With the advent of higher frequency satellite communication systems, there has been an increased demand for algorithms to mitigate propagation induced fades in links as well as for modeling of fades in computer simulations of overall satellite system capacity and performance under such conditions. To test these various algorithms and models, one is often faced to three solutions: an experimental set-up, an analytical model or the use of measured fade levels. The first one tends to be costly and time consuming while the second one tends to raise some doubts about its validity because there are no currently validated analytical models well representing the dynamics of propagation fades. The third one tends also to raise doubts given the limited measurements and the likelihood that the algorithms are only applicable, tuned to the measured fade dynamics.

This paper presents a methodology to generate a time series of propagation fade levels based on a reference time series obtained through beacon measurements or other means. It is not replicating the temporal waveform of the reference time series but instead maintains its attenuation, fade rate and fade duration statistical characteristics.

The proposed methodology is simple as it is based in gathering the statistics of the fade rate within various attenuation level intervals and replicating these using an arbitrarily shaped probability density function random number generator. The results presented showed that not only we replicate the attenuation level statistics of the reference time series but also its fade rate and fade duration statistics. In addition, the methodology allows to generate events not present in the measured time series but statistically probable.

Six measured fade events measured at 30 GHz during the CRC participation to the ACTS propagation measurement campaign were concatenated into a single set of events statistically representative of conditions encountered in Ottawa. Based on these measurements, the fade rate probability density function (PDF) when the fade level is within a 0.1 dB interval is calculated. Using this set of PDF's to drive a random number generator, the fade level at time n (x(n)) is given by the running sum of all previous random fade rates. Results show a very good agreement in fade level, fade rate and fade duration statistics between the generated and measured time series.

The channel state of a satellite RF link is best provided as the Signal to Noise ratio (SNR) estimate of the received signal. This form of channel state is utilised by satellite capacity optimization techniques like Adaptive Coding and Modulation (ACM), which comes under the hood of Fade Mitigation Techniques (FMT). The performance of the SNR estimation algorithm, used as the channel state estimator, directly impacts the efficiency of the FMT. In this paper we analyse the contemporary SNR estimation algorithms considering only the rain attenuation as the fading process.α Their performances are quantified in terms of the number of received symbols needed to obtain an estimate with a given error margin. Their suitability as a channel state estimator for a typical DVB type satellite system is analysed by considering the various assumptions involved in the algorithm, the effect of noise due to interference as encountered in such systems on the estimation algorithm. Index terms - Fade Mitigation Techniques, SNR Estimation Algorithm, DVB type satellite system.

Interference, arising from a reason or another, may be harmful to the communication receiver. There exist a vast number of papers considering interference mitigation techniques. The purpose of this paper is to offer an overview of those methods. In particular, the paper considers interference mitigation techniques usable for single antenna receivers and, on the other hand, adaptive antenna receivers. The major difference between the two receiver types is that with single antenna techniques only narrowband (with respect to the desired signal's bandwidth) interference can be mitigated whereas adaptive antenna techniques can be used to suppress also wideband (with respect to the desired signal's bandwidth) interference. The most usable single antenna techniques are notch or whitening filters. A notch filter has been implemented and tested for GPS, and observed to have an expected performance, i.e., narrowband interference can be successfully mitigated but wideband not. Recent development of notch filters has yield to iterative notch filters, which offer an improved performance through iteration. Notch filters use different means to detect the interference. A recent algorithm is the consecutive mean exciser (CME), of which several versions exist. These can detect interference up to 40 % of the desired signal's bandwidth. However, some versions can, in some extreme cases, detect interference up to 80 % bandwidth. The (iterative) notch filters can also be used to detect impulsive interference, that might be harmful, e.g., to the automatic gain control unit. The traditional whitening filter's (LMS or RLS) performance may be improved using a robust version of it. Adaptive antenna techniques are actually different beam forming methods. They include such methods than Bartlett, null steering, Capon, power inversion (whitening). The paper compares requirements and interference mitigation performance of these methods. E.g., both Bartlett and Capon require direction-of-arrival (DOA) information of the desired signal but the Bartlett is not very efficient against interference. Null steering method requires also DOA information of interfering signal(s). Power inversion is efficient against interference, does not require DOA information, but, on the other hand, does not offer antenna array gain (reduced sensitivity) and have problems with high SNR desired signals. The beam former methods requiring DOA information (steering vector) are sensitive to errors on it and also to other calibration type errors. The paper will discuss also robust beam forming methods which are less sensitive to calibration errors. These include a diagonal loading method. Finally the paper will address also beam formers that do not require DOA information, but need training to adjust the beam forming coefficients.

