Heterogeneous Doppler Spread-based CSI Estimation Planning for TDD Massive MIMO

TL;DR

This paper introduces a novel uplink training scheme for TDD Massive MIMO systems that adapts pilot transmission periodicity based on users' heterogeneous Doppler spreads, significantly improving spectral efficiency.

Contribution

It proposes a two time scale control framework optimizing pilot scheduling and location estimation considering channel and user mobility heterogeneity.

Findings

Significant spectral efficiency gains with the proposed scheme.

Effective training policy adapts to users' mobility profiles.

Partial location knowledge still yields notable improvements.

Abstract

Massive multi-input multi-output (Massive MIMO) has been recognized as a key technology to meet the demand for higher data capacity and massive connectivity. Nevertheless, the number of active users is restricted due to training overhead and the limited coherence time. Current wireless systems assume the same coherence slot duration for all users, regardless of their heterogeneous Doppler spreads. In this paper, we exploit this neglected degree of freedom in addressing the training overhead bottleneck. We propose a new uplink training scheme where the periodicity of pilot transmission differs among users based on their actual channel coherence times. Since the changes in the wireless channel are, primarily, due to movement, uplink training decisions are optimized, over long time periods, while considering the evolution of the users channels and locations. Owing to the different rates of the wireless channel and location evolution, a two time scale control problem is formulated. In the fast time scale, an optimal training policy is derived by choosing which users are requested to send their pilots. In the slow time scale, location estimation decisions are optimized. Simulation results show that the derived training policies provide a considerable improvement of the cumulative average spectral efficiency even with partial location knowledge.