Real Sphere Decoder Algorithm for Signal Detection in Multiple-input Multiple-output Power Line Communication

A scheme of real sphere decoder for signal detection in multiple-input multiple-output power line communication was studied. Firstly, based on the Cholesky decomposition, the problem of searching lattice in a sphere packing of the received signal space was transformed into the problem of searching the lattice in the corresponding ellipsoid in the transmitted signal space. Then, according to the inequality induced by the upper triangular matrix, the boundary of the ellipsoid was determined, and the corresponding integer component was formed in order to span the candidate lattice subset defined by the transmitter ports. Hereafter, the algorithm iteratively tested the lattice points within the sphere packing in the constantly updated interval, and outputted the nearest lattice point as the approximately optimal estimation for the transmitted signal. The simulation results show that the proposed scheme is superior to the zero forcing and the minimum mean square error detector in terms of the mitigation of multi-stream interference whether or not the common mode (CM) reception is adopted, and its complexity is the polynomial function of the number of ports when the signal to noise ratio is higher.