Efficient subcarrier pairing and power allocation in multi-relay cognitive networks

Abstract

This paper presents an efficient algorithm for joint subcarrier pairing, power allocation, and relay selection in an orthogonal frequency-division multiplexing (OFDM)-based cognitive decode-and-forward (DF) multi-relay network. Such a joint consideration entails a mixed integer programming (MIP) problem. To efficaciously resolve this complicated MIP problem, we propose a blended genetic algorithm (BGA), which tackles all of these three issues as a whole and is free of extra assumptions made in previous works to simplify the problem. In our BGA, every chromosome is divided into a number of integer strings for relay selection and subcarrier pairing, and a real number string for power allocation. A proper initialization scheme based on the convex optimization theory and the channel strengths is proposed. Moreover, new crossover and mutation operations are devised for these new chromosomes as well as to manage the interference to the primary users (PUs). Conducted simulations show that the developed BGA provides superior performance, yet with lower computations compared with previous works.