Satellite communication systems at millimetre wave frequencies can be affected by huge rain attenuation. To design them reliably, we need to know both first order statistics (probability distribution of exceeding rain attenuation), and second order statistics (fade duration, rate of change, diurnal variation and so forth). To this end, the availability of reliable rain attenuation time series is the most important information a designer can have. Experimental attenuation time series, however, may not be available and, moreover, when available, they have been recorded only for specific sites, frequencies and elevation angles, and their scaling to other sites is not straight. Instead of using real data collected from expensive and long propagation experiments, the alternative is to generate synthetic rain attenuation.

The synthetic storm technique is a powerful tool that can produce all the necessary first and second order statistics due to rain attenuation. Starting from rain rate time series, collected at a site with a rain gauge, the Synthetic Storm Technique can generate rain attenuation time series at any frequency and polarisation, and for any slant path above about 10°, as long as the hypothesis of isotropy of the rainfall spatial field holds, in the long term.

We have studied how the rain attenuation A (dB) predicted in a slant path of rainy path length L, by the Synthetic Storm Technique, is built up by parts deltaA (dB) due to partial rainy path lengths, deltaL, of the total rainy L. We have found the ratio deltaA/A can be estimated from the ratio deltaL/L independently of the value of A and L, and that this relationship leads to a differential equation from which we can reliably estimate: (a) time series of rain attenuation from rain rate time series measured at a point (with a rain gauge), without applying directly the Synthetic Storm Technique, and (b) the probability distribution of rain attenuation from rain rate probability distribution (i.e., no knowledge of rain rate time series is required). In both cases the predictions are as accurate and reliable as those directly derived by applying the full Synthetic Storm Technique, but in case (b) we do not need to know the rain rate time series so that we can apply the global ITU-R rain maps.

The atmospheric depolarization at microwaves is due to the non spherical shape of hydrometeors, namely raindrops and ice particles. Whereas at Ku band the contribution due to rain is predominant, for Ka and V frequency bands the effect of ice particles becomes more and more relevant. The two classes of particles being characterised by different dynamical and propagation properties the impact on system performances is quite different. Rainfall induced depolarization is obviously linked to rain attenuation and relevant depolarization is experienced in presence of deep fades. As well rain depolarization dynamic is very similar to rain attenuation dynamic. On the other hand ice particles can be present in atmospheric clouds even without concurrent rain. Furthermore, the ice particles being much smaller than raindrops and being affected by the electrostatic fields inside the clouds, the ice depolarization is affected by higher dynamic than rain depolarization. The modelling of atmospheric depolarization can be done either by a statistical model of the relationship between crosspolar discrimination and rain attenuation, derived from crosspolar discrimination measurements, or using the so-called quasi-physical parameters of depolarization (i.e. anisotropy and canting angle for each family of particles), derived from inversion of measurements based on a polarization switching beacon. The former approach is more directly applicable for the design of systems operating at frequencies, polarizations, elevation angles and climatological conditions whereas the latter permits a physical based scaling of the depolarization for system parameters different from the original experimental ones. Nevertheless the availability of full-transfer matrix measurements at Ka band and V bands is almost limited to specific European stations receiving the OLYMPUS and ITALSAT propagation beacons.

This paper presents the results of a reanalysis and modelling of OLYMPUS and ITALSAT measurements carried out in Belgium and Italy respectively, including both new model development for the statistical modelling and the quasi physical parameters approach. The availability of measurements at different elevation angles, frequencies and climatological conditions makes feasible the assessment of the accuracy of each proposed methodology.

Earth-Satellite meteorological conditions along the full tropospheric propagation path have an important impact on the reliability of microwave communication systems. Such propagation impairments increase with frequency, such as attenuation and scintillation intensity. While it is clear that fade mitigation techniques must be implemented to provide adequate quality of service at higher frequencies, attenuation is not the only impairment to be taken into account. Issues such as frequency attenuation scaling can be a source of uncertainty in some fade mitigation systems. These scaling variations are, in rain conditions, due to the unpredictable shape and size of hydrometeors found on the path.

Important experimental information about this "microstructure" channel property can be inferred by the phase balanced complex crosspolar channel discrimination (XPD) measurements using satellite beacons. Such an experiment is being made since two years apart at our site: city of Aveiro, Portugal. The measurements are being made at a 38° elevation path using a beacon at approximately 20<.

The depolarization is caused by non spherical rain drops and ice clouds found on the path. Ice depolarization has been subject of several publications often dealing with the most spectacular XPD changes during thunderstorms/lightning conditions. The ice formation was linked to the presence of ice nuclei that could be more abundant close to the sea (as is our case). Ice depolarization measurements are however scarce and modeling difficult due to a lack of correlation with any ground measurements.

During these two years of continuous measurements (more than 97%) availability an important (ice and rain) depolarization database was collected. We observed that ice depolarization is, at least at our site, not an exceptional occurrence in rainy weather conditions but it is a rather frequent one in different rain regimes and through all the year. Ice is seen to be present and dominate depolarization phenomena with longer duration depolarizing events than rain events. Such observation seems to confirm suspicions obtained during a six month data collection period using Olympus satellite. It must be pointed out that ice XPD gets worse with increasing frequency and some times ice occurs with negligible attenuation.

