IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 1 Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Ranging with IEEE Narrow-Band PHY] Date Submitted: [14 September, 2009] Source: [Wolfram Kluge, Dietmar Eggert] Company: [Atmel] Address: [Koenigsbruecker Strasse 61, Dresden, Germany] [ Re: [Response to Call for Final Proposals] Abstract: [Proposal of using IEEE Narrow-Band PHY for Ranging and Localization] Purpose:[To present the method of performing ranging in a narrow-band transceiver using phase measurements] Notice:This document has been prepared to assist the IEEE P 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

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 2 IEEE PHY usage for Active RFID and Ranging Widely adopted for wireless sensor networks, home control and industrial automation and similar applications Proven technology Although narrow-band, it is suitable for ranging even under multipath environments Less additional hardware needed in existing transceiver design

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 3 IEEE PHY extensions needed Transmitting carrier for short times (blocking modulation) Phase measurement unit State machine to coordinate transmit and receive mode with appropriate timing  can be implemented in hardware or software

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 4 Advantage of Phase-Based Ranging Fits to narrow-band transceiver design – only carrier transmitted Any unknown delay in the transceiver (clock skew, filter group delay,…) has no impact on ranging accuracy No impact of channel filter group delay –Example: 2MHz, tg=325ns+/- 5%  +/- 16ns systematic error contribution by receiver –Corresponds to 4.8m systematic range error in ToA systems! –No impact on phase measurements, since all phases are measured at exactly the same IF frequency 10 to 20 times faster than Time-of-Arrival with IEEE compliant frames Needed to perform ranging measurements at multiple frequencies to mitigate multipath effect Fast scanning of multiple frequency channels allows tracking of moving objects Fast scanning saves power and improves batter life time

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 5 Choice of TX signal for phase measurements Phase measurements can be done with 1.Cross-correlation of IEEE frames Utilizing preamble spreading sequence Complex cross correlation in baseband domain yields phase of received signal One frame per frequency  at least about 300µs 2.CW carrier less than 1us needed for phase measurement Practically µs to allow PLL to settle completely  Much faster, therefore preferred

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 6 Active Reflector Principle (1) Device A initiates ranging measurement Device A transmits carrier  device B performs phase measurement changing transmit direction in both devices Device B transmits carrier  device A performs phase measurement Device B transmits frame with measurement results to Device A Device A is able to calculate range Bidirectional traffic needed for devices with asynchronous time base

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 7 Active Reflector Principle (2) PLL is running at same frequency at TX and RX mode Receiver measures phase between LO signal and received carrier Phase measurement can be done at any down-converted signal since frequency conversion maintains phase information Propose phase measurement at IF frequency in low-IF receiver

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 8 Ranging with Active Reflector TX signal phase of device B (reflector) must be the same as of the received signal.  hard to implement Proposal: Device B measures phase of receives signal relative to own LO signal phase. Phase difference is transferred to device A used as correction factor.

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 9 Ranging Procedure (1)

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 10 Ranging Procedure (2) Device A  Transmitting Ranging Request Frame  Receiving Ranging Ack  Locking AGC  Starting timer after RX end  Setting PLL to 1 st meas. freq.  Starting phase meas. sequence  Setting PLL to orig. freq.  Acking Result Frame  Releasing AGC Lock  Restoring IF position  Distance calculation Device B  Locking AGC after Request Frame receive  Transmitting Ranging Ack  Starting Timer after TX end  Correcting Frequency offset with PLL  Setting PLL to 1 st meas. freq.  Starting phase meas. sequence  Setting PLL to orig. freq.  Transmitting results frame  Receiving Ack  Releasing AGC Lock

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 11 Ranging Request Frame Device A sends Ranging Request Frame to device B. Parameters: Start frequency (2401 MHz … 2483 MHz) Stop frequency (2401 MHz … 2483 MHz) Step frequency (0.5 MHz, 1 MHz, 2 MHz) Slot time (0…255)*1  s Device B acknowledges this frame. Sequence of transmitting CW carriers starts after fixed Tprep measured from acknowledge frame end. After Tprep device B transmits data frame back to device A (two frames if more than 112 frequencies are used). Step frequency sets max. distance that can be measured. Fstep (MHz)0.512 Max. Dist. (m)

