Multipath Fading FR NL UK Broadcom HPE Intel
Multipath Fading (FR, NL, & UK) Broadcom, HPE, Intel, Qualcomm
Agenda • • Executive Summary Multipath Fading Overview Multipath Fade for FR, NL, and UK Links Conclusion 2
Multipath Fade Overview 3
Multipath Fading Overview • As microwave signals travel long distances, the signal can take slightly different paths. If one path is delayed a significant portion of the period with respect to another path, there could be signal attenuation or magnification. • This phenomena is called multipath fading. G. Kizer, “Digital Microwave Communication”, 2013 4
Multipath Fading is at its Worst at Night • Multipath fading typically occurs at night 1 • NTIA asserts that Multipath fading is at its worst between midnight and 8 AM 2 1. G. Kizer, “Digital Microwave Communication”, 2013 2. NTIA Technical Report, “INTERFERENCE PROTECTION CRITERIA Phase 1 - Compilation from Existing Sources” 5
Multipath Fading is at its Worst During Summer Months, when Daytime is Longest • Multipath fading activity increases during the summer and decreases in the Winter Peak activity around August • Minimum activity around January • 1. G. Kizer, “Digital Microwave Communication”, 2013 6
The Deeper the Fade the Shorter the Duration, and Spatial Diversity is Effective in Mitigating against Deep Fades 6 GHz Measurements Made in West Unity, Ohio (41 o. N) G. Kizer, “Digital Microwave Communication”, 2013 Reference: A. Vigants, “Number and Duration of Fades at 6 and 4 GHz, ” The Bell System Technical Journal, Vol. 50, No. 3, March 1971 7
ITU-R Rec. P. 530 Model to Predict Multipath Fade • ITU-R Rec. P. 530 -17 contains a model that enables link designers to predict the probability and severity of flat multipath fade and design their link with adequate fade margin to meet their target availability* • The major driver in this model is the Geoclimatic parameter P 0, which is a function of: • d. N 1: Point refractivity gradient in the lowest 65 m of the atmosphere not exceeded for 1% of the worst month (e. g. , August) in an average year • Lowest (TX/RX) antenna height relative to sea level • Link distance • Path elevation angle * Model has been continuously revised -- most recently as of December 2017 8
The Principal ITU Fade Model (P. 530) Confirms Inverse Relationship between Fade Depth and Time for a Given P 0 Value Model demonstrates that a link with a P 0 value of 1 could achieve 99. 999% reliability with 30 d. B in fade margin 9
Multipath Fade Analysis of 534 FS Links studied in Report 302 10
P 0 Values for FS Links in Report 302 • In Report 302, FS links residing within three administrations were studied: • Using P. 530, P 0 values were assigned for all 534 FS links as shown the following distribution • Links at ~ 5%, 50%, and 95% were chosen for further review 100. 00% 90. 00% 80. 00% 70. 00% 60. 00% CDF • France: 3 links • The Netherlands: 26 links • United Kingdom: 505 links UK: P 0 = 68. 79 NL: P 0 = 2. 48 50. 00% 40. 00% 30. 00% 20. 00% FR: P 0 = 0. 0073 10. 00% 0 20 40 60 80 100 P 0 120 140 160 180 200 11
Deep Fades Occur for Very Limited Time Even During the Worst Month • Percentage of time fade depth will be exceeded in August was calculated for the 5 th, 50 th and 95 th P 0 percentiles • 5 th percentile link (P 0=0. 007) not expected to exceed 10 d. B of its fade margin 9. 9967% of the time • Median link (P 0=2. 48) not expected to exceed 20 d. B of its fade margin for 99. 9775% of the time • 95 th percentile link (P 0=68. 79) not expected to exceed 20 d. B of its fade margin 99. 4501% of the time 100. 00000% 10. 00000% Percentage of time abscissa is exceeded 95%tile % of time Sec/Mo Fade exceedance P 0=0. 007 P 0=2. 48 1. 00000% P 0=68. 79 0. 10000% > 40 d. B 0. 01000% 0. 00100% > 30 d. B 0. 00010% > 20 d. B 0. 00001% 0. 00000% > 10 d. B 0. 00000% 0 5 10 15 20 25 Fade Depth (d. B) 30 35 40 45 50%tile 5%tile % of time Sec/Mo exceedance 6. 88 E-05 184. 2 2. 48 E-06 6. 6 7. 35 E-09 0. 02 6. 88 E-04 1, 842. 3 2. 48 E-05 66. 3 7. 35 E-08 0. 2 5. 50 E-01 14, 728. 53 2. 25 E-02 601. 3 7. 85 E-05 2. 1 2. 9 E+00 77, 822. 33 2. 52 E-01 6, 744. 55 3. 27 E-03 This table assumes 31 day in a month = 2, 678, 400 seconds 12 87. 55
