Document Sample

RAIN MODELS FOR THE PREDICTION OF FADE DURATION AT MILLIMETER WAVELENGTHS MOHD AFZAN BIN OTHMAN UNIVERSITI TEKNOLOGI MALAYSIA 2 UNIVERSITI TEKNOLOGI MALAYSIA BORANG PENGESAHAN STATUS TESIS♦ JUDUL : RAIN MODELS FOR PREDICTION OF FADE DURATION AT MILLIMETER WAVELENGTHS SESI PENGAJIAN: 2006/2007 SAYA MOHD AFZAN BIN OTHMAN (HURUF BESAR) mengaku membenarkan tesis (PSM/Sarjana/Doktor Falsafah)* ini disimpan di Perpustakaan Universiti Teknologi Malaysia dengan syarat-syarat kegunaan seperti berikut: 1. Tesis adalah hak milik Universiti Teknologi Malaysia. 2. Perpustakaan Universiti Teknologi Malaysia dibenarkan membuat salinan untuk tujuan pengajian sahaja. 3. Perpustakaan dibenarkan membuat salinan tesis ini sebagai bahan pertukaran antara institusi pengajian tinggi. 4. **Sila tandakan ( √ ) SULIT (Mengandungi maklumat yang berdarjah keselamatan atau kepentingan Malaysia seperti yang termaktub di dalam AKTA RAHSIA RASMI 1972) TERHAD (Mengandungi maklumat TERHAD yang telah ditentukan oleh organisasi/badan di mana penyelidikkan dijalankan) TIDAK TERHAD Disahkan oleh ___________________________________ ___________________________________ (TANDATANGAN PENULIS) (TANDATANGAN PENYELIA) Alamat Tetap: NO 270 TAMAN PERTAMA, ASSOC PROF DR JAFRI DIN . 272000 KUALA LIPIS, (Nama Penyelia) PAHANG DARUL MAKMUR. Tarikh : 23 NOVEMBER 2006 Tarikh : 23 NOVEMBER 2006 “I hereby declare that I have read this thesis and in my opinion this thesis is sufficient in terms of scope and quality for the award of the degree of Master of Engineering (Electrical-Electronics & Telecommunication)” Signature :………………………….. Name of Supervisor : Assoc. Prof. Dr Jafri Din Date : 23rd November 2006 RAIN MODELS FOR THE PREDICTION OF FADE DURATION AT MILLIMETER WAVELENGTHS MOHD AFZAN BIN OTHMAN A project report submitted in partial fulfilment of the requirements for the award of the degree of Master of Engineering (Electrical-Electronics & Telecommunication) Faculty of Electrical Engineering Universiti Teknologi Malaysia NOVEMBER 2006 ii I declare that this thesis entitled “Rain Models for the Prediction of Fade Duration at Millimeter Wavelength” is the result of my own research except cited in the references. The thesis has not been accepted for any degree and is not concurrently submitted in candidature of any other degree. Signature :…………………………………. Name : MOHD AFZAN BIN OTHMAN Date : 23rd November 2006 iii “To my beloved mother, father, fiancé and family…” Thank you. iv ACKNOWLEDGEMENTS Alhamdulillah. First of all, I would like to express my gratitude to Allah S.W.T because of his blessing; I have managed to complete my master project successfully within the given time without any difficulties. Here, I wish to take this opportunity to express my appreciation and gratitude of myself to my helpful and knowledgeable supervisor, Assoc. Prof. Dr Jafri Din from Wireless Communication Center, UTM, whom I deeply indebted, in the process of finishing this project. Without his brilliant suggestions and encouragement, I would never be able to write this thesis completely as I am right now. Thank you so much for your commitment and your very active participation in helping me doing this project, and again, I am very proud to have you as a supervisor. Thank you so much. Also, a million thanks to all of the lecturers in Faculty of electrical Engineering who involves direct or indirectly in this project. I am really appreciate of your great assistances and would like to truly thank you for your magnificent teaching, and also for the wonderful and worthwhile experience I had during my course of study in UTM. Last but not the least, I would like to say thank you to all of my friends for their helps, support, and valuable hints. Finally, my heartfelt thanks and appreciation goes to my beloved family, especially my Mum and Dad; and most important of all, to my fiancée. I am really grateful to have them in my life. Thanks for all of your helps and supports when I am trapped in a difficult time. Thanks for everything you have done for me. To all, I wish you success in your future endeavors. Thank you v ABSTRACT The planning of radio communications system requires an estimate of the average annual outage due to fading, which at millimeter wavelengths, is generally dominated by the effects of rain attenuation. Current ITU-R recommendations provide algorithm for estimating the exceedance static of rain-induced attenuation on terrestrial links. Another factor of interest, not currently covered by ITU-R recommendations is the distribution of the durations of rain fades. Hence this project involved an extensive review on the available models describing annual rain fade distributions on line-of-sight links. Also, analysis on the rain attenuation data conducted on an experimental 350m, and 38GHz in frequency in UTM campus for 1 year will be used to obtain information on the rain fade statistic. Previous researches pointed out that the distribution of the durations of rain events at different intensities is more fundamental than distributions of link fades. Thus, this project aim is to establish an expression for the average number of events per year of rain intensity greater than a given threshold. This could be achieved from the detail analysis of the 38GHz link signal level profile database. Thus, the rain fade statistic describing annual rain fade distributions on line-of-sight could be developed. To predict rain attenuation on complex multi hop or route diverse links, it is necessary to have a statistical representative of rainfall in time and space, which is accurate over wide ranges of spatial and temporal scales. Various available statistical models will be used to demonstrate the influence of the developed rain fade statistic when applied for more complex microwave links. vi ABSTRAK Pada masa kini, perkembangan dalam bidang telekomunikasi adalah amat memberangsangkan. Untuk menghasilkan system telekomunikasi radio, perkara penting yang amat dititikberatkan ialah kesan isyarat radio terhadap hujan. Di mana seperti yang kita sedia maklum, bagi isyarat yang lebih daripada 10 GHz, kesan hujan terhadap isyarat ini adalah sangat tinggi. Dengan itu ITU-R telah memperkenalkan algorithm bagi menganggarkan kesan hujan ini. Tetapi ada satu lagi faktor yang tidak kelaskan oleh ITU-R, iaitu masa semasa isyarat ini menjadi lemah disebabkan oleh kesan hujan. Oleh itu, projek ini akan merangkumi analisis terhadap data yang diamabil di UTM kampus bagi tempoh 1 tahun iaitu pada 1999, 350m, 38GHz serta mengenal pasti model kesan hujan terhadap isyarat yang telah dicadangkan oleh beberapa penyelidik. Model hujan yang sesuai dengan data yang sedia ada akan diambil sebagai rujukan dan pemalarnya akan diubah mengikut kesesuaian data hujan di Malaysia. Tujuan projek ini dilaksanakan adalah untuk mencari nilai purata ‘event’ ini berlaku bagi tempoh setahun. Ini membolehkan kita membuat ramalan berapa lama isyarat akan menjadi lemah bergantung kepada jumlah hujan yang turun. Maka dengan itu sebuah model yang baru akan dihasilkan untuk meramalkan kesan hujan terhadap isyarat radio ini. vii TABLE OF CONTENTS CHAPTER TITLE PAGE DECLARATION ii DEDICATION iii ACKNOWLEDGEMENTS iv ABSTRACT v ABSTRAK vi TABLE OF CONTENTS vii LIST OF TABLES x LIST OF FIGURES xi LIST OF ABBREVIATIONS xiii LIST OF SYMBOLS xiv LIST OF APPENDICES xv CHAPTER I INTRODUCTIONS 1 1.1 Project Background 1 1.2 Problem Statement 3 1.3 Objective 4 1.4 Scopes of Project 4 1.5 Importance of the Project 5 viii CHAPTER II LITERATURE REVIEW 7 2.1 Rain Attenuation on Satellite Communication 7 2.2 Fade Mitigation Techniques 9 2.2.1 Power Control 10 2.2.2 Signal Processing 11 2.2.3 Diversity 11 2.3 Available Prediction Models 12 2.3.1 ITU-R Model 12 2.3.2 Chris J. Gibbins/ Kevin S. Paulson 17 2.3.3 Kormayos/ Lena Pedersen/ Cyril Sagot/ 19 Janos Bito 2.4 Summary 21 CHAPTER III METHODOLOGY 22 3.1 Methodologies 22 3.2 Data Collection 24 3.3 Data Extractions 26 3.4 Existing Rain Models Programming 29 3.4.1 ITU-R Model Programming 30 3.4.2 Chris J. Gibbins/ Kevin S. Paulson 32 Programming 4.4.3 Kormayos/ Lena Pedersen/ Cyril Sagot/ 33 Janos Bito Programming 3.5 Comparison 33 CHAPTER IV RESULTS 35 4.1 Introduction 35 4.2 Measured Data Interpretations 35 4.2.1 Power Received 35 ix 4.2.2 Measured Data Analysis 37 4.3 Result and Analysis 41 4.3.1 Obtain Rain Fade Statistics 41 4.3.2 Comparison 45 CHAPTER V CONCLUSIONS 49 5.1 Conclusions 49 5.2 Future Works 50 REFERENCES 51 x LIST OF TABLES TABLE NO TITLE PAGE 4.1 Convert AGC (V) to decibel (dBm) 38 4.2 Analysis on fade duration 40 4.3 Number of events which duration exceeds abscissa 42 at 1, 3, 6, 9, 12, 15, 18 and 21 dB levels 4.4 Number of events for each month over one year 44 4.5 Fade duration statistics information 46 xi LIST OF FIGURES FIGURE NO TITLE PAGE 1.1 Features characterizing the dynamic of fade 2 Events[1] 3.1 Research methodology flow chart 23 3.2 Path profile for Ericson Microwave