Random Access Preamble Detection for the PRACH Helsinki
Random Access Preamble Detection for the PRACH Helsinki Finland 19 January, 1999 26 -Sep-20 1
This presentation summarizes Three Contributions • Tdoc 2 x-009 -Random Access Preamble Detection: Relative Power of Preamble and Message Block • Tdoc 2 x 99 -011 Performance in the ITU Channel Model • Tdoc 2 x 010 Random Access Preamble Detection in Doppler: Detection based on a prespecified SNR • 9/26/2020 2
Tdoc 2 x-009 -Random Access Preamble Detection: Relative Power of Preamble and Message Block • Tdoc SMG 2 UMTS-L 1 621/98 estimated that the PRACH preamble was typically detected at lower levels than those recommended for succesful message processing. ; on the order of 5. 5 d. B. • Appeared to disagree with other analyses; e. g. Tdoc SMG 2 UMTS-L 1 670/98, from Ericsson, which estimated that the power levels should be about the same value. • This document explains differences 26 -Sep-20 3
Three Sources of difference Identified • Message Data Rate • Criterion for preamble detection • Model for Frame Error Rate Versus SNR 26 -Sep-20 4
Summary of Differences in Assumptions 9/26/2020 5
Conclusions • The differences between Tdoc 621/98 and Tdoc 670/98 are explainable. • The primary issue of Tdoc 621 was the uncertainty in the power level rather than the absolute value. We believe that these observations are still valid. 9/26/2020 6
2 x 011 Random Access Preamble Detection in Doppler: Performance in ITU Channel Model • Objective- show detection performance in the classic Doppler spectrum as defined in Appendix 1 to Annex 2 of Recommendation ITU-R M. 1225. • Results continue to show the relative merit of differentially encoded signatures over the FDD baseline signatures. • This document also includes a corrected version of the notional block diagram for the differential detection process. 9/26/2020 7
Notional Block Diagram of Differential Preamble Detection 9/26/2020 8
Overview of Simulation • Simulation performed at symbol level. • For each trial, for each specified SNR, 16 complex samples generated by passing signal through the channel model. • Outputs then normalized to yield prespecified SNR for average of 16 samples, • Then add 16 complex random noise samples • Then apply each of the two detection techniques and record pass/fail for each case 9/26/2020 9
Channel Model consistent with ITU-R M. 1225 Vehicular Model Type B • Based on Nokia-furnished model used as basis for Tubo-Coding evaluations • Six of seven paths shown in appendix • Each path has 20 components • Paths 1 and 3 were used for these results • Details in document 9/26/2020 10
Nokia Data base path 1 shows breakdown of coherent detection at 120 and 150 km/hr 9/26/2020 11
Nokia Data base path 3 also shows poor performance for coherent detection at 120 and 150 km/hr 9/26/2020 12
Coherent processing of proposed signatures in ITU model • We have claimed that proposed signatures do not preclude coherent processing for low doppler • Tdoc 620/98 showed results for ideal channel • Next slide shows performance in ITU model 9/26/2020 13
Coherent Detection of proposed signatures at low doppler in ITU model is near optimum 9/26/2020 14
Conclusions • Recommendations of Tdoc 620 are reconfirmed using the ITU channel model. • For the limited number of cases exercized, u u 9/26/2020 Differential Decoding performance appeared to be robust and predictable. Coherent processing suffered large losses for one case and virtually a complete failure for the other case. 15
Tdoc 2 x 99 -010 Random Access Preamble Detection in Doppler: Detection Based on a Prespecified SNR This contribution discusses preamble detection performance under the assumption that preambles are only declared when they are detected at a high threshold, consistent, e. g. with having sufficient energy to support channel estimation, rather than simply at the minimum detectable level, consistent with low false alarm probability. 9/26/2020 16
Background • Tdoc SMG 2 UMTS-L 1 620/98 and Tdoc SMG 2 UMTS-L 1 621/98 assumed that preamble detection occurred at the lowest possible level consistent with acceptable false alarm or false detection probabilities. • In associated discussions, as well as in Tdoc SMG 2 UMTS-L 1 670/98, it was pointed out that it may be desirable and integral to the system concept for PRACH preambles to be declared only if they exceed a prespecified value; e. g. 3 d. B, or higher, to ensure sufficient energy to create a useful channel estimate. 9/26/2020 17
Therefore, this contribution compares the performance of different detection schemes on the assumption that ACK is sent by the base station to the Mobile User only if the estimated SNR is a prespecified number. 9/26/2020 18
Three approaches considered · Coherent Detection for each of the 16 baseline signatures · Differential Detection for each of the 16 signatues · Energy-Sum of 16 energy samples at the specified symbol spacing. · 9/26/2020 In this case signature determination is an auxiliary function, considered only if the energy sample summation passes its threshold. 19
Model • For each of the three detection techniques, empirically determine threshold to give 50% probability of detection at the specified level u (e. g. 3 d. B) for a fixed amplitude signal in AWGN. • For this analysis assume that the range of operation is no more than about 8 km, consistent with round trip time less than 1/16 millisecond. 9/26/2020 20
For each of the thresholds, we estimated the probability of false alarm in noise. · For Differential Detection Pfa = 10 -8 · For Coherent Detection Pfa = 5 x 10 -9 · For Envelope Summation Pfa = 2 x 10 -8 9/26/2020 21
Zero Doppler 9/26/2020 22
Coherent Detection in Doppler 9/26/2020 23
Differential Detection in Doppler 9/26/2020 24
Envelope Summation in Doppler 9/26/2020 25
Detection performance in ITU Channel Model 9/26/2020 26
Coherent Detection with Doppler in ITU Channel Model 9/26/2020 27
Differential Detection with Doppler in ITU Channel Model 9/26/2020 28
Energy Summation with Doppler in ITU Channel Model 9/26/2020 29
Conclusions • For Preamble presence, · Envelope power summation is most reliable · Differential Detection is next best · Coherent Detection is least reliable • In the actual processor it may be advantagous to · First detect presence, based on energy · Then select the best of 16 signatures 9/26/2020 30
and……. • The results of this contribution reenforce the conclusions of Tdoc 2 X 99 -011, that Differential Preamble Detection is superior to Coherent Preamble Detection. 9/26/2020 31
Recommendation • Propose that the signature set of SMG 2 UMTS-L 1 620/98 replace the signature set currently in document xx 05 section 6. 2. 3. 2 Preamble signature 9/26/2020 32
- Slides: 32