A Metaheuristic Approach to Parallel Inversion of Large Random Matrices :: Precoding/Decoding for Large Scale MIMO Systems Using Particle Swarm Optimization

A novel method is presented for the inversion of large random channel matrices that arise in large scale MIMO systems. The proposed method eliminates the need for forming the pseudoinverse altogether. It is based on the idea of inverse vectors in which each column in the channel matrix is viewed as a particle in a swarm that seeks its own solution, the inverse vector. Such an inverse vector has to satisfy two constraints. Firstly, it has to be in the null-space of all the remaining columns. Secondly, it has to form a projection of value equal to one in the direction of selected column. These constraints open up an avenue for metaheuristic methods to enter the realm of large random matrix inversion. A case for Particle Swarm Optimization has been presented which when employed to demonstrate the new approach yields much better performance in terms of residue and computation time. But the opportunity is still there for rest of the methods. Enter the realm of large scale MIMO systems by trying one of your own!