Link at 25 38GHz in WCC, UTM 3.3 Rain rate contour map in Peninsular Malaysia 26 3.4 Read data form text file 27 3.5 Converts AGC (volts) values to dBm 27 3.6 Compile fade duration statistic for attenuation 28 exceeding threshold 3.7 Save fade duration statistics data to corresponding 29 files 3.8 Set parameters with correct values 30 3.9 Matlab programming for corresponding equations 31 3.10 Matlab programming for total numbers of fade 31 durations 3.11 Matlab programming for number of fades NA 32 3.12 Matlab programming for probability numbers of fades 33 durations xii 3.13 Matlab programming for comparison among measure 34 data and studied rain fade models 4.1 Received power level time series on clear weather 36 4.2 Received power level time series on rainy weather 37 4.3 Power received signal attenuation 39 4.4 Analysis on fade duration 40 4.5 Fade duration grouping by attenuation levels 43 4.6 Fade duration grouping by months 45 4.7 Comparison among existed rain models with 47 measured data xiii LIST OF ABBREVIATIONS DTH - Direct To Home FMT - Fade Mitigation Technique MEASAT - Malaysia East Asia Satellite WCC - Wireless Communication Centre AGC - Automatic Gain Control UTM - Universiti Teknologi Malaysia CD - Cumulative Distribution ITU-R - International Telecommunications Union - Radiocommunications sec - seconds xiv LIST OF SYMBOLS f - frequency φ - elevation angle A - attenuation σ - standard deviation γ - power law distribution Dt - boundary time k - fraction time Q - standard cumulative distribution function N - number of fade events R - rain rate r - path reduction factor L - path length V - AGC value (volt) R0.01 - rain intensity at 0.01% of time dBm - decibel meter xv LIST OF APPENDICES APPENDIX TITLE PAGE A FLOW CHART 53 B PROGRAMMING CODES 57 CHAPTER I INTRODUCTIONS 1.1 Project Background The planning of radio annual outage due to fading at millimeter wavelengths is generally dominated by the effects of rain attenuation. This is usually determined from long-term statistic of annual rainfall rates, applying the procedures in ITU-R Recommendation P.530 for terrestrial links. In the design of a variety of telecommunication systems, the dynamic characteristics of fading due to atmospheric propagation are of concern to optimize system capacity and meet quality and reliability criteria. Examples are fixed networks that include a space segment and systems that apply fade mitigation or resource sharing techniques. 2 Several temporal scales can be defined, and it is useful to have information on fade slope, fade duration and interfade duration statistics for a given attenuation level (Figure 1.1). Figure 1.1: Features characterizing the dynamic of fade events [1] Fade duration is defined as the time interval between two crossings above the same attenuation threshold whereas interfade duration is defined as the time interval between two crossings below the same attenuation threshold. Fade slope is defined as the rate of change of attenuation with time. Of particular interest in the context of availability criteria is the distinction between fades of shorter and longer duration than 10 s. Knowledge of the distribution of fade duration as a function of fade depth is also a prerequisite for the application of risk concepts in the provision of telecommunication services. In addition, information about the expected fade slope is essential to assess the required minimum tracking rate of a fade mitigation system. 3 1.2 Problem Statement Nowadays, the advancement in microwave communication technologies especially in telecommunication and broadcasting has resulted in congestion for frequencies below 10GHz. This has forced microwave designers to look for higher frequencies. Unfortunately for the frequencies that greater than 10GHz, rain become the main factor of attenuation especially for tropical and equatorial countries that experience high rainfall rate throughout the year such as in Malaysia. Where is for the frequencies above 10GHz, it will lead to outages that compromise the availability and quality of service, making this one of the most critical factors in satellite link design. Thus, in some cases the use of suitable compensation techniques to counter excessive attenuation will be needed to maintain reliable system operation. The proper design of fade mitigation techniques on satellite links requires not only knowledge of long-term statistics, but also of second-order statistics describing the dynamic behavior of attenuation, such as duration of fades, duration between fades and fade rate. So due to this circumstance it is important to have a predication of fade duration. Usually, fade duration statistics are presented as conditional distributions of the number of fades exceeding certain durations, given that specified fade threshold has been exceeded. This representation provides information on the number of outages and system availability due to propagation on a link, given a fade margin and an availability specification. 4 1.3 Objective The main objective of this project is to estimate of the average number of events per year of rain attenuation greater than a given threshold. As we know, the cumulative probability distribution of the system (take Astro as an example) will down for one year is about 0.01%. This downtime is equal to 52mins/year that the signal will drop due to the attenuation of rain. In this case we don’t know the exact number of events occur per year, how much the signal drop and also the duration for one event occurs. Hence, this project is expected to answer all of the questions stated previously. 1.4 Scope Of Project This project will involve an extensive review on the available models describing annual rain fade distribution on line-of-sight links. There are many fade duration models that was published in order to predict the rain fade duration. The examples of the published prediction models of fade duration are; • ITU-R Model [1] • Chris J Gibbins and Kevin S Paulson [2] • Zsolt Kormanyos/ Lena Pedersen/ Cyril Sagot/ Janos Bito [3] • Mopfouma [4] • A two component rain model [5] Details on ITU-R Model [1], Chris J Gibbins and Kevin S Paulson [2] and Zsolt Kormanyos/ Lena Pedersen/ Cyril Sagot/ Janos Bito [3] will be discussed in the next chapter. 5 Secondly, do analysis on the available rain attenuation data in UTM. This data was conducted on an experimental 350m, 38GHz by Wireless Communication Center (WCC) [6]. At first, the purpose of these rain attenuation data is to produce cumulative distribution (CD) of rain attenuation data. And for the purpose of this project, these data will be used to obtain information on the rain fade statistics. This rain fade statistics will be developed by using Matlab software. The result from this rain fade statistics will be used to compare with the available empirical models. The best empirical models that suit to the obtained rain fade statistic will be adopted as rain fade model in Malaysia. 1.5 Importance Of The Project Fade duration is an important parameter to be taken into account in system design for several reasons; • System outage and unavailability: fade duration statistics provide information on number and duration of outages and system unavailability due to propagation on a given link and service; • Sharing of the system resource: it is important from the operator’s point-of-view to have an insight into the statistical duration of an event in order to assign the resource for other users. For example, nowadays most of the telecommunication systems are based on bandwidth on demand. In this case, when signal is dropped then we must assign small bandwidth to users in order to make sure that C/N 6 (Carrier to Noise ratio) is high. If not, user will no longer receive any signal during that event; • FMT (Fade Mitigation Technique): fade duration is of concern to define statistical duration for the system to stay in a compensation configuration before coming back to its nominal mode. This FMT will be discussed in more detail later; • System coding and modulation: fade duration is a key element in the process of choosing forward error correction codes and best modulation schemes; for satellite communication systems, the propagation channel does not produce independent errors but blocks of errors. Fade duration impacts directly on the choice of the coding scheme (size of the coding word in block codes, interleaving in concatenated codes, etc.). CHAPTER II LITERATURE REVIEW 2.1 Rain Attenuation on Satellite Communication Satellite transmissions are carried on one of two frequencies: C-band or Ku- band. When operating at the higher frequency Ku-band, the strength of the satellite signal may be temporarily reduced under severe rain conditions. To compensate for these potential effects, earth stations located in heavy rain areas are designed with more transmit power. C-band transmissions are virtually immune to adverse weather conditions. Signal attenuation due to rain is a characteristic of both microwave and satellite transmissions. It is the interference caused by raindrops on electromagnetic signals traveling through the atmosphere. When this phenomenon occurs, the transmission is weakened by absorption and scattering of the signal by raindrops. 