Date: Meeting: 8th May 2002 PPWG Document Filename: Paper Number: RADIOCOMMUNICATIONS AGENCY Project title System Signature Prediction Model Author: M RELMY Date: April 2002 533544781 PPWG (02-02)/024 System Signature Prediction Model CONTENTS Acknowledgements…………………………………………………………...1 ABSTRACT…………………………………………………….…………….2 1. INTRODUCTION……………………………………………………………3 1.1 Fresnel Zone……………………………………………………………..3 1.2 Dispersive fading………………………………………………………..3 1.3 Digital radio signature………………………………………………….4 1.3.1 Despersive fade model…………………………………………………………....4 1.3.2 Signature curves…………………………………………………………………..5 1.3.3 Selective outage probability………………………………………………………6 2. Methodology…………………………………………………………………..6 3. Results…………………….…………………………………………………...7 3.1 Selective outage probability improvement using adaptive equaliser…8 3.2 Selective outage probability for different modulation schemes………8 3.3 Selective outage probability for 10-3 and 10-6 BER…………………...10 3.4 Selective outage probability for systems with adaptive equalisation..11 3.4.1 Selective outage probability for the 4 GHz band………………………………………11 3.4.2 Selective outage probability for the 6 GHz lower band……………………………….12 3.4.3 Selective outage probability for the 6 GHz upper band ………………………………12 3.4.4 Selective outage probability for the 7.5 GHz band…………………………………….14 3.4.5 Selective outage probability for the 13 GHz band……………………………………..15 3.5 Selective outage probability for the 38 and 52 GHz bands…………...15 4. Conclusion and recommendation…………………………………………..17 4.1 Conclusion……………………………………………………………...17 4.2 Outage probability investigation……………………………………...17 4.3 Recommendation………………………………………………………19 5. Appendix…………………………………………………………………….21 6. Bibliography ………………………………………………………………...24 M RELMY i System Signature Prediction Model Acknowledgements The author would like to thank first the manufactures for providing theirs radio relay systems signature parameters, which were necessary, for this project. The author was very grateful for the help and the support during this project from the following persons: Dr David Bacon Mr Steve Lynch Mr Nicholas Woollard Mr Ian Flood Mr James Richardson Mr Duncan Gallon Mr Stan Murphy M RELMY 1 System Signature Prediction Model ABSTRACT A major concern for microwave system users is how often and how long a system might be out of service. Various statistical models and analysis methods have been developed in order to predict and measure the outage and availability over a period of time. This report deals with the calculation of the selective outage probability using radio relay systems signatures. The system signature is the measurement of the tolerance of a digital radio system to multipath fades and was raised for consideration during the review of the link length policy within fixed links consultative committee (FLCC) and Point to Point Working Group formerly RSSPWG. The selective outage probability is defined as the probability that the Bit Error Ratio (BER) is larger than a given threshold which is based on the system signatures. The methodology used in the calculation is based on the ITU-R 530-9 Recommendation. The calculations were performed on all the UK fixed terrestrial (point-to-point) links in the 4, lower 6, upper 6, 7.5 and 13 GHz bands. The selective outage probability was performed then plotted against path length using developed software for all the UK fixed links (FL) in the bands under study. The objective of the study was achieved and it was clear from the results that longer links have a selective outage probability greater than the acceptable value (10-4). Also in this report it was concluded that to improve the selective outage probability for longer links it is necessary to 1) increase the height of both antennas, and/or 2) increase the inclination angle Option 2 gives greatest benefit for the reduction of selective outage probability. M RELMY 2 System Signature Prediction Model 1. Introduction 1.1 Fresnel Zone In microwave line of sight systems it is necessary to provide clearance of around 0.6 Fresnel zone to ensure that attenuation due to obstacles, in or near the ray beam path, is negligible. The first Fresnel zone is the zone within which the reflected wave has an indirect path of a half wavelength longer than the direct path. Fresnel zone Direct path Indirect path rn Receiver Transmitter d1 d2 Figure 1: Fresnel zone The radius of nth Fresnel zone (rn) is derived as follow: (d1 + rn)1/2 + (d2 + rn)1/2 = (d1 +d2) +n(/2) For rn << d1, d2 we can use (1 + x )n 1 + nx for x very small Then 1 n d1 d2 rn = d1 d2 2 1.2 Dispersive fading Multi-path propagation causes dispersive fading when microwave signals reach a receiver having