Efficient Genetic-Based Detection Algorithm for Large-Scale MIMO Systems

In this paper, a genetic-based signal detection algorithm is proposed for large-scale multi-input multi-output (MIMO) systems. First of all, the random Gaussian noise is utilized to serve for the population initialization in Genetic algorithm (GA), where candidates in the population can be easily generated. Then, the Euclidean distance $\Vert \mathrm{y}-\mathrm{H}\mathrm{x}\Vert$ is applied as the fitness function for the candidate selection. After that, two-point recombination as well as random mutation is introduced for the following evolution, thus completing an iteration of the proposed algorithm. Meanwhile, a flexible trade-off is established between detection performance and complexity, which can be adjusted by the population size and the iteration numbers. Furthermore, a pre-detection stage that relies on decoding radius is also proposed for the efficient detection without any performance loss. Finally, simulation results confirm that considerable performance gain can be achieved in a low complexity cost.