Semi-Blind Source Separation based on Multi-Modulus Criterion - Application for Pilot Contamination Mitigation in Massive MIMO Communications Systems.

This paper presents an efficient semi-blind approach for demixing instantaneous mixtures, in a massive Multiple-Input Multiple-Output (MIMO) communications system, with pilot contamination. A hybrid cost function is defined based on the Multi-Modulus (MM) criterion, for the blind part, and the Least Squares (LS) criterion for the pilot-based part. A simple but efficient semi-blind block gradient descent procedure is put forward, in which the step size, which globally minimizes the cost function along the search direction, is algebraically computed at each iteration for each user. Simulation results show that the use of pilots enhance the convergence and allow to mitigate the inherent ambiguity of fully-blind methods. Moreover, the optimal step size notably accelerates the convergence and can further reduce the impact of local extrema. Furthermore, the proposed approach is very promising for source recovering when pilot contamination occurs in massive MIMO communications systems.