A Pilot Allocation Scheme Based on Joint User Classification and Coalition Game for Massive MIMO Three-Tier HetNets

Abstract

Many researches demonstrate that pilot contamination is a major factor that affects the overall performance of massive MIMO systems. Based on this, this paper proposes a dynamic pilot allocation scheme based on joint user classification and coalition game (JUC-CG) to mitigate pilot contamination for Massive MIMO Three-tier heterogeneous networks (HetNets). It contains two steps: (1) According to the constraint in definition 2, the users are divided into two classes, namely A1 and A2. Users in pico-cell and sufferred from strong interference are classified into class A1, and remaining users of HetNets are classified into class A2. (2) The users of class A1 use a set of orthogonal pilots, and the users of class A2 reuse another set of orthogonal pilots by means of coalition game. In the coalition game for the users of class A2, users are divided into several disjoint user sub-coalitions, users belonging to different sub-coalitions are allocated different orthogonal pilot sequences, and users in the same sub-coalition reuse the same pilot sequence. The corresponding pilot allocation algorithm of JUC-CG is given. From the algorithm, we can see that JUC-CG scheme is flexible and can be used for scenarios that all users and base stations (BSs) are randomly distributed (such as multi-tier HetNets, ultra-dense HetNets and so on). In addition, the pilot allocation algorithm can obtain the overall optimal solution through cyclic searching. The simulation results demonstrate that the JUC-CG scheme can effectively reduce the average mean square error (MSE) of user signal detection and improve spectral efficiency (SE) in the uplink.