Novel Markov channel predictors for interference alignment in cognitive radio network

Abstract

In cognitive radio (CR) network, how to mitigate interference between different users is a key task. Interference alignment (IA) is a promising technique to tackle the multi-user interference in communication system. Compared with other interference management methods (such as zero-forcing), IA can not only effectively eliminate the interference, but also greatly increase the system capacity. However, the perfect channel state information (CSI) is required for both transmitters and receivers to apply the IA algorithm, which is hard to achieve in practical applications. In this paper, the effect of imperfect CSI on IA in CR system is analyzed in terms of signal to interference plus noise ratio and achievable sum rate. A linear finite state Markov chain (LFSMC) predictor, which incorporates the finite state Markov chain into the AR predictor, is proposed to reduce the impact of imperfect CSI on system performance of CR network. Moreover, for the sake of simplifying the initialization of LFSMC predictor, a simplified LFSMC (S-LFSMC) predictor is provided. Simulation results indicate that both of the LFSMC and S-LFSMC predictor can greatly improve the performance of IA system with the inaccurate CSI. Specifically, the LSFMC predictor can achieve satisfied performance compared with other predictors mentioned in this paper. And the LSFMC predictor which is simpler and its performance is still much better than traditional predictors. Therefore, we can choose a suitable predictor (LAFMC or S-LSFMC) based on the different requirements.