Broadband fixed wireless access (BFWA) has been recognised as an effective first kilometre solution for delivering broadband services to residential and business customers. BFWA can also function as a backbone for lower capacity systems (both wire line and wireless) such as WLAN, WiMAX and 2/3-G base stations. The large bandwidths available above 20 GHz make radio systems with very high capacities possible. Users can be offered bit rates in the order of several hundred Mbit/s. Such radio links can, in many cases, be an alternative to optical fibre in terms of capacity.

This article presents a space-time dynamic channel model for BFWA operating above 20 GHz. The developed channel model represents the time varying wideband channel within an antenna sector for a number of simultaneous users. BFWA systems are affected by several degradation factors, including multipath propagation, rain and vegetation attenuations as well as scintillation effects. A combined dynamic channel model for BFWA systems which includes these effects is reported in [1]. In this paper the main focus is on space - time rain attenuation correlation for the different users in a sector.

Rain causes absorption and scattering of radio waves, and the resulting attenuation shows a considerable temporal and spatial variability. It is frequently observed that during a shower, high intensity rain is localised in a very small area surrounded by a region of more uniform, low intensity rain. Figure 1 shows a wireless access system served by four 90 degree base station sectors subjected to rain precipitation in the coverage area. In this scenario, information on space-time correlation of rain attenuation can be used by the system to select the most suitable base station for the users. Such channel information is also required when optimising the radio resource management (RRM) within a single sector to obtain high traffic throughput with given quality of service (QoS) constraints. The RRM adapts for example modulation and coding, and access to time- frequency slots, to improve the overall system performance. If providing layered multicast services, space-time channel knowledge may improve the system performance as well.

In this paper the space-time correlation of rain attenuation within a sector is studied based on an extension of the Maseng-Bakken rain attenuation model, as reported in [2] for slant paths. A large star network database of rain attenuation measurements at 42.1 GHz is available for testing purposes. In addition to providing information for route diversity, the channel model is suitable for studying advanced RRM techniques including adaptive coding and modulation as well as interference cancellation, optimising the traffic throughput within an antenna sector.

BFWA, Sector RRM, Spatial dynamic channel model, Rain attenuation

References:

[1] M. Cheffena, L. E. Bråten, T. Tjelta and T. Ekman, "Time dynamic channel model for broadband fixed wireless access systems," 15th IST Mobile & Wireless Communications Summit, Myconos, 4 – 8 June, 2006.

[2] B. Gremont and M. Fukuo, "Spatio-temporal rain attenuation model for application to fade mitigation techniques," IEEE Trans. Ant. Prop., vol. 52, pp. 1245-1256, May 2004.

Mobile Satellite Systems (MSS) are becoming an increasingly important part of the global communication infrastructure, cooperating with and complementing the existing terrestrial facilities. As the demand for traffic to be routed through these networks increases, it is necessary to devise techniques able to extend system capacity. In satellite networks, a possible solution consists in reducing the frequency reuse factor. However, this strategy results in an increasing interference among different users in the uplink segment.

In fact, the signal transmitted by each user is received not only through the reference beam, but also through a number of adjacent beams, according to the sidelobes in the antenna pattern. By exploiting the knowledge of the antenna pattern at the gateway, the interbeam interference can be reduced by adopting Multi User Detection (MUD) techniques.

In the recent years, great interest has been devoted to this topic, and a plethora of techniques has been devised. However, their practical application still has to be proved, as they are typically complex and prone to estimation errors and system non idealities, such as non-linear distortion. In this framework, the goal of this paper is to assess the applicability of MUD techniques to multi-beam mobile satellite systems based on the GMR-1 standard. In particular, spatial MMSE (SMMSE) combining with turbo interference cancellation (SMMSE-IC) techniques are applied at the gateway of the GMR-1 system. Physical layer simulations are used to prove the robustness of the proposed techniques, under realistic propagation conditions including independent Ricean fading for each user and non linear propagation effects, and taking into account the effect of non-ideal parameter estimation.

In order to assess the benefits provided by the use of MUD techniques, full frequency reuse is considered, for 3 co-channel users located in adjacent cells. The corresponding matrix of normalized channel gains is shown in the table. The figure reports the BER performance of Turbo SMMSE-IC in TCH6 GMR-1 transmission for each detected user at iteration 1 and 2, respectively. These results are obtained considering a Rice fading channel with Rice factor 10 dB and user terminal speed equal to 50 Km/h. The figure clearly illustrates the performance gain due to the introduction of Turbo SMMSE-IC techniques. Specifically, the highest gain is experienced by user 2, where the traditional receiver has a performance floor at BER = 0.3. Turbo SMMSE-IC successfully removes the floor at first iteration and obtains a further significant gain after the cancellation iteration. To further validate the robustness of SMMSE techniques in the considered system, non-linear distortion induced by the mobile equipment power amplifier and by estimation errors at the receiver have been considered. The corresponding results are reported in the full paper, showing the robustness of the proposed techniques.

