Firefly Algorithm for Beamforming Design in RIS-aided Communication Systems

This paper studies a non-convex power minimization problem for reconfigurable-intelligent-surfaces-aided communication systems whose constraints are multivariate functions of two independent optimization variables, i.e., active and passive beamforming vectors. A widely adopted alternative optimization (AO) approach approximates the originally non-convex problem by two convex sub-optimization problems where each suboptimization problem deals with one variable considering the other variable as a constant. The solution for the original problem is obtained by iteratively solving these sub-optimization problems. Although the AO approach converts the original NP-hard optimization problem to two convex sub-problems, the solutions attained by this method may not be the global optimal solution due to the approximation process as well as the inherent non-convexity of the original problem. To overcome the issue, this paper adopts a nature-inspired optimization approach and introduces a novel Firefly algorithm (FA) to simultaneously solve for two independent optimization variables of the originally nonconvex optimization problem. Computational complexity analyses are provided for the proposed FA and the AO approaches. Simulation results reveal that the proposed FA approach prevails its AO counterpart in obtaining a better solution for the under-studied optimization problem with the same order of computational complexity.