8 The level of attenuation is the product of a number of variables, and to minimize its effect, a rain fade margin should be included when designing satellite services and equipment. The rain fade margin is the amount of extra power of satellite adds to the signal strength to compensate for the possibility of rain attenuation. In most cases the reduction in signal strength due to rain does not surpass the rain fade margin and does not have any noticeable effect on transmission. Generally, rain attenuation increases as the signal frequency increases. Therefore, transmissions at 6/4 GHz will experience insignificant attenuation, while transmissions at 14/12 GHz will experience greater attenuation. For 6/4 GHz signals to be affected would require rain storms approaching hurricane conditions. Signals at higher frequencies can be affected by less severe storms. This is due to the wavelength of each frequency and the size of the raindrop through which the signal has to pass. Transmissions at 6/4 GHz have a longer wavelength than transmissions at 14/12 GHz, and are less susceptible to rain attenuation. For example, a 6/4 GHz frequency has a wave-length of approximately 7 cm, and a 14/12 GHz frequency has a wavelength of approximately 2 cm. Any raindrop in the path of either signal which approached half the wavelength in diameter will cause attenuation. How long a transmission will be affected by rain attenuation and how deep the attenuation will be is determined by the amount of rainfall. Generally, signal strength can be affected for two to three minutes during an average rainfall, and up to 15 minutes for extremely heavy rain periods. However, attenuation periods of up to 15 minutes are extremely rare, and although signal strength may be affected, there will be no noticeable effect on transmission as long as the attenuation does not exceed the allocated rain fade margin. 9 To offset the effects of external forces on satellite transmissions, satellite provider must builds a link margin into its calculations when designing satellite services. This margin is the amount of extra transmission power of signal strength satellite provides so the service is not affected by rain attenuation during normal rainfalls. The rain fade margin is a component of the link margin, and is a calculation of expected rain attenuation over one year. It is based on rainfall data, elevation angle, and weather patterns. This margin gives each customer more power than is needed at any given time, so that when rain attenuation occurs, it rarely affects the service. Based on the link margin and the built-in rain attenuation margin, each customer should typically meet or exceed space segment performance specifications 99.9 per cent of the time over one year for 14/12 GHz service, and 99.95 per cent of the time over one year for 6/4 GHz service. This reduces the possibility of rain attenuation affecting your service, and confines the effects to very heavy and very infrequent rain periods. 2.2 Fade Mitigation Techniques Many research groups have investigated the use of fade countermeasures. It is important for the system to stay in compensation configuration during the fade duration. This means, users still can get the signal even though at that time the rain attenuation happens. As a result the fade mitigation techniques (FMT) have been proposed. This FMT can be divided into three main classes; 10 • Power control • Signal Processing • Diversity 2.2.1 Power Control The transmitter power is adjusted to compensate for change in signal attenuation. For example; if the signal dropped about 10dB so the system must transmit at 10dB also in order to compensate the change. The uplink is the critical portion of the connection; the connection is enhanced if the uplink operates with an increased EIRP in rain (ULPC). There are two type of ULPC; • Closed loop ULPC – signal power is detected at the satellite and a control signal is sent back to earth station to adjust the power. • Opened loop ULPC - The fade on the downlink signal is used to predict the likely fade level occurring on the uplink. Closed loop operation is always more accurate but is more expensive to implement. It is due to the operations must always looping until the signal back to it normal mode. Hence, most ULPC systems are, at present is open loop. 11 2.2.2 Signal Processing For this type of fade mitigation, an on board processing (OBP) have been introduced. Where the OBP is used to translates the digital carriers arriving at the satellite to base band for processing and onward transmission back to earth. For example, when the rain attenuation occurs, the OBP will adjust the bandwidth so that all users get the same bandwidth. It is doesn’t matter if the bandwidth is too small as long as users still can use the service during that time. The use of OBP separates the uplink from the downlink and each part of the link can be treated separately in developing a link budget. 2.2.3 Diversity There are three types of diversity available. They are; • Time diversity – it is only used when OBP are being used on the satellite. Additional slots in the TDMA frame can be assigned to the rain affected link so that the same signal can be sent at a slower rate; essentially lowering the bandwidth and raising the C/N. • Frequency diversity – switch from high frequency to the lower frequency. For example; a rain affected Ku-band link could be switched to C band, which is not attenuated significantly by rain. This technique only can be done if the satellite operates in a number of frequency bands. 12 • Site diversity - A technique whereby two or more earth stations are located sufficiently far apart. Where or this technique, the signal will be divert to the earth station where it is uncorrelated with the rain effect. The paths through the rain that are uncorrelated and so the technique is more accurately described as Path diversity 2.3 Available Prediction Model There are a lots of fade duration models that was published by the researchers in the recent years. These models are used to predict the fade duration of rain attenuation. So as the purpose of this project, the comparison and analyze of the available models will be made with the available rain data. Then the best model that suit with the measured data will be adopted for used in Malaysia. 2.3.1 ITU-R Model ITU-R Model [1] is the newest prediction model that proposed by ITU-R. A few reasons have been considered in order to develop this model [1]. Below are three main reasons that have been considered for developing this model [1]. 13 • For a variety of radio communication services, they required the information on the dynamic of outage events • For the evaluation of parameters associated with the risk of failure to provide a certain quality and reliability of services, the probability of occurrences of fades of a certain duration must be known • There is a need to provide engineering information for the calculation of fade duration According to [1], the model are consists of a log-normal distribution function for long fades and a power-law function for short fades. The boundary between short and long fades is given by the threshold duration Dt calculated in the model [1]. The power- law model is valid for fade durations longer than 1 s. Fades of shorter duration do not contribute significantly to total outage time. The following provides estimates of the parameters required for the model [1] and finally defines for distribution functions, i.e. the occurrence probability P and the exceedance probability (or fraction of time) F. The model [1] is expected to be valid for duration longer than 1 s. The following parameters are required as input to the model [1]; f: frequency (GHz): 10-50 GHz ϕ: elevation angle (degrees): 5-60° A: attenuation threshold (dB) 14 The step by step calculation of the fade duration distribution is as follows; Step 1: Calculate the mean duration D0 of the log-normal distribution of the fraction of fading time due to fades of long duration, given that the attenuation is greater than A, as: D0 = 80ϕ −0.4 f 1.4 A −0.39 (1) Step 2: Calculate the standard deviation ⌠ of the lognormal distribution of the fraction of fading time due to fades of long duration as: σ = 1.85 f −0.05 A −0.027 (2) Step 3: Calculate the exponent γ of the power-law distribution of the fraction of fading time due to fades of short duration as: γ = 0.055 f 0.65 A−0.003 (3) Step 4: Calculate the boundary between short and long fade durations, Dt, as: 2 Dt = D0e p1σ + p 2σ − 0.39 (4) where; p1 = 0.885γ − 0.814 (5) p2 = −1.05γ 2 + 2.23γ − 1.61 (6) Step 5: Calculate the mean