Systems operating at high frequency bands are submitted to strong propagation impairments due to the presence of oxygen, water vapour, liquid water (in clouds and rain) in the troposphere. To cope with such high propagation impairments without limiting too much system availability, Fade Mitigation Techniques (FMTs), are necessary.

Various methods exist to counteract propagation effects, called Fade Mitigation techniques (FMT). The relevant FMTs should take into account operating frequency bands, system performance objectives and the main design parameters of the system architecture (network organisation, type of communication payload, multiple access schemes, ). FMTs can be divided into:

- Power Control: transmitting power level fitted to propagation impairments,
- Adaptive waveform: compensating fade by a more efficient modulation and coding scheme,
- Diversity: avoiding fade by the use of another less impaired link,
- Layer 2: coping with the temporal dynamics of the fade.

FMT at layer 2 level are techniques which do not aim at mitigating a fade event but instead rely on the re-transmission of the message. Two different techniques can be envisaged at layer 2: Automatic Repeat Request (ARQ) and time diversity. With ARQ, the message is resent until the message reaches successfully the receiver. Time diversity aims to re-send the information when the state of the propagation channel allows to get through. This technique benefits from the use of propagation mid-term prediction model in order to estimate the most appropriate time to re-sent the message without repeating the request. This way, the TD technique is suitable for specific applications like file transfer or data transfer process that does not require a real time communication. In these cases, it is assumed that the user can wait for the end of the fade event before receiving the requested information.

Regarding time diversity, no model extensively validated has been proposed up to now in the literature to estimate the performance of time diversity. The aim of this paper is to present a new prediction model allowing time diversity performances to be assessed, which relies on a bi-dimensional joint log-normal distribution. Then, experimental statistics generated from Olympus and Italsat data collected in Belgium and Italy are presented before to carry out a testing activity of the proposed model.

This study has been carried out in the framework of the ESA study n°17760/03/NL/JA: Characterisation and modelling of propagation effects in 20-50 GHz band.

Time diversity is a promising rain attenuation countermeasure in satellite broadcasting systems at Ka band (20-30 GHz). If a digital stream, such as a television or radio broadcast, is transmitted twice during a rain attenuation event, with a suitable delay between the two transmissions, then the receiver can accumulate and use the information from the times with the best receiving conditions. A first physical and statistical assessment of the performance of a time diversity system can be obtained by measuring or predicting the diversity gain, for a given rain attenuation, as a function of time delay. To be statistically reliable, the diversity gain should be measured for a very long time, a very expensive activity, or could be simulated with a very good rain attenuation time series simulator. We present long-term experimental results obtained from rain attenuation time series collected at Ka Band in a 37.7° slant path to Italsat, at Spino d’Adda (Italy) in the years 1994-2000. We have calculated the time diversity gain for several rain attenuation thresholds and time delays up to 60 minutes, and we have compared the results with the corresponding statistics predicted for the same radio link by the Synthetic Storm Technique, reported in a companion paper, and found good agreement.

The site diversity system has been proposed as a mitigating technique for the rain attenuation on satellite links using frequencies above about 10 GHz. In this system, signals of two earth stations located apart each other are switched. Although two prediction methods (Boithias method [1] and Hodge method [2]) for site diversity improvement are recommended as Rec. ITU-R P.618 [3], these methods do not consider regional climatic parameters that will affect diversity improvement. These methods may be further improved in prediction accuracy by considering the regional climatic parameters. A modified version of the existing prediction methods were investigated in this paper, using KIT (Kitami Institute of Technology) satellite link site diversity databank for 108 datasets in 12 countries and regional climatic parameters.

Boithias method can predict diversity improvement factor defined as the ratio of single-site time percentage to the diversity time percentage at the same attenuation level. The prediction parameter β is derived from the data for frequencies of 10�`20 GHz (mostly 11�`13.6 GHz) and elevation angles above 10 deg., and depends only on the distance between two earth stations [1]. On the other hand, Hodge method can predict diversity gain as the product of component gains which mean the contribution by site separation, frequency, elevation angle, and base line orientation, respectively [2].

Based on our research, above two methods proved to be improved by using the regional climatic parameters such as the thunderstorm ratio and so on. The thunderstorm ratio is defined as the ratio of annual total rainfall arising from thunderstorm (convective rain) to the annual total rainfall. In this paper, we used the value of the thunderstorm ratio given by the equation proposed by Dutton et al. [4].

In the first place, the regional dependence of β in Boithias method was investigated, and a prediction equation of β was obtained using parameters such as site separation, frequency, elevation angle, base line orientation, latitude and longitude (modified Boithias method).