Case Study on 5 th Percentile, 50 th Percentile, and 95 th Percentile P 0 Links • To determine the impact from RLANs operating at a 2% activity factor, we conducted a case study of the 5 th, 50 th, and 95 th percentile P 0 value links • link is attempting to achieve 100% capacity at 128 QAM and 64 QAM (per table 17 in Report 302) and there are no gaps between FS packet transmissions (i. e. , continuous broadcast) • multipath fade magnitude and duration in accordance with P. 530 • 2% duty cycle RLAN interference (I/N) at 5, 10, 15, 20, and 25 d. B • if all forms of interference exceed the Link Margin, the link has adaptive modulation and coding (ACM), its throughput is reduced during the duration of the interference event • outage will only occur after ACM is exhausted Carrier power SNR Requirement C/N Fade Margin Link Margin Excess Link Margin Noise power 13 As Interference Increases C/N Decreases • For initial analysis, the following simple assumptions apply: Model
Projected SNR Requirement and Data Rate Ranges for FS Operating at 30 MHz Bandwidth • We obtained the projected data rate and SNR requirement for each modulation between 4 -QAM and 128 QAM from three publicly available data sheets assuming 5 d. B noise figure and 30 MHz operation Required SNR and Data Rate 4 -QAM 16 -QAM 32 -QAM 64 -QAM 128 -QAM Projected Data rate (Megabits/sec) 37 73 93 123 147 -148 SNR Required (d. B) 5. 7 -6. 2 11. 7 -12. 2 16. 2 -16. 7 15. 7 -19. 7 18. 7 -23. 2 Data rate and required SNR derived from SAF Integra DS, Redline RDL 5000 DS, and ALFOplus 2 datasheets assuming 5 d. B Noise Figure and 30 MHz operation 14
Assumed C/N based on Fade Margin Necessary to Achieve 99. 99% and 99. 999% at Target Modulation • We calculated link C/N as follows: • Each link had only the necessary fade margin to achieve desired availability for the target modulation • Target modulation is based on publicly available data • Added 1 d. B for safety margin • We note that most links we’ve studied have excess margin Percentile 5% (4. 68%) 50% (49. 06%) 95% (94. 94%) P 0 Value 0. 0073 2. 48 68. 9 Link Location France The Netherlands The United Kingdom Fade Margin for 99. 99% 8 23. 6 38. 2 • Assumed C/N for 64 QAM 28. 7 44. 3 58. 9 • Assumed C/N for 128 QAM 32. 2 47. 8 62. 4 Fade Margin for 99. 999% • Assumed C/N for 64 QAM • Assumed C/N for 128 QAM 12. 5 33. 8 48. 2 33. 2 54. 5 68. 9 36. 7 58 72. 4 15
Expected Throughput Degradation from Multipath Fade for 50%tile P 0 Link (No RLAN Interference) at 99. 999% Availability for 128 -QAM • Assume C/N = 58 d. B • The link is expected to be capable of operating at 128 QAM for all but 22 seconds during the worst month • The link has outage for 0. 4 seconds during the worst month resulting in outage availability of 99. 99998% (= 1 – 0. 4/2, 678, 400) 31 days = 744 hours = 44, 640 minutes = 2, 678, 400 seconds Fade Sec. / Worst Mo Remaining Margin (d. B) Modulation (QAM) Throughput (Mbits/sec) ≤ 10 d. B > 10 d. B, ≤ 34. 8 d. B > 34. 8 d. B, ≤ 38. 3 d. B 2, 671, 655. 4 6, 722. 59 12. 2 ≥ 48 < 48, ≥ 23. 2 < 23. 2, ≥ 19. 7 128 64 148 123 > 38. 3 d. B, ≤ 41. 3 d. B > 41. 3 d. B, ≤ 45. 8 d. B > 45. 8 d. B, ≤ 51. 8 d. B Total monthly 4. 89 3. 2 1. 31 2, 678, 399. 6 < 19. 7, ≥ 16. 7 < 16. 7, ≥ 12. 2 < 12. 2, ≥ 6. 2 32 16 QPSK 93 73 37 396, 402, 179. 31 16