duration D2 of the log-normal distribution of the probability of occurrence of fading events of long duration as: 15 2 D2 = D0e −σ (7) Step 6: Calculate the fraction of time k due to fades of duration less than Dt as: −1 ⎡ ⎛ ln( Dt ) − ln( D0 ) ⎞ ⎤ ⎢ D0 D2 (1 − γ )Q⎜ ⎟⎥ ⎝ σ ⎠⎥ k = ⎢1 + (8) ⎢ ⎛ ln( Dt ) − ln( D 2) ⎞ ⎥ Dt γQ⎜ ⎟ ⎢ ⎣ ⎝ σ ⎠ ⎥ ⎦ where; Q: standard cumulative distribution function for a normally distributed variable. ∞ 1 1 − x2 2π ∫ Q( z ) = e 2 dx (9) z Step 7: Calculate the probability of occurrence of fade events of duration d longer than D given that attenuation a is greater than A as: (10) For 1≤ D ≤ Dt P(d > D|a > A) = D–γ ⎛ ln( D) − ln( D2 ) ⎞ Q⎜ ⎟ −γ ⎝ σ ⎠ For D > Dt P(d > D | a > A) = Dt ⋅ (11) ⎛ ln( Dt ) − ln( D2 ) ⎞ Q⎜ ⎟ ⎝ σ ⎠ Step 8: Calculate the cumulative probability of exceedance, i.e. the total fraction of fade time due to fades of duration d longer than D: 1− γ ⎡ ⎛D⎞ ⎤ For 1≤ D ≤ Dt F (d > D | a > A) = ⎢1 − k ⎜ ⎟ ⎥ ⎜D ⎟ (12) ⎢ ⎣ ⎝ t⎠ ⎥ ⎦ 16 ⎛ ln( D) − ln( D0 ) ⎞ Q⎜ ⎟ σ ⎠ For D > Dt F (d > D | a > A) = (1 − k ) ⋅ ⎝ (13) ⎛ ln( Dt ) − ln( D0 ) ⎞ Q⎜ ⎟ ⎝ σ ⎠ Step 9: If required, the total number of fades of duration d longer than D for a given threshold A can be calculated from: N(D, A) = P(d > D|a > A) x Ntot(A) (14) Where D is the mean duration time of fade duration The total fading time due to fades of duration d longer than D for the threshold A is: T(d > D|a > A) = F(d > D|a > A) x Ttot(A) (15) Where Ttot(A) is the total time the threshold A is exceeded and Ntot(A) is the total number of fades exceeding the minimum duration of 1s. These parameters can be obtained in the following way: Ttot(A) should be obtained from local data. If this long- term statistic is not available, an estimate can be calculated from Recommendation ITU- R P.618 [7]. In this case the procedure consists in calculating the CDF of total attenuation, deriving the percentage of time the considered attenuation threshold A is exceeded and then the associated total exceedance time Ttot(A) for the reference period considered. Once Ttot(A) has been obtained, Ntot(A) can be calculated as: k 1− γ N tot ( A) = Ttot ( A) ⋅ ⋅ (16) γ Dt1 − γ 17 The above method [1] was tested against the Radio communication Study Group 3 fade duration data bank for frequencies between 11 and 50 GHz and for elevation angles between 6° and 60°. The arithmetic mean of the logarithmic error (ratio of the predicted to the measured fade duration at the same probability level) was found to be 30% for fade durations shorter than 10 s and between –25% and –80% for fade durations longer than 10 s. As far as the standard deviation is concerned, it was found to range between 80% and 150%, demonstrating the high natural variability of this parameter. 2.3.2 Chris J. Gibbins / Kevin S. Paulson The model [2] is proposed to predict the system outages due the rain attenuation. Where is, it presents two algorithms for estimating the distribution of the durations of rain fade. According to [2], the distribution of the durations of rain events at different intensities is more fundamental than distributions of link fade. An expression for the average number of events per year of rain intensity greater than a give threshold has been developed from rain duration data gathered in the South East UK over three years. As mentioned earlier, in this model [2], two mathematical models have been considered. These models are the lognormal model and power law mode. The lognormal model was chosen due to it fit to each of the distributions of durations with given rain rates. The lognormal model as shown below; −1.76 (ln t d − 2) 2 N = 1.70 x10 R 4 exp(− ) (17) 3.86 − 0.0409 R 18 where R is the rain rate, N represent of number of rain events per year and td is the duration of rain rate. In order to develop this model [2], the lognormal model above had been extended to provide an estimate number of rain fading events, using an analysis of fade duration at 38 GHz over a 9 km path. The model [2] is based on the rainfall rate R (mm/h) which gives rise to a given path attenuation, A, from the expression: A = k Rα.d.r(dB); where k and a are constants obtained from ITU-R Rec.P.838 [8] and d is the path length in km. The path length reduction factor r takes account of the fact that rain is not uniformly distributed along the path. In developing the model, two expressions for the path length reduction factor were considered, that in ITU-R Rec.P.530 [9] and the RAL model for the path reduction factor developed from radar data [10], in which r is given by the minimum of the following two expressions. r = 1.35 + s(d) log R and r=1 (18) Where R is the rainfall rate and s(d) = 2d-0.053 – 2.25. For a given fade depth A, in dB, the appropriate point rainfall RA can then be determined. In analogy with the path length reduction factor, a similar time dilation factor is necessary to convert between the durations experienced by a point rain rate measurement device, such as rain gauge, and the duration of rain fade experienced by a link. Fade durations are longer than equivalent point rain duration as the rain can interact with the link anywhere along its length. A proportion of rain events will travel along the length of the link with the prevailing wind and cause fade much longer duration than the rain duration measured at a point. This duration effect was found to fit the following expression, (19); t d = 273R −0.89 + (0.166 + 0.0194 R )t ' d (19) where t’d is the fade duration experienced by the link.Using this procedure, the model for rain rate durations was fitted to the 38 GHz fade data, with the best fit being 19 provided by the RAL path reduction factor, to yield the following expression for the number of fades NA exceeding a depth of A dB and duration of td seconds; [ln(273R A.39 + (0.166 + 0.0194 R A )t ' d −2] 2 0 N A = 1.70 x10 4 R −1.76 exp(− ) (20) 3.86 − 0.0409 R A From this model [2], the average number of fading events per year can be estimated; taking account of the fact that rain is generally not uniformly distributed along a path, so using the path length reduction factor as a solution. It should be in mind, however, that the above expression, (20) was derived from point rainfall rate measurements. For practical link lengths, it may underestimate the number of fade events exceeding a given duration and attenuation for the following reason. Rain cells typically have diameters of the order of 2km. if such rain cell passes transversely across a radio link, fade duration of certain period will occur. If however, the same rain cell travels along the link, the durations of events with the same level of attenuation would be somewhat longer than in the transverse case, depending on the length of the link and the speed which the rain cell travels. 2.3.3 Kormanyos/ Lena Pedersen/ Cyril Sagot/ Janos Bito According to [3], for mobile operators it is very important to plan highly reliable millimeter wave links. Where, there are several proposed methods for predicting rain attenuation statistic. For example the International Telecommunication Union (ITU) has issued a recommendation for rain attenuation prediction [9]. The method is clearly of high quality, but it is still of interest to check the method and investigated a new attenuation prediction method [3]. 20 The new method [3] is based on Moupfouma’s semiempirical rain intensity distribution prediction [4]. The method given in [9] is valid for temperate zones, it calculates attenuation distribution for 38 GHz in frequency and it also applies the measured R0.01 value; b 1 α ⎛ A0./01 + (k ⋅ L. ⋅ r )1 / α ⎞ ⎜ 1/ α P ( Ai ) = ⎜ ⎟ ⋅ 10 − 4⋅a (21) 1/ α ⎟ ⎝ Ai + (k ⋅ L. ⋅ r ) ⎠ 1/ α ⎛ A ⎞ ⎡ ⎛ A ⎞1 / α ⎤ b=⎜ i ⎟ ⎜A ⎟ ⋅ ln ⎢1 + ⎜ i ⎟ ⎥ (22) ⎝ 0.01 ⎠ ⎢ ⎜ A0.01 ⎟ ⎥ ⎣ ⎝ ⎠ ⎦ 1 − ( Ai / A0.01 ) 1/ α a = 1− (23) 1 + 4.56 ⋅ ( Ai / A0.01 ) 1.03 / α 1 α Where k, A, and r are taken from ITU recommendation [9]. A0.01 = k ⋅ R0./01 ⋅ L ⋅ r ,where L and r are the path length and the path reduction factor. This prediction model [3] is based on rain intensity prediction. Beside the rain attenuation distribution, fade duration and other dynamic statistics are also important for mobile operators. They [3] have presented rain attenuation rate of change statistic. In this contribution to evaluate the dynamics of rain attenuation, the number of events and fade duration statistics have been determined for links in Hungary, for the long link in Norway and for a short link in Ireland. Fade duration statistics have been compiled for attenuation exceeding 4, 8, 12, 16 and 20 dB levels and plotted graph shown the number of events for which duration exceeds abscissa at the level given. As it was expected the number of events on longer links is higher than on the shorter ones, and also the duration of fade events is longer. From the analysis that being made, they found out that at 4dB level the number of events with short duration (10-100 sec) is remarkably smaller in Hungary. At higher levels these 21 differences in number of fades disappear for long links. On the links events with long duration (greater than 1000 sec) are presumably caused by melting snow on antennas. 