In the second place, the inclusion of an additional component gain by the thunderstorm ratio into Hodge method was investigated, and a modified Hodge method was obtained.

It is proved that the modified Boithias method have a better prediction accuracy than the existing Boithias method. For modified Hodge method, a similar result is obtained. When two modified prediction methods are compared, it is found that modified Boithias method has better accuracy below 0.01% of single-site time percentage and modified Hodge method has better accuracy above 0.01%. In conclusion, it is verified that the modified methods proposed in this paper have better accuracy than existing methods by using the regional climatic parameters such as the thunderstorm ratio and so on.

[1] L. Boithias, �gRadio Wave Propagation,�h North Oxford Academic, 1987. [2] D. B. Hodge, Radio Science, Vol.17, No.6, pp.1393-1399, 1982. [3] ITU-R, Rec. ITU-R P.618-8, ITU, Geneva, 2003. [4] E. J. Dutton et al, NTIS Rep. ACC-ACO-16-74, 1974.

Time diversity is a promising rain attenuation countermeasure in satellite broadcasting systems at Ka band (20ƒ{30 GHz). If a digital stream, such as a television or radio broadcast, is transmitted twice during a rain attenuation event, with a suitable delay between the two transmissions, then the receiver can accumulate and use the information from the times with the best receiving conditions. The size of the delay must be designed to best suit the prevailing weather conditions, time of day, diurnal variations of rain attenuation, and link power budget.

A first physical and statistical assessment of the performance of a time diversity system can be obtained by measuring or predicting the diversity gain, for a given rain attenuation, as a function of time delay. To be statistically reliable, the diversity gain should be measured for a very long time, a very expensive activity, or could be simulated with a very good rain attenuation time series simulator.

To provide a reliable formula of the gain we have simulated rain attenuation time series in a 37.7 degs slant path (Italsat), at Spino d'Adda (Italy) from a large data bank of rain rate time series spanning the years 1994-2000 by applying the Synthetic Storm Technique, a reliable rain attenuation prediction tool that yields results that can be considered as experimental. The Synthetic Storm Technique can generate a rain attenuation time series at any frequency and polarisation, and for any slant path above about 10 degs from a rain rate time series measured with a rain gauge. The paper will report a general formula of the gain (dB) as a function of time delay (1 min to 60 min), carrier frequency (10 to 100 GHz) and rain attenuation (0 to 30 dB).

In the Ka- and V-bands, attenuation caused by rain is too severe to be accounted for by a fixed margin in the link budget. Because of this, Fade Mitigation Techniques ('FMTs') are developed, which compensate for rain attenuation by adaptively improving the quality of the link only when the signals are degraded.

For the design of FMT systems, knowledge of not only the depth of fades to be expected is essential, but also of the rapidity at which these can change. In order to provide the necessary information for FMTs, a study of the dynamics of rain attenuation has led to the development of the 'Two-Sample Model'.

The Two-Sample Model predicts the probability distribution of rain attenuation a short time after a measured value, dependent on the values of two previous samples of rain attenuation. This model can be used in the design of FMT systems, to determine the required tracking speed, and can be implemented in the FMTs themselves, to predict the probability of fades in the very near future. Furthermore, it can be used for data simulation models of rain attenuation, useful for channel modelling, to test FMTs.

Because the Two-Sample Model uses two previous samples, when used for short-term prediction of rain attenuation it predicts the attenuation distribution more accurately than similar models which use only one previous sample value. When used in simulation of time-series data, the result of the Two-Sample Model represents better the dynamic behaviour of a rain attenuation event.

The Two-Sample Model has three parameters, which characterise the dynamic properties of the rain attenuation event, and depend on wind speed and the type of rain. The parameters valid for a particular link can be calculated from time series of attenuation measured on the same link, which allows their evaluation in different environments and climates, and during different wind and rain conditions. However, for many climatic configurations, propagation measurements are not available.

Because of this, theoretical expressions have been derived to calculate the Two-Sample-Model parameters from rain rate measurements. The expressions depend on wind speed, wind direction, vertical air velocity and rain height, and on the elevation and azimuth angles and the frequency of the link. The theoretical expressions have been compared to measurements of attenuation on a satellite link in the UK, and give good results on the long term. The verification of the dependence on meteorological parameters is limited by the resolution of the meteorological data used.

With these expressions, the Two-Sample-Model parameters for different configurations can be derived using only rain rate measurements and other meteorological data. This will provide a great opportunity for the assessment of the model parameters on many sites on earth where expensive satellite link measurements have not been performed, which will facilitate the design of FMTs for these sites. Furthermore, the results enable a wide study of the climatic dependence of the dynamic properties of rain attenuation, which can lead to the development of a global dynamic rain attenuation model.