[Paper Review] A Rate Splitting Strategy for Mitigating Intra-Cell Pilot Contamination in Massive MIMO

Christo Kurisummoottil Thomas, Bruno Clerckx+2 more|arXiv (Cornell University)|Jan 1, 2020

Advanced MIMO Systems Optimization18 references15 citations

TL;DR

This paper proposes a single-layer Rate Splitting (RS) strategy to mitigate intra-cell pilot contamination in Massive MIMO systems where all users share the same pilot sequence. By splitting messages into common and private streams and using maximum ratio precoding for private streams and a weighted channel estimate-based precoder for the common stream, the scheme achieves higher spectral efficiency than conventional maximum ratio precoding, especially at high SNR, with numerical results showing non-saturating SE gains due to effective interference management.

ABSTRACT

The spectral efficiency (SE) of Massive MIMO (MaMIMO) systems is affected by low quality channel estimates. Rate-Splitting (RS) has recently gained some interest in multiuser multiple antenna systems as an effective means to mitigate the multi-user interference due to imperfect channel state information. This paper investigates the benefits of RS in the downlink of a single-cell MaMIMO system when all the users use the same pilot sequence for channel estimation. Novel expressions for the SE achieved in the downlink by a single-layer RS strategy (that relies on a single successive interference cancellation at each user side) are derived and used to design precoding schemes and power allocation strategies for common and private messages. Numerical results are used to show that the proposed RS solution achieves higher spectral efficiency that conventional MaMIMO with maximum ratio precoding.

Motivation & Objective

• To address the spectral efficiency (SE) degradation caused by pilot contamination in single-cell Massive MIMO systems where all users use the same pilot sequence.
• To design a rate splitting (RS) strategy that improves SE by managing interference from imperfect channel state information (CSI) due to pilot contamination.
• To derive closed-form expressions for SE under RS and develop a power allocation algorithm that optimizes common and private stream power distribution.
• To demonstrate through simulations that RS outperforms conventional maximum ratio precoding, especially at high SNR, with non-saturating SE gains.

Proposed method

• Derives novel closed-form expressions for downlink spectral efficiency (SE) using the hardening bound for both common and private messages in a single-layer RS scheme.
• Applies maximum ratio (MR) precoding to private streams and a weighted combination of all users' channel estimates for the common stream precoder.
• Develops a novel power allocation algorithm that jointly optimizes power among common and private streams under a total transmit power constraint.
• Uses a successive interference cancellation (SIC) process at each user: first decoding the common stream while treating private streams as noise, then removing it before decoding the private stream.
• Employs a concave approximation technique to handle the non-concave nature of the SE expression, enabling iterative power allocation via Lagrangian optimization.
• Derives analytical expressions for key statistical expectations involving channel estimates and precoders using matrix trace and covariance properties.

Experimental results

Research questions

• RQ1 Can a single-layer rate splitting strategy effectively mitigate intra-cell pilot contamination in Massive MIMO when all users share the same pilot sequence?
• RQ2 How does the spectral efficiency of the proposed RS scheme compare to conventional maximum ratio precoding in the presence of pilot contamination?
• RQ3 What is the optimal power allocation strategy between common and private streams that maximizes sum spectral efficiency under a total power constraint?
• RQ4 Does the RS scheme achieve non-saturating spectral efficiency at high SNR, and if so, why?
• RQ5 How does the performance of RS scale with the number of users and antennas in a finite-M Massive MIMO setup?

Key findings

• The proposed RS scheme achieves significantly higher spectral efficiency than conventional maximum ratio precoding, especially at high transmit power levels.
• At high SNR, the sum spectral efficiency with RS does not saturate, unlike conventional MaMIMO, due to effective interference suppression via the common stream.
• With M = 100 and K = 10, the RS scheme outperforms conventional MaMIMO by up to 30% in sum SE at 20 dBm transmit power.
• The performance gain diminishes as the number of users increases, due to the common message needing to be decodable by all users, which limits the common rate.
• The system with RS maintains a higher sum DoF (1 + (K−1)δ) compared to conventional MaMIMO (Kδ), indicating better interference management.
• The proposed power allocation algorithm successfully converges and achieves near-optimal SE by jointly optimizing common and private stream power.