2.4 Summary This proposed model will be based on the available prediction models that have been stated as above. Where for the purpose of this project, to obtain the annual rain fade distributions on line-of-sight from the 38GHz link profile and determine the appropriate rain fade model to be adopted in Malaysia. CHAPTER III METHODOLOGY 3.1 Methodologies The methodologies of this project as shown in the flow chart in figure 3.1 below. In the beginning, there are two tasks that need two do simultaneously. First task is to do some literature reviews on the concept of the rain fade duration as well as to find out the available of existing models that proposed by other researchers. This literature reviews will help to understand in more details what fade durations and rain attenuations are. Beside it can help to identify the comparison of the previous prediction methods that have been proposed. So it can give some ideas to develop the empirical models efficiently. The second task to do is study and mastering the Matlab software. In this part, the main agenda here is how to make the programming. Since this project doesn’t need the simulation, so it will based on programming only. For simplicity of the analysis of the data, necessary graphs are recommended to be plotted. 23 Start Literature Reviews Mastering in Matlab programming Analyisis of the rain data in UTM Identifying existing models [1] Obtain rain fade statistics [2] Result analysis and comparison between [1] and [2] End Figure 3.1: Research methodology flow chart Then do analysis on available rain attenuation data in UTM campus. This analyzing will help to know deeply about the rain attenuation. Hence from the analyzing, the best value to determine the threshold can be identified. This threshold is used to determine how many dB the signal attenuate. This is the most critical part of this research, because once the threshold set wrongly then the incorrect result will get. After the threshold has been set, then proceed with programming in Matlab in order to obtain rain fade statistic. At this stage, measured data will be extracted into particular signal drops (for attenuation exceeding 1, 3, 6, 9, 12, 15, 18, 21 dB levels). Once the data completed extracted, the number of events and fade duration statistics can 24 be determined by plotting graphs. These graphs then will be compared with the existing models in the next stage. From the literature reviews and analysis have been made, we can identify the best existing model that can be adopted in Malaysia. The model that has to choose must perform well and has wide validity range with the locally measured data. Finally do result analysis and comparison. In this stage, tasks that have to do are do analysis on the result and compare the result with available statistical models. And finally determine the appropriate rain fade model to be adopted in Malaysia. 3.2 Data Collection The rain attenuation data collection can be divided into terrestrial and satellite rain attenuation data. The terrestrial rain attenuation data was measured from an experimental point-to-point microwave link operating at 38GHz. While the satellite rain attenuation data was collected from Malaysia-East Asia Satellite (MEASAT). MEASAT is a broadcasting satellite catering Direct-To-Home (DTH) broadcasting services to the South East Asia region operating at Ku band (14/12GHz). But for this project only data measured from terrestrial rain attenuation is concerned. The availability of terrestrial rain attenuation data is for 1 year period from January 1, 1999 to December 31, 1999. The set-up of the terrestrial receiver is shown on Figure 3. 25 Figure 3.2: Path profile for Ericson Microwave Link at 38GHz in Wireless Communication Centre (WCC) University Teknlologi Malaysia The experimental setup of the link consists of a transmitter and a receiver 350m apart. The diameter of both antennas is 0.6m. The operating frequency is 38 GHz. The link is applying horizontal polarization. The Automatic Gain Control (AGC) level (in Volts) of the receiver is connected to a data acquisition system continuously. The signal level is collected in one-second integration time. The signal level is then converted into dBm for analysis. The 350m path length suggests that rain can be assumed homogenous along the signal path. Thus no correction factor both horizontal and vertical is needed. The experimental link setup located in Wireless Communication Center (WCC), UTM Skudai. The data used is the signal level collected by the data acquisition system throughout the year 1999. These rain attenuation data collected were analyzed to produce cumulative distribution (CD) of rain attenuation data. In addition to the rain attenuation data collected, one-minute integration time of rainfall data was also measured downloaded from TRMM satellite system. A total of one year data were collected from January 2000 to February 2002 where the TRMM data is centred at Skudai, Malaysia (1.473°, 03.745°). 26 Figure 3.3 below show the rain contour map in Peninsular Malaysia; Figure 3.3: Rain Rate Contour Map (exceeding depicted rain intensity at 0.01 % of time, R0.01) in Peninsular Malaysia 3.3 Data Extractions As mentioned previously, collected data will be extracted in order to obtain number of events per year and also fade duration statistics. Matlab software is used to extract and manipulate these data. Below are some Matlab programming on how the data are manipulated. 27 First step to do is import data from text file (*.txt) into Matlab. This importing data can be done by using textread function. The programming code to read text file as below; %read data from text file [voltan] = textread('nov.txt','%*s %*s %*s %f'); Figure 3.4: read data from text file Second step is converting read data from AGC values (volt) to dBm values. This can be performed by using this formula; A = (40 × V ) − 120 (24) where is; A = Attenuation (dBm) V = AGC value (volt) And by using Matlab programming, the syntax as below can be writing; %converts AGC values to dBm voltage = 40*voltan-120; Figure 3.5: Converts AGC (volts) values to dBm Third step is set the threshold value into the programming. For this project, the value of threshold chooses is -25.5dB. Where is, the threshold value can be determined 28 from (25). After threshold is set, then compiled fade duration statistics for attenuation exceeding the 1, 3, 6, 9, 12, 15, 18, 21 dB levels and plotted. 1 α A = k ⋅ R0./01 (25) where k and α are taken from ITU recommendation [9] %set threshold value = -25.5dB threshold = -25.5; while x<bil k=k+1; if voltage(k) < threshold if voltage(k) <= voltage(k-1) c=c+1; if (voltage(k)-threshold)< v0 %attenuation exceeding 1db c0=c0+1; else end if (voltage(k)-threshold)< v1 %attenuation exceeding 3db c1=c1+1; else end . . . . end end end Figure 3.6: Compile fade duration statistic for attenuation exceeding threshold 29 Finally save the compiled fade duration statistics into particular file in order to be plotted and manipulated later. This save file can be writing by using save function. %save file y0=[volt0',masa0']; save('1db.txt','y0','-ascii','-append') %save file for attenuation exceeding 1dB save('1209.txt','y0','-ascii','-append') %save file through month of events save('overall.txt','y0','-ascii','-append') %save all fade duration statistics here y1=[volt1',masa1']; save('3db.txt','y1','-ascii','-append') %save file for attenuation exceeding 3dB save('1209.txt','y1','-ascii','-append') %save file through month of events save('overall.txt','y1','-ascii','-append') %save all fade duration statistics here . . . Figure 3.7: Save fade duration statistics data to corresponding files 3.4 Existing Rain Models Programming In this topic, it will discuss how to develop the programming for the existing rain models that mentioned earlier. This programming must be done in order to make comparison between measured data with the published rain models. After the comparison has been made, the best prediction model can be determined and will be adopted for the prediction of fade duration in Malaysia. As mentioned in previous chapter only three models are considered for this project. Those models are; 30 • ITU-R Model [1] • Chris J Gibbins and Kevin S Paulson [2] • Zsolt Kormanyos/ Lena Pedersen/ Cyril Sagot/ Janos Bito [3] 3.4.1 ITU-R Model Programming As discussed previously, this model [1] is consisting of a log-normal distribution function for long fades and power-law function for short fades. The boundary between short and long fades is given by the threshold duration Dt. There are three parameters are required as input for the model [1]; f: frequency (GHz): 10-50 GHz ϕ: elevation angle (degrees): 5-60° A: attenuation threshold (dB) For the purpose