travelled over more than a single path with significant relative delays and comparable signal amplitudes. This may happen due to the presence of reflective layers in the atmosphere or reflective areas on the ground. Signals arriving via the reflective layers or areas are delayed relative to the direct line-of-sight signal. All signals combine at the receiver antenna can cause amplitude and phase distortion of M RELMY 3 System Signature Prediction Model the direct signal. Normally this distortion is minimal and easily handled by the receiver because the direct signal is much stronger than all the reflected signals. However, under certain conditions, the direct signal may be attenuated and the distortion increased to the point where frequency selective notches result and dispersive fading is said to be present. 1.3 Digital radio signatures 1.3.1 Dispersive fading model W.D. Rummler of Bell laboratories has developed a 3-path model of multi-path propagation, see figure 2. The transfer function for this model is given by the following expression: H() = a[1 – b.e j( - 0)] Where: a is the attenuation due to flat fading (randomly varying flat fade across the channel) = 2f, centre frequency of the notch in radians 0 = 2f0 , centre frequency of the RF channel in radians b is the ratio of the amplitude of the main signal (direct) to the delayed (indirect) signal, b (0 b 1) determines the notch depth. = 6.3ns [1] which the standard delay time between the two signals The signs in the exponent correspond to non-minimum phase and minimum phase fades respectively, which are explained in the next section. Direct path 1 2 Tx 3 Indirect path Rx Figure 2: Diagram of the 3-path model showing the direct (1) and indirect (2&3) paths [1] Rummler has shown that 6.3 ns is approximately the delay time measured on real microwave links in the USA . M RELMY 4 System Signature Prediction Model 1.3.2 Signature curves It is easier to understand and measure the performance of a microwave radio under multi-path receptions when considering just one direct signal and one indirect signal having a different amplitude and relative time delay (as shown in figure 3). When the amplitude of the direct signal is higher than that of the indirect signal, the notch is called a minimum-phase notch. Conversely, when the delayed signal has higher amplitude than the direct signal, the notch is a non-minimum phase notch. Because a receiver can respond differently to these types of notches, it is important to test the radio under both minimum and non-minimum phase conditions. In general, nonminimum is more severe than minimum phase dispersive fading, but under most conditions, typically the direct signal is stronger (minimum phase notch). Spectrum Analyzer Variable attenuation Tx 3dB coupler 3dB coupler MOD Rx DEM Fixed delay Variable attenuation Pattern Generator (2n-1) Phase shifter Error Detector (2n-1) Figure 3: Block diagram of a test bench for the experimental measurement of signature The measurement is performed by setting the centre frequency of the notch somewhere in the RF channel of the particular radio system under test. The depth is adjusted until BER degrades to 10-3 or 10-6. The phase shifter is then adjusted so that the notch appears at a different frequency (in the RF channel) and again the notch depth is adjusted to a reference BER (T). This process is repeated at a number of frequencies resulting in a signature curve by plotting of non-minimum and minimum phase notch versus notch depth. The signature curve shows the depth of the notch required, at a particular frequency relative to the centre of the RF channel to cause the BER to degrade to T. The signature of a radio relay system (Figure 4) for a given BER can be idealised by a rectangular block. The system signature parameters are: Wx: the width of the system signature in MHz. Ndx: The signature depth in dB. whith x denoting either minimum phase (m) or non-minimum phase (nm) fades. : Reference delay used to obtained the signature. M RELMY 5 System Signature Prediction Model The system will suffer outage if the multipath notch intrudes in to the signature. Freq in MHz Ndx Selective fade attenuation in dB (Notch) Wx Idealised system signature Figure 4: The selective fading attenuation characteristic and the idealised system signature 1.3.3 Selective outage probability A major concern for microwave system users is how often and how long a system might be out of service. Various statistical models and analysis methods have been developed in order to predict and measure the outage and availability over a period of time. This report is about the calculation of the outage probability / outage time based on ITU-R P.530-9 for radio relay systems in the 4, lower 6, upper 6, 7.5 and 13 GHz FS bands. The selective outage probability is defined as the probability that the BER is larger than a given threshold (T) (Note: This probability does not translate directly to the unavailable time). The calculation of selective outage probability is based on system signatures. The next section will describe the steps that were followed to calculate the selective outage probability based on ITU-R P.530-9 Recommendation. 