Interference alignment with iterative channel estimation for the reciprocal M×2 MIMO X Network

This paper investigates an interference alignment (IA) scheme for a reciprocal multi-input multi-output (MIMO) M×2 X network where the knowledge of channel state information (CSI) is required. In our proposed approach, singular vectors, calculated from the singular value decomposition (SVD) of channel matrices, are used to compute precoding and zero-forcing (ZF) decoding matrices at transmitters and receivers, respectively. The orthogonality between precoding and decoding vectors that results from SVD is an advantage for realizing IA scheme because we can rely on an iterative scheme, known as iterative power method (IPM). The singular vectors resulting from the IPM approach converge to the actual ones after multiple iterations assuming a common “virtually static” channel between each link. However, due to the fast fading nature of the channel, computed precoding and ZF decoding vectors will be different from those resulting from the SVD of the actual channel. To this end, the IPM applied to get an estimate of precoding and ZF decoding vectors allows a better tracking of the time-varying channel. The bit error rate of the proposed scheme is evaluated by means of Monte Carlo simulations and compared with that achieved by a perfect CSI based system.