of this project, the frequency used is 38GHz, elevation angle used is 38.98° (this value got from MEASAT) and attenuation threshold is -25.5dB. First step to do is set all these three values into the Matlab programming. Followed by develop all the equations involved in model [1] into Matlab. Then simulate the model and plotting graph. f=38; %frequency GHz e=0.6803; %elevation angle degress a=26; %attenuation threshold (dB) . Figure 3.8: Set parameters with correct values 31 d0 = 80 * 38.98^(-0.4) * 38^1.4 * 26^(-0.39); %equation (1) sigma = 1.85 * 38^(-0.05) * 26^(-0.027); %equation (2) y = 0.055 * 38^0.65 * 26^(-0.003); %equation (3) p1 = 0.885*y - 0.814; %equation (5) p2 = (-1.05)*y^2 + 2.23*y -1.61; %equation (6) dt = d0 * exp(p1*(sigma^2) + p2*sigma - 0.39); %equation (4) d2 = d0*exp(-(sigma^2)); %equation (7) q1 = (log(dt) - log(d0)) / sigma; q2 = (log(dt) - log(d2)) / sigma; denom1 = (0.661*q1) + (0.339*sqrt((q1^2)+5.51)); %equation (9) Q functions denom1 = denom1*sqrt(2*pi); denom2 = (0.661*q2) + (0.339*sqrt((q2^2)+5.51)); denom2 = denom2*sqrt(2*pi); Q1 = (1/denom1)* exp(-(q1^2) / 2); %equation (9) Q functions Q2 = (1/denom2)* exp(-(q2^2) / 2); k = (1 + ((sqrt(d0*d2) * (1-y) * Q1) / (dt * y * Q2)))^(-1); %equation (8) Figure 3.9: Matlab programming for corresponding equations if masa3(n)<dt %for 1<D<Dt P(n)= masa3(n)^(-(y)); %equation (10) else q3 = (log(masa3(n)) - log(d2)) / sigma; %for D>Dt denom3 = (0.661*q3) + (0.339*sqrt((q3^2)+5.51)); denom3 = denom3*sqrt(2*pi); Q3 = (1/denom3)* exp(-(q3^2) / 2); %Q functions P(n)= (dt^(-(y)))*(Q3/Q2); %equation (11) end N(n)=P(n)*event3(n); %total number of fade %duration; equation (14) Figure 3.10: Matlab programming for total number of fade durations 32 3.4.2 Chris J Gibbins and Kevin S Paulson Model Programming This model [2] has considered two mathematical models – the log-normal model and power-law model. The log-normal was chosen due to it fit to each of the distributions of durations with given rain rates. The model [2] is based on the rainfall rate R (mm/h) which gives rise to a given path attenuation, A from the expression: A = k Rα.d.r(dB); where k and a are constants obtained from ITU-R Rec.P.838 [8] and d is the path length in km. Take notes, for this project path length reduction factor, r is neglected due to the length from transmitter and receiver is short only 350m. As usual first step to do is develop Matlab programming for all equations involved to this model [2]. After that, simulate the model and plotting graph. %the number of fade exceeding threshold,A; equation (9) event(1)=1.7*(10^4)*(R^(-1.76))* exp(-(log(td1)-2)^2/(3.86-(0.0409*R))); Figure 3.11: Matlab programming for number of fades NA 33 3.4.3 Zsolt Kormanyos/ Lena Pedersen/ Cyril Sagot/ Janos Bito Model Programming The model [3] is based on Mopfouma’s semiemprical rain intensity prediction [4]. Where is, the model [3] is based on rain intensity prediction and it also valid for temperate zones. The Matlab programming codes as stated below. b=(((Ai/A)^(1/alfa))-1)*log((1+((Ai/A)^(1/alfa)))); %equation (22) a=1-((1-((Ai/A)*(1/alfa)))/(1+(4.56*((Ai/A)^(1.03/alfa))))); %equation (21) %probability numbers of fades duration; equation (20) P=((((A^(1/alfa))+((k*L*r)^(1/alfa)))/((Ai^(1/alfa))+((k*L*r)^(1/alfa))))^b)*(10^(- 4*a)) Figure 3.12: Matlab programming for probability numbers of fades duration 3.5 Comparison Finally do comparison with measured data and these three models. This comparison can be done by plotting the graph for each models and measured data. From this graph, the appropriate rain fade model can be determined where is the graph is closely to the measure data graph. Below is some Matlab programming codes on how to do this comparison. 34 %read fade duration statistic data (measured data) [masa, db] = textread('overall-time.txt','%f %f'); itutrydt %call Itu-R model [1] empirical %call gibbins model [2] bito %call bito model [3] figure(2); loglog(masa,db,time2,N2,time,N,masa1,event) %plot log graph legend('measured data','Bito model','Itu-r model','Paulson model') %legends xlabel('Duration [sec]') %x-axis label ylabel('No of events exceeding abscissa') %y-axis label title('Anuual Fade duration in UTM 99 (38GHz)') %graph title Figure 3.13: Matlab programming for comparison among measure data and studied rain fade models CHAPTER IV RESULTS 4.1 Introduction This chapter will discuss the results that obtained from the project that has been studied. Where is, it consist of two section. First section is how to interpret the measured data into fade duration statistic. Second section is the comparison that being made among fade duration statistic (measured data) with studied models. 4.2 Measured Data Interpretation 4.2.1 Power Received 36 Figure 4.1 and 4.2 below shows the differences between signal attenuation on clear weather and signal attenuation on rainy weather. On the clear day, the signal received at receiver is above the attenuation threshold, -25.5 dB. So on this weather the satellite and telecommunication systems are free from any outage due to rain attenuation. But on the rainy weather the signal received at receiver are below the attenuation threshold, -25.5 dB. This happens due to interference cause by raindrops and electromagnetic signal traveling through the atmosphere. When this phenomenon occurs, the transmission is weakened by absorption and scattering of the signal by raindrops. As a result any telecommunication and satellite systems that operate during this duration will experience outage due to rain attenuation. The more signal attenuated below threshold, the more system will experience the outage. Thus, the use of suitable compensation techniques such as FMT will be needed to maintain reliable system operation. In order to use the suitable compensation techniques, one must know the information of fade duration statistics. This fade duration statistics can be obtained from studying the number and duration the signal received at receiver below the attenuation threshold. Figure 4.1; Received power level time series on clear weather 37 Figure 4.2: Received power level time series on rainy weather 4.2.2 Data Analyzed Measure data will be converted first from AGC (V) value to decibels (dBm) values before analysis can be preceded. As mentioned previously, power received (dBm) can be obtained from equation below; Power received (dBm) = 40 * V -120 (26) Table 4.1 shows the sample of measured data that been converted from AGC (V) values to decibel (dBm) values. 38 Table 4.1: Convert AGC (V) to decibel (dBm) After converting have been done, the next step plotting signal attenuation graph in order to ease in analyzing the measured data. 39 Figure 4.3: Power received signal attenuation Figure 4.4 shows how the fade durations have been analyzed. For example, the fade duration statistic have been compiled for attenuation exceeding the 1 and 2dB levels (take notes; for the purpose of this project the exceeding levels are divided for attenuation exceeding the 1, 3, 6, 9, 12, 15, 18, 21 dB levels. Attenuation exceeding 1 and 2dB levels only shows for example only. But the process is still the same.) 40 -25 -25.5 -26 Power Received [dBm] 3 sec 9 sec -26.5 -27 2 sec -27.5 -28 -28.5 13 49 58 13 50 03 13 50 08 13 50 13 13 50 18 13 50 23 Time [Hr:min:sec] Figure 4.4: Analysis on fade duration Table 4.2: Analysis on fade duration Table 4.2 and Figure 4.4 show how the fade duration statistics have been analyzed. First let’s look at attenuation exceeding 1dB level. There are two events where the power received signal drops exceeding 1dB level. First event occurred at about 13:50:08 hours and ended about 13:50:11 hours, fade duration taken approximately 3 sec. While the second event for exceeding 1dB level is about started at 13:50:12 and ended at about 13:50:21 where is, the fade duration taken approximately 9 sec. 41 There is only one event for attenuation exceeding 2dB level. This event occurred about at 13:50:15 and fade duration taken approximately 2 seconds. From this information, the number of events longer than abscissa can be obtained. This is shown in Table 4.2. Where is for attenuation exceeding 1dB level, there are two events number of fades longer than 1 sec and only one event number of fade longer that 5 sec. While no event occur for fade duration longer than 10 sec. For attenuation exceeding 2dB level, only one event occurs at fade duration longer than 1sec. This analysis processes will be carried on to other data with different durations and dB levels. 