Calculation to Demonstrate the Impact of RLAN Interference • • Conducted sensitivity analysis to assess impact from RLAN interference Parametrically assessed RLAN interference at I/N=5, 10, 15, 20 and 25 d. B Assume RLAN interference occurs for 2% of the time Calculated the total throughput due to • Multipath fading (98% of the time), plus • Multipath fading and interference (2% of the time) • Computed reduction in throughput by comparing results with and without interference (e. g. previous slide) • Computed resulting outage availability • An outage occurs when the lowest modulation (i. e. QPSK) doesn’t close • Total Outage = Outage in presence of Interference on (2% duty cycle is factored in) + Outage (without Interference) 17
Results for P 0 = 50 th Percentile C/N = 44. 3 d. B (64 -QAM) (using min fade margin for 64 -QAM 99. 99% availability) Without Interference Outage (sec. /worst-month)= 10. 3 Outage Availability = 99. 9996% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 38. 11 33. 89 29. 16 24. 26 19. 29 Reduction in Throughput 0. 00002% 0. 0006% 0. 0021% 0. 0130% 0. 5159% Outage (sec. /worst month) 10. 5 12. 3 16. 4 28. 5 68. 9 Availability 99. 9996 99. 9995% 99. 9994% 99. 9989% 99. 9974% C/N = 47. 8 d. B (128 -QAM) (using min fade margin for 128 -QAM 99. 99% availability) Without Interference Outage (sec. /worst-month)= 4. 6 Outage Availability = 99. 9998% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 41. 61 37. 39 32. 66 27. 76 22. 79 Reduction in Throughput 0. 0002% 0. 0005% 0. 0017% 0. 0098% 0. 3038% Outage (sec. /worst month) 4. 9 5. 5 7. 5 13. 1 30. 2 Availability 99. 9998% 99. 9997% 99. 9995% 99. 9989% 18
Results for P 0 = 50 th Percentile (Cont. ) C/N = 54. 5 d. B (64 -QAM) (using min fade margin for 64 -QAM 99. 999% availability) Without Interference Outage (sec. /worst-month)= 1. 0 Outage Availability = 99. 99996% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 48. 31 44. 09 39. 36 34. 46 29. 49 Reduction in Throughput 0. 00002% 0. 000063% 0. 00019% 0. 00057% 0. 00199% Outage (sec. /worst month) 1. 0 1. 2 1. 6 2. 9 6. 9 Availability 99. 99996% 99. 99994% 99. 99989% 99. 99974% C/N = 58 d. B (128 -QAM) (using min fade margin for 128 -QAM 99. 999% availability) Without Interference Outage (sec. /worst-month)= 0. 4 Outage Availability = 99. 99998% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 51. 81 47. 59 42. 86 37. 96 32. 99 Reduction in Throughput 0. 000016% 0. 000051% 0. 00015% 0. 00046% 0. 0016% Outage (sec. /worst month) 0. 5 0. 7 1. 3 3. 2 Availability 99. 99998% 99. 99997% 99. 99995% 99. 99988% 19
Results for P 0 = 95 th Percentile C/N = 58. 9 d. B (64 -QAM) (using min fade margin for 64 -QAM 99. 99% availability) Without Interference Outage (sec. /worst-month)= 9. 9 Outage Availability = 99. 9996% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 52. 71 48. 49 43. 76 38. 86 33. 89 Reduction in Throughput 0. 0002% 0. 0006% 0. 0019% 0. 0053% 0. 0133% Outage (sec. /worst month) 10. 01 11. 87 16. 15 29. 68 72. 47 Availability 99. 9996% 99. 9994% 99. 9989% 99. 9973% C/N = 62. 4 d. B (128 -QAM) (using min fade margin for 128 -QAM 99. 99% availability) Without Interference Outage (sec. /worst-month)= 4. 4 Outage Availability = 99. 9998% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 56. 21 51. 99 47. 26 42. 36 37. 39 Reduction in Throughput 0. 000164% 0. 0005% 0. 00158% 0. 004354% 0. 0109% Outage (sec. /worst month) 4. 7 5. 32 7. 21 13. 26 32. 37 Availability 99. 9998% 99. 9997% 99. 9995% 99. 9988% 20
