints the quality of the crystals, not the channel-sounding segment