4.3 Result and Analysis 4.3.1 Obtain Rain Fade Statistics After all measured data have been processed; the next task is do analysis and comparison for each data processed. This task can be done by plotting graphs. Where from this graph, information on fade duration statistics can be obtained more precisely and details. Table 4.3 shows the number of events for which duration exceeds abscissa at 1, 3, 6, 9, 12, 15, 18 and 21 dB levels. From this table known that when the attenuation is increased the time duration will decrease. This is due that the heavy rains which cause higher attenuation level only happen in short duration. It is because the higher attenuation depending on rain drops size and also the rain intensity. Then all information represents in this table will be plotted to graph in order to get the details scenario of the fade duration statistics. 42 Table 4.3: Number of events for which duration exceeds abscissa at 1, 3, 6, 9, 12, 15, 18 and 21 dB levels Figure 4.5 shows the number of events grouped by exceeding levels versus duration (sec). As it was expected the number of events of attenuation exceeding 1dB level is higher than other levels, and also the duration of fade events is longer. This is due for attenuation exceeding 21dB level; it needs bigger rain intensity and longer raining time compare to attenuation exceeding 1dB level. 43 Figure 4.5: Fade duration grouping by attenuation levels Figure 4.6 shows the number of fade events grouped by months. As expected, August, September, October and November are months that have higher number of fade events compare to other months. This is due to monsoon season during these 4 months in Malaysia. Where is, in monsoon season the intensities of rain are bigger than normal season and the rain drop size also bigger. So no wonder during this season, the number of fade events is higher and the fade durations are longer too. Table 4.4: Number of events for each month for one year 45 Figure 4.6: Fade duration grouping by months 4.3.2 Comparison Finally comparison among measured data and available rain models have been made in order to determine the best model that suit with locally measured data to be adopted as prediction model in Malaysia. 46 Table 4.5: Fade duration statistics information Table 4.5 above shows the fade duration statistics information. This table represented the number of fade events which exceeds abscissa per year. The information of this fade duration statistics will be transferred to graph in order to make comparison with studied models. Figure 4.7 below shows the comparison among the available rain models with fade duration statistics that we obtained previously. 47 Figure 4.7: Comparison among existed rain models with measured data From Figure 4.7 above, ITU-R rain model [1] obviously have wide validity range and performed very well with the locally measured data compare to Paulson model [2] and Bito model [3]. It is because; ITU-R model has used wide range of rain database in order to develop the model. So as a result it suit to any climate around of the world. Besides, ITU-R model also depends on local data parameters such as elevation angle, frequency being used and also the local rain rate. The second best model perform well with locally measured data is Bito model. This is because Bito model only used to predict the rain fade duration for temperate climate only, but the studied area is in extreme climate where is Malaysia received rain throughout the year. This model can still be used in Malaysia but needed some 48 modification of the coefficients of this model. Paulson model not perform very well because this model is developed by using UK rain database only. Most of the coefficients from this model cannot be used in Malaysia. Besides, this model depends on the length of the link and the speed with which the rain cell travels. So as a result, the ITU-R model is the appropriate rain fade model to be adopted in Malaysia as a rain fade prediction model. CHAPTER V CONCLUSSION 5.1 Conclusions The conclusion that can be made from this project is the rain fade duration statistics can be obtained from measured data. Where is, from locally measure data, these data have been interpreted into number of groups of attenuation exceeding given threshold. After the data have been interpreted, then do analysis on that data and obtain the rain fade duration statistics for one year duration. Finally, the number of events per year can be estimated from these measured data. Secondly, ITU-R prediction model [1] is the best model that can be adopted for Malaysia climate. This can be proved from observation and comparison that being made previously. Besides, the ITU-R model is updated regularly for every two years by using more rain database. So no wonder ITU-R model becomes a global model that suit to any climates around the world. It is because ITU-R model is dependently on local parameters such as rain rate, frequency used and elevation angle. 50 This prediction model is important in system design especially to have flexible and broadband feeder network. As we know, nowadays all telecommunication systems are based on bandwidth demand. So this prediction model will help the service provider to provide best service during raining day. 5.2 Future Work From this project, there are three suggestions that I can suggest in order to enhance this prediction models. They are; • Proposed for rain fade satellite measurement. It is because this project is based on line-of-sight measurement. • Analysis with different frequency. Instead of using 38GHz as frequency, why not use other frequencies especially Ku and Ka band that usually being used in satellite and telecommunication systems. • Obtain new Fade Mitigation Techniques (FMT) REFERENCES [1] ITU-R P.1623-1, “Prediction method of fade dynamics on Earth-space paths”, ITU, Geneva, 2005 [2] Paulson, K.S.; Gibbins, C.J.; “Rain models for the prediction of fade duration at milimetre wavelengths”, IEEE Journal, Volume 147, Issue 6, Dec. 2000 Page(s):431 – 436 [3] Zsolt Kormayos, Lena Pedersen, Cyril Sagot, Janos Bito; “Rain attenuation and fade duration statistic at 38 GHz derived from long term radio link measurement in Hungary, Norway and Ireland”, AP2000, Davos, 13 April 2000 [4] F. Moupfouma and L. Martin, “Point rainfall rate cumulative distribution function valid at various locations,” Electronic Letters, vol. 29, no. 17, pp. 1503- 1505, 1993 [5] M.M.J.L. van de Kamp; “Short-term prediction of rain attenuation using two samples”, Electronic Letters, vol. 38, no. 23, pp. 1476-1477, 7th November 2002 [6] R. Singliar, J. Din, L. Csurgai, A.R. Tharek, P. Horvath and J. Bito, “Comparison of 38 GHz rain fade dynamics between Malaysia and Hungary”, 15th IST Mobile & Wireless Communication Summit Myconos 4-8 June 2006 52 [7] ITU-R P.618-6, “Propagation data and prediction methods required for the design of earth-space telecommunication systems”, ITU, 1999 [8] ITU-R P.838-6, “Specific attenuation model for rain for use in prediction methods”, ITU, 1999 [9] ITU-R P.530-8, “Propagation data and prediction method required for the design of terrestrial line of sight systems”, ITU, 1999 [10] J.W.F. Goddard and M. Thurai, “Radar derived path reduction factors for terrestrial systems,” Proc. Tenth International Conference on Antenna and Propagation, ICAP’ 97, 14-17 April 1997, Edinburg UK, pp. 2.218-2.221 APPENDIX A FLOW CHART 54 APPENDIX A FLOW CHART FADE DURATION STATISTICS PROGRAMMING Start Set threshold Read measured data from text file Convert AGC (V) to dBm value A< threshold A= Yes 1dB? Count fades duration time No Save data into particular file D B 55 D B A= Yes 3dB? No Count fades duration time Save data into particular file A= Yes 6dB? No Count fades duration time Save data into particular file A= Yes 9dB? No Count fades duration time Save data into particular file A= Yes 12dB? No Count fades duration time Save data into particular file D C 56 D C A= Yes 15dB? No Count fades duration time Save data into particular file A= Yes 18dB? No Count fades duration time Save data into particular file A= Yes 21dB? No Count fades duration time Save data into particular file APPENDIX B PROGRAMMING CODES 58 APPENDIX B Fade Duration Statistic (Measured Data) [voltan] = textread('dec.txt','%*s %*s %*s %f'); %read data from text file voltage = 40*voltan-120; %convert AGC values to dBm bil = length(voltage); bil=bil-1; threshold = -25.5; %set threshold x=0; k=1; c=0;c0=0;c1=0;c2=0;c3=0;c4=0;c5=0;c6=0;c7=0; v0=-1;v1=-3;v2=-6;v3=-9;v4=-12;v5=-15;v6=-18;v7=-21; n0=1;n1=1;n2=1;n3=1;n4=1;n5=1;n6=1;n7=1; b0=1;b1=1;b2=1;b3=1;b4=1;b5=1;b6=1;b7=1; while x<bil k=k+1; if voltage(k) < threshold %Attenuation < threshold if voltage(k) <= voltage(k-1) %Attenuation(k) <= Attenuation(k-1) c=c+1; if (voltage(k)-threshold)< v0 %Attenuation exceeding 1dB c0=c0+1; else end if (voltage(k)-threshold)< v1 %Attenuation exceeding 3dB c1=c1+1; else end if (voltage(k)-threshold)< v2 %Attenuation exceeding 6dB c2=c2+1; else end if (voltage(k)-threshold)< v3 %Attenuation exceeding 9dB c3=c3+1; else end if (voltage(k)-threshold)< v4 %Attenuation exceeding 12dB c4=c4+1; 59 else end if (voltage(k)-threshold)< v5 %Attenuation exceeding 15dB c5=c5+1; else end if (voltage(k)-threshold)< v6 %Attenuation exceeding 18dB c6=c6+1; else end if (voltage(k)-threshold)< v7 %Attenuation exceeding 21dB c7=c7+1; else