Research On Signal Detection For Generalized Spatial Modulation

Generalized spatial modulation?GSM?overcomes not only the low spectrum efficiency default of the traditional spatial modulation?SM?technology,but also the constraint that the number of transimit antennas has to be a power of two.In GSM systems,multiple antennas are activated at the same time and transmit the different modulation symbols.GSM can get a better tradeoff in the aspect of both deployment cost and spectrum efficiency by adjust the number of activated antennas.Therefore,GSM technology has become one of the most popular candidate technologies for the fifth generation?5G?mobile communication systems.For GSM,the maximum likelihood?ML?detection algorithm is the optimal,which involves the exhaustive search of all possible combination of the activated antennas and transmitted modulation symbols.Its complexity increases exponentially with the number of transmit antennas and modulation levels,so it's difficult to apply in the actual systems.Compared with ML algorithm,the traditional linear detection algorithm has a lower complexity,but its performance also has a significant loss.Due to this reason,the GSM signal detection algorithm was deeply studied in the thesis,and the main contents and innovations are given as follows.1.Introduction to GSM technologies and the related detection algorithmsBased on the introduction to the SM systems and GSM systems,sereval classical low-complexity detection methods are summarized,which include ML,maximum ratio combination?MRC?,normalized MRC?NMRC?for SM system,and linear detection algorithm,lattice reduction,QRD-M for GSM system.The computational complexity and detection performance of these algorithms are simulated and comparied as well.2.Regularized orthogonal matching pursuit?ROMP?detection algorithm for Generalized Spatial ModulationIn the orthogonal matching pursuit?OMP?algorithm,the wrong activated antenna index may be decided.To solve this problem,a low-complexity detection algorithm is proposed,termed as regularized OMP algorithm.The algorithm starts by producing an estimate of the plausible active antennas indices according to correlation between the MIMO channels and the residual,then removes wrong indexes based on the regularization criteria,which lead toshrinking search space.Finally,GSM signal is estimated by solving the least squares problem.Simulation results show that the proposed algorithm can improve the detection performance at the cost of slightly increased complexity compared with the conventional OMP detector,and a better tradeoff between the performance and complexity is achieved.3.Proximal gradient?PG?–based detection algorithm for large-scale multiuser GSM-MIMO systemsFor large-scale multiuser GSM signal detection,the complexity of the ML algorithm is high and the sparse reconstruction algorithm has great performance loss.The proposed PG algorithm transforms the nonconvex ₀l-norm optimization problem into its optimal convex approximation,namely,the least squares problem with ₁l-norm penalty.Then the process of calculating the solution vector is deduced based on the gradient descent method.The simulation results show that the proposed algorithm has certain performance improvement compared with the zero forcing?ZF?algorithm and the compressive sampling matching pursuit?CoSaMP?algorithm based on sparse reconstruction.