2. Methodology ITU-R P.530-9 Recommendation was used to calculate the selective outage probability that the system performs more poorly than a specified BER threshold (T). In this study the threshold for the BER is 10-6 (T = 10-6). The outage probability is calculated using the following steps and is in appendix A of this report. Step 1: Calculation of the mean time delay from: M RELMY 6 System Signature Prediction Model d m 0.7 50 1.3 Where d is the path length (km) Step 2: Calculation of the multipath activity parameter from: 0.75 1 e 0.2 po Where po = pw/100 is the multipath accurrence factor corresponding to the percentage of time pw (%) of exceeding A = 0 dB in the average worst month. pw K ( d) 3.0 1 p 1.2 10 0.033 f 0.001 hl Where f is the frequency (GHz), hl is the altitude of the lowest antenna K is the geoclimatic factor (in UK a typical value is 8.241 10-5) and p = (height of the transmitter – height of the receiver) / d : is the inclination angle in mrad (1 mrad 0.03) Step 3: Calculation of the selective outage probability from: Ndm Ndnm 2 3 20 3 20 m Ps 2.15 Wm 10 10 Wnm 10 10 Where: Wx: signature width (GHz) Ndx: signature depth (dB) with x denoting either minimum phase (m) or non-minimum phase (nm) fades. : the reference delay (ns) used to obtained the signature 3. Results Mathcad tool was used to calculate and plot the selective outage probability for all the FS links in UK in the following bands: 4 GHz band Lower 6 GHz band Upper 6 GHz band 7.5 GHz band 13 GHz band Results are shown in the following section. The algorithm for the calculation for the selective outage probability for 4 GHz band is in the appendix B. For other bands the procedure remains the same but with changed parameters. In the calculation of the selective outage probability the fade margin depth is set for M RELMY 7 System Signature Prediction Model 0 dB. The signature parameters of the radio relay system with or without adaptive equaliser used in the calculation are obtained from manufacture’s data. Manufactures are not identified for the sake of confidentiality. 3.1 Selective outage probability improvement using adaptive equaliser Selective outage probabilities were calculated for all UK FL in the band under study using system signature with and without adaptive equaliser. The calculation performed shows that selective outage probability for system without adaptive equaliser is improved by approximately 94.3 % using adaptive equaliser as illustrated in figure 5. Figure 5: Selective outage probability for 4 GHz band Blue crosses for radio relay system without adaptive equaliser Red crosses for radio relay system with adaptive equaliser M RELMY 8 System Signature Prediction Model 3.2 Selective outage probability for different modulation schemes Calculations of selective outage probability were performed for all UK FL in the band under study for systems with 64 and 128 TCM. The results as in figures 6 and 7 (7.5 GHz band) show that the selective outage probability is the same using different modulation schemes. Figure 6: Selective outage probability against path length for a 7.5 GHz band Red crosses for 10-6 BER for a 128 TCM system Blue crosses for 10-6 BER for a 64 TCM system Figure 7: Selective outage probability against path length for a 7.5 GHz band Red crosses for 10-3 BER for a 128 TCM system Blue crosses for 10-3 BER for a 64 TCM system M RELMY 9 System Signature Prediction Model 3.3 Selective outage probability for 10-3 and 10-6 BER The calculations of the selective outage probability were performed on the UK FL bands under study using system signatures for a 10-3 and 10-6 BER for different modulation schemes. The results as illustrated in figures 8 and 9 shows that the selective outage probability for a 10-3 BER is improved by approximately 40 % than a 10-6 BER. Figure 8: Selective Outage probability for 7.5 GHz band using 64 TCM system Red crosses for a 10-3 BER Blue crosses for a 10-6 BER M RELMY 10 System Signature Prediction Model Figure 9: Selective Outage probability for 7.5 GHz band using 128 TCM system Red crosses for a 10-3 BER Blue crosses for a 10-6 BER 3.4 Selective outage probability for systems with adaptive