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 12 Ranging Results Device A calculates range by averaging method. The result is a 16-bit integer value representing the phase difference between Fstart and Fstop. Where c is the speed of light and phase is measured with an 8-bit integer value (2  == 256). This avoids division calculation in the  C. Integer result allows post processing of ranging date from different devices within the network by devices based on micro controllers (avoiding floating point calculation). Return value 0xFFFF proposed for invalid result.

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 13 Implementation Example of Phase Measurement Example: Low-IF receiver Phase difference measured between IF signal and divided clock signal Capturing time difference between signal edges (zero crossing of sine signals) Phase difference independent of time (for zero frequency offset between devices)

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 14 Distance Calculation by Averaging for line-of-Sight channel  Simple method to cope with multipath effects  Adding all  to reconstruct phase over 80MHz bandwidth  Distance calculation: Is identical to average group delay Issue:  f must be small enough to avoid cycle slip for largest distance

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 15 Multipath Propagation Most significant error in ranging measurements Narrow-band measurement (2MHz bandwidth) very prone to multipath channel (Corresponds to sampling of channel group delay curve at arbitrary frequency) Solution: gathering information over as a wide frequency band as possible About 80MHz in 2.4GHz ISM band Flexibility: Depending on severity of multipath propagation (ratio of LOS signal power to signal power in delay paths) the number of frequencies used can be chosen Issue with statistical channel models (like JTC-B) LOS component amplitude is Rayleigh distributed producing to some percentage a LOS signal below any detection level. In this case the channel does not contain the information to get the correct ranging results.

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 16 Distance Calculation by IFFT Measuring phase difference and RSSI value for each frequency Accumulating phase differences to reconstruct phase  (f) Generating complex baseband signal x(f) = RSSI(f)*exp(j  (f)) IFFT shows channel impulse response. Selecting 1st tap to identify LOS component Restriction: 1.875m resolution due to 80MHz bandwidth of ISM band (double distance measured) Higher computational effort than averaging, but more robust under harsh multipath environments (office or industrial environment)

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 17  80 phase  100 channel representations  Distance error for LOS multi-path environment (D=37.5m, Td=125ns)  LOS path with const. amplitude=1 in I and Q  Tap delays vary with 5ns rms Distance Error Simulation for line-of-Sight Channel Distance calculated from average group delay Distance Error less then +/- 0.5m

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 18 Distance Error Simulation for JTC-B Channel Model  Td=125ns (D=37.5m)  IFFT with 80samples   f=1MHz  1000 channels  Amplitude with 3dB resolution vs. no quantization  both 23 errors In 2.3% of statistical channel samples distance estimation fails. Reason: partly due to Gaussian amplitude of LOS path!!!

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 19 Distance Measurement Results Algorithm verified with cable connection Phase measured at 80 frequencies with 1MHz stepping 8 phase values measured per frequency Steepness is proportional to distance

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide b PHY as option to active RFID Option 1 RFID tag with UWB TX and 2.4GHz RX Best location accuracy Inherent robustness against multipath propagation Best choice for fixed industrial environment Option 2 RFID tag TX and RX 2.4GHz at RX Moderate location accuracy Longer range Utilizing existing PHY standard Utilizing existing MAC as protocol Enables location awareness in sensor networks Adds RFID functionality to sensor networks Allows RFID application w/o need of synchronous backbone infrastructure

IEEE f Submission Nov 2009 Wolfram Kluge, AtmelSlide 21 Summary Ranging with phase measurements fits to narrowband transceiver hardware utilized in IEEE devices Less hardware extensions needed to perform phase measurements Distance resolution not prone to transceiver group delay – no transceiver calibration needed Ranging at multiple channel frequencies allows mitigation of multipath effects