Evolutionary algorithms for near-optimum detection of multi-beam satellite signals

Multi-beam satellites represent one of the enabling technologies for future terabit satellite systems. It is known that, increasing the reuse factor of frequency sub-bands, it is possible to boost multi-beam satellite capacity, provided that co-channel interference is conveniently reduced at the Earth station side. To this aim, suitable multi-user detection techniques are required. The optimum detection is based on the maximum-likelihood (ML) criterion, which involves a prohibitive computational burden for large reuse factors. Alternative state-of-the-art suboptimal solutions are based on Minimum Mean Square Error (MMSE) detection and iterative interference cancellation. In this work, we aim at testing near-optimum multi-beam detection techniques based on evolutionary algorithms, namely: Genetic Algorithms (GAs) and Particle Swarm Optimization (PSOs), both them well suited for solving complex nonlinear optimization problems with tolerable computational complexity. In particular, we sha...

Multi-beam satellites represent one of the enabling technologies for future terabit satellite systems. It is known that, increasing the reuse factor of frequency sub-bands, it is possible to boost multi-beam satellite capacity, provided that co-channel interference is conveniently reduced at the Earth station side. To this aim, suitable multi-user detection techniques are required. The optimum detection is based on the maximum-likelihood (ML) criterion, which involves a prohibitive computational burden for large reuse factors. Alternative state-of-the-art suboptimal solutions are based on Minimum Mean Square Error (MMSE) detection and iterative interference cancellation. In this work, we aim at testing near-optimum multi-beam detection techniques based on evolutionary algorithms, namely: Genetic Algorithms (GAs) and Particle Swarm Optimization (PSOs), both them well suited for solving complex nonlinear optimization problems with tolerable computational complexity. In particular, we shall consider GA and PSO-assisted maximum likelihood detection, thus exploiting the capability of GA and PSO of finding near-optimum ML solution within a search space of reasonable cardinality. A simulation testbed will be implemented by considering a multi-beam Single Feed per Beam Antenna System (SFBA) with combined transmission and reception antennas. Results obtained by evolutionary-aided ML detection will be compared with those yielded suboptimal MMSE multi-beam detection and with the single-user bound. Simulation results show the near-optimal potential of the proposed evolutionary techniques, in particular when the theoretical ML detection exhibits unaffordable computational burden.