equalisation 3.4.1 Selective outage probability for the 4 GHz band Figure 10: Selective outage probability that BER > 10-6 for all UK FL in the 4 GHz band M RELMY 11 System Signature Prediction Model 3.4.2 Selective outage probability for the 6 GHz lower band Figure 11: Selective outage probability that BER > 10-6 for all UK FL in the lower 6 GHz band 3.4.3 Selective outage probability for the 6 GHz upper band Figure 12: Selective outage probability that BER > 10-6 for all UK FL in the upper 6 GHz band M RELMY 12 System Signature Prediction Model Figure 13: Selective outage probability that BER > 10-6 for all UK FL in the upper 6 GHz band (16x2 Mb/s system signature parameters) 3.4.4 M RELMY Selective outage probability for the 7.5 GHz upper band 13 System Signature Prediction Model Figure 15: Selective outage probability that BER > 10-6 for all UK FL in the 7.5 GHz band Figure 16: Selective outage probability that BER > 10-6 for all UK FL in the 7.5 GHz band 3.4.5 Selective outage probability for the 13 GHz band M RELMY 14 System Signature Prediction Model Figure 17: Selective outage probability that BER > 10-6 for all UK FL in the 13 GHz band 3.5 Selective outage probability for the 38 and 52 GHz bands The next figures show the selective outage probability function (E) against the height of antenna (transmitter) for 38 and 52 GHz bands. The parameters from left to right of the E function are the maximum link path length in km obtained from Second MPL Report v5, the height of transmitter ASL in metres and the height of the receiver ASL in meters. The signature parameters of the radio relay systems in the frequency range of 4-14 GHz were used to estimate the selective outage probability for the 38 and 52 GHz bands. Figures 18 and 19 shows that the selective outage probability is below 10-10. The peak represents the worst case (inclination angle is 0 i.e. antennas height are equal). There is a small sensitivity to increasing the height of the receiver. M RELMY 15 System Signature Prediction Model Figure 18: Selective outage probability against height of transmitter ASL for 38 GHz band Figure 19: Selective outage probability against height of transmitter ASL for a 52 GHz band M RELMY 16 System Signature Prediction Model 4 Conclusion and recommendation: 4.1 Conclusion From the results, most of the selective outage probabilities for UK fixed links in the bands under study are below 10-4 which maybe, acceptable. However there are two links in the lower 6 GHz and the 7.5 GHz bands where the selective outage probability is greater than the acceptable value (10-4). Also the results shows that the selective outage probability for the fixed links is: 1) Nearly the same for signatures obtained from different manufactures. 2) The same for different modulation schemes (64 and 128 TCM). 3) Improved by 94.3 % when adaptive equaliser is used. 4) Difference of about 40 % between 10-3 and 10-6 BER. The improvement in the selective outage probability is due to the increase in the notch depth of the signature which reduces the possibility of the multipath notch intruding in to the signature. This will reduce outage in the system. In summary the results show that the selective outage probability is higher for longer path links. 4.2 Outage probability investigation The parameters from left to right of the Ps function in figures 20 and 21 are path length, centre frequency, the height of the lowest antenna, the inclination angle and the point refractivity gradient. 3 Selective outage probability 4.55210 Ps di 3.9 0 0 40 Outage probability for BER > 10^-6 0.006 0.0054 0.0048 Ps di 6.175 0 0 40 0.0042 0.0036 Ps di 6.775 0 0 40 0.003 Ps di 7.65 0 0 40 0.0024 Ps di 13 0 0 40 0.0018 0.0012 106 10 3.72710 4 0 0 0 15 30 45 60 75 90 di Path length in km 105 120 135 150 150 Figure 20: Outage probability for the worst case ( hl = 0 and = 0) for the 5 bands under study against path length M RELMY 17 Se le ct iv e o uta g e p ro b ab ili ty System Signature Prediction Model 7.107 10 Outage probability for BER > 10^-6 4 8 10 4 4 Ps d i 3.9 10 10 40 7.2 10 4 6.4 10 Ps d i 6.175 10 10 40 5.6 10 4 4.8 10 Ps d i 6.775 10 10 40 4 4 10 Ps d i 7.65 10 10 40 4 3.2 10 4 Ps d i 13 10 10 40 4.234 10 4 2.4 10 4 1.6 10 5 11 8 10 0 0 15 30 45 0 60 75 90 105 120 135 di Path length in km 150 150 Figure 21: Outage probability (hl = 10 and = 10) for the 5 bands under study against path length Figure 21 and 22 shows grate sensitivity to increasing the inclination. Also increasing the height of the lowest antenna can reduce the outage probability. Figure 22 shows the variation of the outage probability set at 60 km path length, the lowest antenna height set at 100 m for