Date Submitted: [24 June 2005] 1 December 2018 Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [Non-coherent ranging results with a-priori knowledge of noise variance] Date Submitted: [24 June 2005] Source: [Ismail Guvenc, Zafer Sahinoglu, Andy Molisch, Philip Orlik, Mitsubishi Electric] Contact: Zafer Sahinoglu Voice:[+1 617 621 7588, E-Mail: zafer@merl.com] Abstract: [This document provides performance results of non-coherent ranging receivers, under the assumption that noise variance is accurately estimated and available a-priori] Purpose: [To help objectively evaluate ranging proposals under consideration] Notice: This document has been prepared to assist the IEEE P802.15. 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 P802.15.

Option-III (Ternary Sequences) Proposed System Parameters (With Same # Pulses per unit time) (by MERL) Option-I (Burst PPM) The Other Bit One Bit Always Empty Always Empty Always Empty 4-pulses 4 pulses Option-III (Ternary Sequences) ………………………… 1 2 3 31 4 5 6 7 8 30 Pulse Repetition Interval ~ 62.5ns Option-IV (Pulse PPM) Tp = 4ns Tf = ~125ns PRP ± TH

Energy Detection Receiver Architectures "Path-arrival dates" table 1D to 2D Conversion Assumption path synchronization Matrix Filtering + Assumption/path selection Time base 1-2ns accuracy Time stamping Analog comparator FT R&D BPF ( )2 LPF / 2-4ns integrator ADC TOA Estimator Sliding Correlator Energy combining across symbols interference suppression 1D-2D Conversion 2D-1D Conversion Energy image generation Bipolar template I2R 1D to 2D Conversion Length-3 Vertical Median or Minimum Filtering Removes interference 2D to 1D Conversion with Energy Combining Energy image generation MERL

Simulations Observation window = 512ns TOA Ambiguity = 256ns 1 December 2018 Observation window = 512ns TOA Ambiguity = 256ns Ts3 = 2048ns* Option 3 (16 pulses per 2us) Option 1 ** (16 pulses per 2us) Option 4 (16 pulses per 2us) In our simulations, a base signal in each option is considered to be 2us long. TOA ambiguity of 256 ns is assumed. In other words, the delay is uniformly distributed within the first 256ns. Observation window is equal to the symbol duration in option-1 and option-4. It corresponds to 1 quarter of the symbol duration in option-3. Therefore, we expect the performance of ternary signaling to degrade to some extent when half a symbol duration TOA ambiguity is assumed. Ts1 = Ts4 = 512ns * Since option-3 uses 31 chip sequences, 1984ns symbol duration is used for option-3 to have multiples of 4ns sampling duration. However, total energy used within 4ms duration are identical for all cases. ** A training sequence of all 1’s are used. Random training sequence will introduce self interference that will degrade the performance. Saturday, December 01, 2018

Threshold Selection Assume that µn and σn2 mean and the variance of the noise respectively Probability that a noise only sample greater than a threshold ε is Probability of threshold crossing within K consecutive noise only samples The corresponding threshold is PFA ε

Results PFA = 0.1, TB = 4ns

Results PFA = 0.05, TB = 4ns

Results PFA = 0.01, TB = 4ns

Results PFA = 0.005, TB = 4ns

Results PFA = 0.001, TB = 4ns

Results PFA = 0.1, TB = 2ns

Results PFA = 0.05, TB = 2ns

Results PFA = 0.01, TB = 2ns

Results PFA = 0.005, TB = 2ns

A-priori knowledge of noise variance improved ranging performance Conclusion A-priori knowledge of noise variance improved ranging performance Threshold is set according to the noise variance and probability of missing a block, not according to the percentage of the highest signal energy block This made option-4 suffered. Option-1 (after processing gain) performed the best both in terms of 3ns confidence level and mean absolute error (MAE). 3ns confidence level can be 90% around 15dB, at 4ns sampling interval, and around 13dB at 2ns sampling interval MAE is around 2ns at 15dB at 4ns sampling interval, and at 13dB at 2ns sampling interval

Transmitted Time-hopping Sequence ACF of the Transmitted Time-hopping Zero Correlation Zone Multipath components Peak Leading Edge Received energy samples (after processed with the time-hopping code) Search-back the leading edge Saturday, December 01, 2018

When the signal energy increases, the leading edge energy increases Peak SB When the signal energy increases, the leading edge energy increases However, energy of the sidelobes increases as well It becomes more likely to pick multipath components from the ACF’s sidelobe when searching back the leading edge Therefore, increasing the Eb/No may hurt after some point