Development, Modelling and Simulation of Advanced Techniques to Estimate and Improve the Performance of Satellite Telecommunications Systems in Non-Linear Channels

Abstract. Non-linear amplifiers that are used near saturation in satel-lites, cause severe distortions of the transmitted signal. This dissertation proposes novel signal processing techniques that counteract these distor-tions to improve performance. Firstly, the fact that the non-linearities in-troduced by memoryless bandpass amplifiers preserve the symmetries of theM-ary Quadrature Amplitude Modulation (M-QAM) constellation is exploited, to present a Cluster-Based Sequence Equalizer (CBSE). The proposed equalizer exhibits enhanced performance compared to other techniques, including Volterra equalizers and neural network equaliz-ers. Moreover, this gain in performance is obtained at a substantially lower computational cost. Secondly, a new constellation shaping tech-nique, which efficiently and effectively reduces the Peak to Average Power Ratio (PAPR) of Orthogonal Frequency Division Multiplexing (OFDM) systems, is proposed. The proposed technique requires minimal imple-mentation complexity, while it offers considerable performance gains. Closed form analytical expressions for the distribution of the PAPR and the Bit Error Rate (BER) are derived and their accuracy is verified via simulations. Finally, a detailed comparative study of two single-carrier frequency-division multiple access schemes is presented, namely localized FDMA scheme (LFDMA) and interleaved FDMA scheme (IFDMA), ver-sus orthogonal scheme (OFDMA), for a satellite up-link based on the dig-ital video broadcasting via satellite (DVB-S) standard. Considering two state-of-the-art high power amplifiers, operating in the K- and S-bands, the performance of synchronous and asynchronous LFDMA, IFDMA and OFDMA is evaluated in a multi-user environment including inter-block interference