end else if (voltage(k)-threshold)<= v0 %Attenuation exceeding 1dB c0=c0+1; if k > bil masa0(n0) = c0; volt0(b0)= -1; c1=0; n0=n0+1; b0=b0+1; end else if c0 == 0 else masa0(n0) = c0; volt0(b0)= -1; c0=0; n0=n0+1; b0=b0+1; end end if (voltage(k)-threshold)<= v1 %Attenuation exceeding 3dB c1=c1+1; if k > bil masa1(n1) = c1; volt1(b1)= -3; c2=0; n1=n1+1; b1=b1+1; end 60 else if c1 == 0 else masa1(n1) = c1; volt1(b1)= -3; c1=0; n1=n1+1; b1=b1+1; end end if (voltage(k)-threshold)<= v2 %Attenuation exceeding 6dB c2=c2+1; if k > bil masa2(n2) = c2; volt2(b2)= -6; c3=0; n2=n2+1; b2=b2+1; end else if c2 == 0 else masa2(n2) = c2; volt2(b2)= -6; c2=0; n2=n2+1; b2=b2+1; end end if (voltage(k)-threshold)<= v3 %Attenuation exceeding 9dB c3=c3+1; if k > bil masa3(n3) = c3; volt3(b3)= -9; c4=0; n3=n3+1; b3=b3+1; end else if c3 == 0 else masa3(n3) = c3; volt3(b3)= -9; c3=0; 61 n3=n3+1; b3=b3+1; end end if (voltage(k)-threshold)<= v4 %Attenuation exceeding 12dB c4=c4+1; if k > bil masa4(n4) = c4; volt4(b4)= -12; c5=0; n4=n4+1; b4=b4+1; end else if c4 == 0 else masa4(n4) = c4; volt4(b4)= -12; c4=0; n4=n4+1; b4=b4+1; end end if (voltage(k)-threshold)<= v5 %Attenuation exceeding 15dB c5=c5+1; if k > bil masa5(n5) = c5; volt5(b5)= -15; c6=0; n5=n5+1; b5=b5+1; end else if c5 == 0 else masa5(n5) = c5; volt5(b5)= -15; c5=0; n5=n5+1; b5=b5+1; end end if (voltage(k)-threshold)<= v6 %Attenuation exceeding 18dB 62 c6=c6+1; if k > bil masa6(n6) = c6; volt6(b6)= -18; c7=0; n6=n6+1; b6=b6+1; end else if c6 == 0 else masa6(b6) = c6; volt6(b6)= -18; c6=0; n6=n6+1; b6=b6+1; end end if (voltage(k)-threshold)<= v7 %Attenuation exceeding 21dB c7=c7+1; if k > bil masa7(n7) = c7; volt7(b7)= -21; c8=0; n7=n7+1; b7=b7+1; end else if c7 == 0 else masa7(n7) = c7; volt7(b7)= -21; c7=0; n7=n7+1; b7=b7+1; end end end else %Attenuation > threshold if c0==0 %Attenuation exceeding 1dB else masa0(n0) = c0; volt0(b0)= -1; 63 c0=0; n0=n0+1; b0=b0+1; end if c1==0 %Attenuation exceeding 3dB else masa1(n1) = c1; volt1(b1)= -3; c1=0; n1=n1+1; b1=b1+1; end if c2 == 0 else %Attenuation exceeding 6dB masa2(n2) = c2; volt2(b2)= -6; c2=0; n2=n2+1; b2=b2+1; end if c3 == 0 else %Attenuation exceeding 9dB masa3(n3) = c3; volt3(b3)= -9; c3=0; n3=n3+1; b3=b3+1; end if c4 == 0 else %Attenuation exceeding 12dB masa4(n4) = c4; volt4(b4)= -12; c4=0; n4=n4+1; b4=b4+1; end if c5 == 0 else %Attenuation exceeding 15dB masa5(n5) = c5; volt5(b5)= -15; c5=0; n5=n5+1; b5=b5+1; end if c6 == 0 else %Attenuation exceeding 18dB 64 masa6(n6) = c6; volt6(b6)= -18; c6=0; n6=n6+1; b6=b6+1; end if c7 == 0 else %Attenuation exceeding 21dB masa7(n7) = c7; volt7(b7)= -21; c7=0; n7=n7+1; b7=b7+1; end end x=x+1; end %save file y0=[volt0',masa0']; save('1db.txt','y0','-ascii','-append') %save file for attenuation exceeding 1dB save('1209.txt','y0','-ascii','-append') %save file through month of events save('overall.txt','y0','-ascii','-append') %save all fade duration statistics here y1=[volt1',masa1']; save('3db.txt','y1','-ascii','-append') %save file for attenuation exceeding 3dB save('1209.txt','y1','-ascii','-append') %save file through month of events save('overall.txt','y1','-ascii','-append') %save all fade duration statistics here y2=[volt2',masa2']; save('6db.txt','y2','-ascii','-append') %save file for attenuation exceeding 6dB save('1209.txt','y2','-ascii','-append') %save file through month of events save('overall.txt','y2','-ascii','-append') %save all fade duration statistics here y3=[volt3',masa3']; save('9db.txt','y3','-ascii','-append') %save file for attenuation exceeding 9dB save('1209.txt','y3','-ascii','-append') %save file through month of events save('overall.txt','y3','-ascii','-append') %save all fade duration statistics here y4=[volt4',masa4']; save('12db.txt','y4','-ascii','-append') %save file for attenuation exceeding 12dB save('1209.txt','y4','-ascii','-append') %save file through month of events save('overall.txt','y4','-ascii','-append') %save all fade duration statistics here y5=[volt5',masa5']; save('15db.txt','y5','-ascii','-append') %save file for attenuation exceeding 15dB save('1209.txt','y5','-ascii','-append') %save file through month of events save('overall.txt','y5','-ascii','-append') %save all fade duration statistics here y6=[volt6',masa6']; save('18db.txt','y6','-ascii','-append') %save file for attenuation exceeding 18dB 65 save('1209.txt','y6','-ascii','-append') %save file through month of events save('overall.txt','y6','-ascii','-append') %save all fade duration statistics here y7=[volt7',masa7']; save('21db.txt','y7','-ascii','-append') %save file for attenuation exceeding 21dB save('1209.txt','y7','-ascii','-append') %save file through month of events save('overall.txt','y7','-ascii','-append') %save all fade duration statistics here Calculate k and α value kh=0.263;kv=0.233; ah=0.979;av=0.963; elevation=38.98; t=45; R=120; L=0.30132; r=1; k=(kh+kv+((kh-kv)*(cos (elevation))^2*(cos(2*t))))/2; alfa=((kh*ah)+(kv*av)+(((kh*ah)-(kv*av))*(cos(elevation))^2*(cos(2*t))))/(2*k); Attenuation=k*(R^alfa)*L*r ITU-R Rain Model f=38; %frequency GHz e=0.6803; %elevation angle degress a=26; %attenuation threshold (dB) d0 = 80 * 38.98^(-0.4) * 38^1.4 * 26^(-0.39); %equation (1) sigma = 1.85 * 38^(-0.05) * 26^(-0.027); %equation (2) y = 0.055 * 38^0.65 * 26^(-0.003); %equation (3) p1 = 0.885*y - 0.814; %equation (5) p2 = (-1.05)*y^2 + 2.23*y -1.61; %equation (6) dt = d0 * exp(p1*(sigma^2) + p2*sigma - 0.39); %equation (4) d2 = d0*exp(-(sigma^2)); %equation (7) q1 = (log(dt) - log(d0)) / sigma; q2 = (log(dt) - log(d2)) / sigma; 66 denom1 = (0.661*q1) + (0.339*sqrt((q1^2)+5.51)); %equation (9) Q functions denom1 = denom1*sqrt(2*pi); denom2 = (0.661*q2) + (0.339*sqrt((q2^2)+5.51)); denom2 = denom2*sqrt(2*pi); Q1 = (1/denom1)* exp(-(q1^2) / 2); %equation (9) Q functions Q2 = (1/denom2)* exp(-(q2^2) / 2); k = (1 + ((sqrt(d0*d2) * (1-y) * Q1) / (dt * y * Q2)))^(-1); %equation (8) [masa3,event3] = textread('overall-time.txt','%f %f'); %read measured data total = length(masa3); bil=total; x=0; n=1; while x<bil if masa3(n)<dt %for 1<D<Dt P(n)= masa3(n)^(-(y)); %equation (10) else q3 = (log(masa3(n)) - log(d2)) / sigma; %for D>Dt denom3 = (0.661*q3) + (0.339*sqrt((q3^2)+5.51)); denom3 = denom3*sqrt(2*pi); Q3 = (1/denom3)* exp(-(q3^2) / 2); %Q functions P(n)= (dt^(-(y)))*(Q3/Q2); %equation (11) end F(n)=(1 - (k * ((masa3(n)/dt)^(1-y)))); N(n)=P(n)*event3(n); %total number of fade time(n)=masa3(n); %duration; equation (14) n=n+1; x=x+1; end y1=[time',N']; loglog(time,N) %plot ITU-R rain model Chris J Gibbins and Kevin S Paulson Rain Model R=1.2; td=10; x=0; n=2; td1=10; 67 %the number of fade exceeding threshold,A; equation (9) event(1)=1.7*(10^4)*(R^(-1.76))* exp(-(log(td1)-2)^2/(3.86-(0.0409*R))); masa1(1)=1; while x<400 event(n)=1.7*(10^4)*(R^(-1.76))* exp(-(log(td)-2)^2/(3.86-(0.0409*R))); masa1(n)=td; x=x+1; td=td+(10); n=n+1; end loglog(masa1,event) %plot Gibbins rain model Zsolt Kormanyos/ Lena Pedersen/ Cyril Sagot/ Janos Bito Rain Model atten1 %call k and α functions Ai=1; %Attenuation exceeding threshold in dB A=Attenuation; b=(((Ai/A)^(1/alfa))-1)*log((1+((Ai/A)^(1/alfa)))); a=1-((1-((Ai/A)*(1/alfa)))/(1+(4.56*((Ai/A)^(1.03/alfa))))); P=((((A^(1/alfa))+((k*L*r)^(1/alfa)))/((Ai^(1/alfa))+((k*L*r)^(1/alfa))))^b)*(10^(- 4*a)) Pnew = P*100; [masa2,event2] = textread('overall-time.txt','%f %f'); total2 = length(masa2); bil2=total2; x=0; n=1; while x<bil2 N2(n)=Pnew*event2(n); %number of fade durations time2(n)=masa2(n); %total time durations n=n+1; x=x+1; end loglog(time2,N2) %plot bito rain model 68 Comparison and Analysis %read fade duration statistic data (measured data) [masa, db] = textread('overall-time.txt','%f %f'); itutrydt %call Itu-R model [1] empirical %call gibbins model [2] bito %call bito model [3] figure(2); loglog(masa,db,time2,N2,time,N,masa1,event) %plot log graph legend('measured data','Bito model','Itu-r model','Paulson model') %legends xlabel('Duration [sec]') %x-axis label ylabel('No of events exceeding abscissa') %y-axis label title('Anuual Fade duration in UTM 99 (38GHz)') %graph title

DOCUMENT INFO

Shared By:

Categories:

Tags:

Stats:

views: | 157 |

posted: | 11/29/2011 |

language: | Malay |

pages: | 86 |

OTHER DOCS BY yurtgc548

How are you planning on using Docstoc?
BUSINESS
PERSONAL

By registering with docstoc.com you agree to our
privacy policy and
terms of service, and to receive content and offer notifications.

Docstoc is the premier online destination to start and grow small businesses. It hosts the best quality and widest selection of professional documents (over 20 million) and resources including expert videos, articles and productivity tools to make every small business better.

Search or Browse for any specific document or resource you need for your business. Or explore our curated resources for Starting a Business, Growing a Business or for Professional Development.

Feel free to Contact Us with any questions you might have.