In the paper we present:

The use of frequencies above 20 GHz for Earth-Space communication has many advantages, primarily substantially increased capacity. Unfortunately at these frequencies the attenuating effects of the troposphere become very large; clouds, rain and atmospheric gases all contribute significant attenuation. The Italsat propagation experiment demonstrated the large fade margins that would be required to achieve a high availability network (e.g., Polonio and Riva, 1998). At the University of Bath we have developed a technique that is able to forecast link fades (in terms of cloud attenuation, rain attenuation and the attenuation from water vapour and gaseous oxygen and an estimate of the scintillation variance) a number of days into the future. This technique exploits concepts from numerical weather prediction (NWP) systems in order to characterise the meteorological environment. A propagation model is then applied to forecast the fade level. Previously we have demonstrated that this approach can generate spatially consistent timeseries that display the correct 1st and 2nd order statistics in terms of attenuation cumulative distribution function and conditional fade slope probabilities [1,2]. In this paper we will present some initial results from a large scale comparison study between our model and beacon measurements. We will present comparisons with measurements from the 20.7 GHz GBS beacon measured at the Rutherford Appleton Laboratory facility at Chilbolton and the 39.6 GHz Italsat beacon measurements made at Spino D'Adda by the Politecnico Di Milano. The paper will cover the fundamentals of the modelling approach and an assessment of the forecast capability of the system for various attenuation levels by comparing the forecasts with the beacon databases. References 1. D. D. Hodges, R. J. Watson, and G. Wyman, "An attenuation time series model for propagation forecasting", IEEE Trans.Antennas Propagat., to appear, June 2006 2. R. J. Watson and D. D. Hodges, "A real-time propagation forecasting system using numerical weather predictions and radar measurements", IEEE AP-S International Symposium, 2005.

DGA decided to use the EHF payload of Syracuse 3 to carry out an EHF propagation experiment in order to improve propagation knowledge at these frequencies and to prepare future military European telecommunication systems.

Syracuse 3 is an operational military satellite which has been launched in October 2005 and positioned at 47° E. It comprises a powerful communication payload operating in two frequency bands: SHF with four spotbeams and nine 40 MHz bandwidth channels; EHF with two spotbeams and six 40 MHz bandwidth channels.

The EHF propagation experiment will be carried out with a 20 GHz beacon and a transparent repeater (44GHz uplink, 20 GHz downlink) during three years. The 44/20 GHz repeater performs a frequency conversion of a 40 MHz channel from the receive 44-45 GHz band to the transmit 20.2-21.2 GHz band. The signal is then amplified and is combined with the 20 GHz beacon signal and is finally transmitted by the 20 GHz antenna. The EHF spot covers the French metropolitan territory.

The experiment uses three ground stations : M1, located at CELAR near Rennes in Western France, M3 located near Chartres in the Center of France and M4 near Carcassonne in the South of France. The three stations will transmit simultaneously an EHF signal and will receive the three signals from their own station and from the two other ones . The data will be recorded at a 1 Hz sampling rate. With an about 17° elevation angle, the dynamic range will be greater than 30 dB for the 20 GHz and 44/20 GHz attenuation measurements.

Radiowave measurements will be associated with local meteorological measurements or meteorological data provided by Meteo France

On the CELAR site, a dual frequency radiometer will provide the 0 dB reference level and the liquid and vapour water contents in order to calculate gaseous and cloud attenuation. A dual-beam infrared spectropluviometer is installed near the earth station. It wil provide data about the fine structure of rain (size, fall velocity and time distributions). These measurements will be very useful to study the effects of the microphysical structure of rain on a short term frequency scaling.

The first results of the measurements are presented. The objectives of the experiment are to obtain attenuation statistics for the operational Syracuse sites, to have measured data to study propagation effects, to evaluate the effects on system performances, to derive information on the dynamic behaviour of the propagation channel in order to design Fade Mitigation Techniques.

The purpose of this paper is to give a presentation of the propagation experiment with the Syracuse 3 satellite. Modelisation work using the data for separation of atmospheric effects on attenuation and comparison of rain attenuation models will be described in another paper ( 313846) presented by CETP during the EUCAP 2006 conference.

This contribution reports the main achievements gained during the Italsat campaign as for depolarisation. The Italian Italsat receivers at Spino d'Adda and Pomezia allowed a full matrix identification of the dual-polarisations transfer channel at 49.5 GHz during a long campaign in which the satellite transmitted, switched (at 933 Hz) between vertical and horizontal polarization. The Earth stations were equipped with two coherent receivers which measured the complex amplitudes of the co- and cross-polar signals. Due to the phase memory of the receivers, also the relative amplitudes and phases of the co-polar signals were measured even if transmitted in different time intervals; this quantity, known as co-polar unbalance, when worked out jointly with the two depolarisation ratios, allowed the full-matrix identification.

