Research On Pilot Decontamination In Multi-cell Multi-user Massive MIMO Systems

Massive MIMO is one of the key technologies of the fifth generation communication system.It can provide a remarkable increase in reliability and data rate with simple signal processing.Massive MIMO has many advantages over existing communication technology,but pilot contamination seriously restricts the performance of multi-cell Massive MIMO and becomes a more serious bottleneck problem.We categorize the different proposed mitigation techniques for pilot contamination using high-precision channel estimation algorithm and effective pilot assignment scheme.Firstly,we give a brief description on massive MIMO system model,and then we describe the time division duplex scheme.Based on this model,we study two communication schemes,uplink training and data transmission.The use of non-orthogonal pilot signals in multi-cell system introduces the problem of pilot contamination.Other possible causes of pilot contamination,such as hardware impairment and non-reciprocal are also pointed out in existing literature.And then,we analyze the impact of pilot contamination and develop methods that mitigate this contamination.The simulation results show that the effect of pilot contamination on area throughput and energy efficiency using different pilot reuse factors.Secondly,we review existing methods proposed to eliminate or reduce the effects of pilot contamination in multi-cell system with time division duplex scheme.The proposed methods have been classified into two categories namely,pilot-based estimation approach and subspace-based semiblind estimation approach.In the pilot-based estimation approach,least square and minimum mean square error channel estimation methods are studied.Then we introduce an improved minimum mean square error method,it can reduce the computational complexity of system effectively.Semiblind channel estimation algorithm only needs little pilot,so as to avoid the pilot contamination.We analyze and discuss the semiblind channel estimation algorithms based eigenvalue decomposition and singular value decomposition proposed in the literature.Furthermore,to improve the performance of that approaches,we combine them with the iterative least square with projection algorithm.Numerical results verify the effectiveness of this channel estimation approach.Lastly,a new time-shifted with different code types of pilot and cell sectorization pilot assignment joint scheme is proposed.This scheme can mitigate pilot contamination by reducing the reuse of pilot sequences through time domain,code domain and space domain.And we find approximate closed-form expressions for uplink and downlink achievable rates.Numerical results show that the proposed training strategy significantly improves both the achievable uplink and downlink rates when compared with those of a conventional pilot strategies.