varying inclination angle against the frequency. Ps Ps Ps O u t a g e p ro b a b i l i t y Ps Ps Ps Ps Ps Ps Ps Ps Ps Ps Ps Ps 2 10 5 1.9 10 5 1.8 10 5 1.7 10 5 1.6 10 5 1.5 10 5 1.4 10 5 1.3 10 5 1.2 10 1.1 10 5 1 10 5 9 10 6 8 10 6 7 10 6 6 5 1.941 10 60 60 60 60 60 60 60 60 60 60 60 60 f i 10 0 5 40 60 60 60 f i 10 0 13 0 40 6 10 5 10 f i 10 0 14 0 4 10 6 3 10 6 2 10 6 1 10 6 40 40 40 40 40 40 40 40 f i 10 0 10 40 f i 10 0 20 f i 10 0 30 f i 10 0 40 f i 10 0 50 f i 10 0 60 f i 10 0 70 f i 10 0 80 f i 10 0 90 40 f i 10 0 10 0 40 f i 10 0 12 0 f i 10 0 15 0 4.843 10 40 40 7 Outage probability for BER > 10^-6 5 6 0 0 0 1.6 3.2 4.8 6.4 8 fi Frequ ency in GHz 9.6 11 .2 12 .8 14 .4 16 16 Figure 22: Outage probability for a 60 km path length, 100 m height for the lowest antenna and varying inclination angles against the frequency Figure 23 shows the outage probability against the path length (red crosses) and the inclination angle in mrad (blue crosses) for the UK fixed links in the 4 GHz band. M RELMY 18 System Signature Prediction Model Figure 23: Selective outage probability against path length in km (Red crosses) and inclination angle in mrad (Blue crosses) for the UK fixed links in the 4 GHz band 4.3 Recommendation Based on the investigation in the above section, the selective outage probability of the links can be improved by increasing the inclination angle or increasing the altitude of the two antennas keeping the same inclination angle. This is shown in the graphs (24 and 25) for the longest fixed link in the lower 6 GHz band. The other recommendation is the limitation on the path length. The solution is dependent on the location of the fixed link and if the recommendations are feasible. M RELMY 19 System Signature Prediction Model Figure 24: Selective outage probability for a BER > 10-6 for a 140.3 km link in the 7.5 GHz band (company B signature) against the inclination angle Figure 25: Selective outage probability for a BER > 10-6 for a 140.3 km link in the 7.5 GHz band (company B signature) against the height of the lowest antenna M RELMY 20 System Signature Prediction Model 5. Appendix: 5.A Methodology used to calculate the selective outage probability 1. Centre frequency f1: the lower frequency in GHz f2: the upper frequency in GHz f: the center frequency in GHz f1 f2 2 f 2. System Signature parameters obtained from manufactures Nd: Notch depth in dB Ndnm: notch depth for non -minimum phase fade Ndm: notch depth for minimum phase fade W: Signature width in MHz Wnm: signature width for non -minimum phase fade Wm: signature width for minimum phase fade Reference delay used to obtain the signature in ns : 3. Point to Point Links data ht is the transmitter hight above sea level in m hr is the receiver hight above sea level in m d is the path length in km h is the path heigth antennas difference hl is the altitude of the lower antenna in m hl min ( ht hr) 4. Calculation of the selective outage probability 4.1 The mean time delay in ns d m 0.7 50 1.3 4.2 Point refractivity gradient in the lowest 65 m of the atmosphere not exceeded for 1% of an average year dN1 40 This is a typical value for UK 4.3 The geoclimatic factor for the average worst month K 10 4.20.0029 dN1 M RELMY 21 System Signature Prediction Model 4.4 Magnitude of the path inclination in mrad p h d 4.5 The percentage of time pw pw K ( d) 3.0 1 p 1.2 10 0.033 f 0.001 hl 4.6 The multipath occurrence factor po pw 100 4.7 Multipath activity parameter 1 e 0.75 0.2 po 4.8 The selective outage probability Ndm Ndnm 2 m Ps 2.15 Wm 10 3 10 20 Wnm 10 3 10 20 4.9 The outage time in s / year Ot = Ps x second in a year ( s / year ) Ot = Ps x 31.536 x 10^6 M RELMY 22 System Signature Prediction Model 5.B Calculation and plotting the selective outage probability against path length for the FS UK links in the 4 GHz band f 3.9 5 6.3 K 8.24 10 E( d ht hr) h ht hr Wm 28 Wnm 28 Ndnm 22 Ndm 27 hl min( ht hr) d 50 1.3 m 0.7 p h d 3.0 pw K d po 1 p 1.2 0.033f 0.001hl 10 pw 100 0.75 0.2 po 1e Ndm Ndnm 2 3 20 3 20 m Ps 2.15 Wm10 10 Wnm10 10 i 0 rows( A) 1 Psi E Ai 0 Ai 1 Ai 2 1 10 5 1 10 6 1 10 7 1 10 8 1 10 9 1 10 10 1 10 11 1 10 Psi 1 10 12 13 1 10 14 1 10 15 1 10 16 1 10 17 1 10 18 1 10 19 1 10 20 1 10 Selective outage probability 4 0 10 20 30 40 50 60 70 Ai 0 Path length in km M RELMY 23 System Signature Prediction Model 6. Bibliography ITU- R P.530-9 ITU-R F.1093-1 Western Multiplex: Dispersive fade margin (WesternMultiplex@http://wmux.com) IEEE 802.16sc-99/13: Dispersive fade margin Telettra: Introduction to high capacity digital transmission systems Radio System Design For Telecommunications (ROGER L. FREEMAN) Principles of radio communication ( Fraidoom Mazda) M RELMY 24