The data editing consisted in the recording of the 4 complex terms of the transfer matrix after forcing the two terms of the secondary diagonal to coincide (full compensation of the satellite effects, including the non-ideal antenna polarization purity and the missing phase equalization between co-polar transmitted signals). The copolar attenuation CPA was assumed as 'independent conditioning variable' and the following quantities have been calculated as 'dependent variables':

conditional quasi-physical parameters (anisotropy and canting angle).

The data points in the scatter plot of Figure 1 show the combined effect of rain and ice. In principle, it would correspond on the plot to a quite definite peak at about 0 dB attenuation but, in practice, as it is clear shown in the figure this is not the case. The reason of this behaviour is explained and it brings to the possibility to mathematically integrate any joint CPA-XPD distribution over an 'equi-performance' domain in order to get the total degradation due to both attenuation and XPD.

This model has been compared with the current models in the literature.

Figure 1 - Scatter plot of XPD versus CPA (ITALSAT experiment)

Satellite communication networks play an important role on the increasing demand of broadband communication services. Over the last years, the spectral congestion in lower frequency bands as well as the increasing demand for more bandwidth imposed by applications with high data rate requirements has led to the utilization of higher frequency bands, namely the Ku (12/14GHz), the Ka band (20/30GHz) and the V band (40/50GHz). However, radio transmission above 10GHz suffers from severe propagation impairments due to various atmospheric phenomena that result in significant BER degradation [Panagopoulos et al, IEEE Comm. Surveys and Tutorials, 3rd Quarter]. In these frequency bands, the performance of the operating satellite communication networks is mainly aggravated due to the induced severe rain attenuation, rain and ice crystals depolarization on the propagation slant path. The depolarization phenomena affect, of course, dual polarized channels, which are usually employed to double the transmitted capacity. The purpose of this paper is the presentation of a general method for the prediction of the outage statistics and error performance analysis of a dual polarized fixed satellite system using Weibull distribution.

The most accepted models for describing point rainfall rate statistics and rain attenuation data are the lognormal and gamma distributions. The lognormal model provides a good approximation in regions of low rainfall rates, while the gamma model is more accurate in regions of high rainfall rates. On the other hand, the Weibull distribution provides a very good approximation for both high and low rainfall/attenuation values [Panagopoulos et al, IEEE Trans. Antennas Prop., 53 (7),pp. 2307-2313, Jul 2005]. The above statements have been extensively tested with the recently released ITU-R rainmaps and ITU-R databank of Study Group 3. Due to lack of available experimental data referring to dual polarized broadband satellite links, the presented numerical results are concentrated on the impact of the polarization tilt angle, the elevation angle, the antenna cross-polarization discrimination XPD and frequency of operation on the outage probability. To quantify the performance deterioration occurred by the employment of dual polarization as a method to increase the capacity, Dual-Polarization Aggravation (DPA) is also introduced. Some qualitative remarks are deduced for the reliable design of modern satellite communication services.

Propagation through the Earth’s atmosphere has a major impact on system design, and the various propagation effects, such as attenuation and scintillation, increase in importance compared with lower frequency bands, requiring a high degree of accuracy and comprehensiveness in their prediction, to encompass all possible propagation impairments, the results of which can be critical in the assessment of system feasibility.

Fast fluctuations of received carrier signal-levels on earth-space links, known as amplitude and phase scintillations, are characterised by random successive fades and enhancements associated small-scale refractive-index inhomogeneities within the atmospheric volume intercepted by the Fresnel ellipsoid of microwave links. Above a few GHz, tropospheric turbulence becomes the prevailing source of scintillation, while the contribution of the ionosphere is no more noticed. Amplitude scintillation effects may reduce the effectiveness of beacon-assisted antenna tracking systems and other open-loop fade mitigation techniques.

The aim of this work is to predict scintillation all around the world by using radiosonde data. The classical radiosonde data from British Atmospheric Data Centre (BADC) are processed by interpolation so that the data are available at equidistant height intervals of 50 m. The computation of the expected value of C_n^² is done for each layer. It can be calculated for one or several months, a season or a year. The outputs of the software include the mean of the expected value of C_n^², the median and the standard deviation. The scintillation parameters m and s [1] are estimated by taking the frequency, the elevation angle, the antenna efficiency, the physical diameter and the estimated values of C_n^². The output parameters allow the prediction of long-term scintillation statistics. A problem has however to be solved: C_n^² is calculated by using the profiles of physical parameters: potential temperature and its derivate, humidity and its derivate, as well as the derivate of directional wind-speeds. Taking the classical two point derivative increases the effect of the noise present in the measured radiosonde data and can present extremely high values influencing very much the cumulative statistics of scintillation. The same problem is present in high resolution radiosoundings. The influence of the various parameters and of the derivation method used will be presented in the final paper. A comparison between the use of high and low resolution radiosoundings for the prediction of scintillation will also be presented.

[1] ITU-R Recommendation P.618-8, "Propagation data and prediction methods required for the design of Earth-space telecommunication systems", Recommendations of the ITU, International Telecommunications Union, ITU-R, 2003.