Results for P 0 = 95 th Percentile (Cont. ) C/N = 68. 9 d. B (64 -QAM) (using min fade margin for 64 -QAM 99. 999% availability) Without Interference Outage (sec. /worst-month)= 1. 0 Availability = 99. 99996% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 62. 71 58. 49 53. 76 48. 86 43. 89 Reduction in Throughput 0. 00002% 0. 000065% 0. 000204% 0. 000647% 0. 001955% Outage (sec. /worst month) 1. 05 1. 19 1. 62 2. 97 7. 25 Availability 99. 99996% 99. 99994% 99. 99989% 99. 99973% C/N = 72. 4 d. B (128 -QAM) (using min fade margin for 128 -QAM 99. 999% availability) Without Interference Outage (sec. /worst-month)= 0. 4 Outage Availability = 99. 99998% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 66. 21 61. 99 57. 26 52. 36 47. 39 Reduction in Throughput 0. 0000164% 0. 000052% 0. 000164% 0. 00052% 0. 0016% Outage (sec. /worst month) 0. 47 0. 53 0. 72 1. 33 3. 24 Availability 99. 99998% 99. 99997% 99. 99995% 99. 99988% 21
Results for P 0 = 5 th Percentile C/N = 28. 7 d. B (64 -QAM) (using min fade margin for 64 -QAM 99. 99% availability) Without Interference Outage (sec. /worst-month)= 1. 1 Outage Availability = 99. 99996% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 22. 51 18. 29 13. 56 8. 66 3. 69 Reduction in Throughput 0. 0054% 0. 5069% 0. 8607% 1. 4086% 2% Outage (sec. /worst month) 1. 21 1. 73 9. 57 913. 35 53, 569. 08 Availability 99. 99995% 99. 999936% 99. 999643% 99. 965900% 97. 99996% C/N = 32. 2 d. B (128 -QAM) (using min fade margin for 128 -QAM 99. 99% availability) Without Interference Outage (sec. /worst-month)= 0. 5 Outage Availability = 99. 99998% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 26. 01 21. 79 17. 06 12. 16 7. 19 Reduction in Throughput 0. 0038% 0. 3495% 0. 8480% 1. 5733% Outage (sec. /worst month) 0. 53 0. 65 1. 62 24. 67 7, 853. 16 Availability 99. 99998% 99. 999976% 99. 999940% 99. 999079% 99. 706797% 22
Results for P 0 = 5 th Percentile (Cont. ) C/N = 33. 2 d. B (64 -QAM) (using min fade margin for 64 -QAM 99. 999% availability) Without Interference Outage (sec. /worst-month)= 0. 4 Outage Availability = 99. 999985% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 27. 01 22. 79 18. 06 13. 16 8. 19 Reduction in Throughput 0. 000088% 0. 0039% 0. 5145% 0. 9030% 1. 4179% Outage (sec. /worst month) 0. 42 0. 5 1. 09 11. 68 1, 738. 53 Availability 99. 999984% 99. 999981% 99. 999959% 99. 999564% 99. 935091% C/N = 36. 7 d. B (128 -QAM) (using min fade margin for 128 -QAM 99. 999% availability) Without Interference Outage (sec. /worst-month)= 0. 2 Outage Availability = 99. 999993% With Interference I/N = 5 d. B I/N = 10 d. B I/N = 15 d. B I/N = 20 d. B I/N = 25 d. B C/(N+I) (d. B) 30. 51 26. 29 21. 56 16. 66 11. 69 Reduction in Throughput 0. 000063% 0. 0027% 0. 3538% 1. 0144% 1. 5003% Outage (sec. /worst month) 0. 19 0. 21 0. 35 1. 56 35. 98 Availability 99. 999993% 99. 999992% 99. 999987% 99. 999942% 99. 998657% 23
Conclusion 24
Conclusion • For links with high P 0 values, multipath fade dominates • We find that links with adaptive modulation and coding that have a high P 0 value because they are susceptible to higher levels of multipath fade are robust against even very high levels of interference because they have a high C/N • For links with low P 0 values, high levels of RLAN interference dominates • We find that links with adaptive modulation and coding that have a low P 0 value because they are less susceptible to higher levels of multipath fade are less robust against high levels of RLAN interference because they have a lower C/N • However, links with low P 0 usually have larger link margins due to their short distances • For links with higher order modulation capability, impact on throughput from even very high levels of RLAN interference is much less than links that are only capable of lower order modulation 25
Thank You 26
- Slides: 26