Relay selection in multi-user amplify-forward wireless relay networks

For multi-user (MU) amplify-and-forward (AF) cooperative networks, their spectral efficiency can be upgraded within the orthogonal transmission of each source node to an assigned subset of all available relays while their information throughput can be improved through optimized power allocation. We consider the joint optimization in both relay assignment for each source-destination pair and power allocation, which is in fact among the hardest problems in optimization. This is the minimization of a nonconvex objective function subject to mixed integer constraints. The existing numerical algorithms could rarely address to its solutions through computationally affordable procedures. Even the conventional relaxation of the integer constraints by linear constraints does not lead to convex optimization, so the standard convexification does not work either. Nevertheless, we show that it can be effectively solved in the d.c. (difference of two convex) programming context. Numerical simulation confirms the effectiveness of our setting.