[Paper Review] Scheduling and Pre-Conditioning in Multi-User MIMO TDD Systems

Jubin Jose, Alexei Ashikhmin+2 more|ArXiv.org|Sep 28, 2007

Advanced MIMO Systems Optimization11 references4 citations

TL;DR

This paper proposes a joint scheduling and pre-conditioning framework for multi-user MIMO TDD systems with reciprocal channels, optimizing training sequence length and user selection to maximize net downlink throughput. It derives tighter sum capacity bounds for homogeneous users and an optimized pre-conditioning matrix for heterogeneous users, showing significant gains in weighted-sum rate through scheduling and channel reciprocity exploitation in interference-limited, high-mobility scenarios.

ABSTRACT

The downlink transmission in multi-user multiple-input multiple-output (MIMO) systems has been extensively studied from both communication-theoretic and information-theoretic perspectives. Most of these papers assume perfect/imperfect channel knowledge. In general, the problem of channel training and estimation is studied separately. However, in interference-limited communication systems with high mobility, this problem is tightly coupled with the problem of maximizing throughput of the system. In this paper, scheduling and pre-conditioning based schemes in the presence of reciprocal channel are considered to address this. In the case of homogeneous users, a scheduling scheme is proposed and an improved lower bound on the sum capacity is derived. The problem of choosing training sequence length to maximize net throughput of the system is studied. In the case of heterogeneous users, a modified pre-conditioning method is proposed and an optimized pre-conditioning matrix is derived. This method is combined with a scheduling scheme to further improve net achievable weighted-sum rate.

Motivation & Objective

• To address the tight coupling between channel training, estimation, and system throughput in interference-limited, high-mobility multi-user MIMO TDD systems.
• To develop a scheduling and pre-conditioning scheme that maximizes net achievable sum or weighted-sum rate without assuming perfect channel state information at the base station.
• To optimize training sequence length and user set size to maximize net system throughput under practical constraints.
• To derive a computationally efficient, optimized pre-conditioning matrix for heterogeneous users under the M-large assumption.
• To demonstrate the performance gains of joint optimization of scheduling and pre-conditioning in realistic low-SINR environments.

Proposed method

• Proposes a scheduling scheme for homogeneous users that improves the lower bound on sum capacity by exploiting user channel variations and reducing pre-conditioning complexity.
• Introduces a training sequence length optimization problem to maximize net throughput, subject to constraints on training overhead and coherence interval.
• Derives an optimized pre-conditioning matrix for heterogeneous users under the M-large assumption, minimizing effective interference while respecting user-specific weights and SINRs.
• Combines the optimized pre-conditioning with a simple scheduling strategy to enhance the weighted-sum rate in multi-user scenarios.
• Uses a TDD-based reciprocal channel model where the reverse channel is the transpose of the forward channel, enabling low-overhead channel estimation.
• Employs a time-division duplex (TDD) system model with block fading and OFDM to account for frequency selectivity, assuming i.i.d. Rayleigh fading on channel entries.

Experimental results

Research questions

• RQ1 How can scheduling and pre-conditioning be jointly optimized in multi-user MIMO TDD systems with reciprocal channels to maximize net downlink throughput?
• RQ2 What is the optimal training sequence length that maximizes net system throughput in interference-limited, high-mobility environments?
• RQ3 How does user heterogeneity in forward and reverse SINRs affect the design of an effective pre-conditioning matrix and scheduling policy?
• RQ4 To what extent do scheduling gains improve sum or weighted-sum rate when training overhead is minimized and channel reciprocity is exploited?
• RQ5 Can a computationally efficient pre-conditioning method be derived that maintains high performance across diverse user SINR and weight configurations?

Key findings

• The proposed scheduling scheme for homogeneous users achieves a tighter lower bound on sum capacity than prior work, with significant performance gains even at low forward SINR (0 dB).
• Net achievable sum rate increases with the number of base-station antennas (M), and the optimized training sequence length τ_rp* is found to equal the optimal number of users K* across all tested forward SINR levels.
• In the heterogeneous users scenario, combining optimized pre-conditioning with scheduling (Scheme-3) yields the highest net achievable weighted-sum rate, with substantial gains when K ≈ M.
• The optimal number of users K* and training length τ_rp* are found to be equal and increase with forward SINR, indicating a strong dependence on channel quality.
• The performance gain from scheduling is most pronounced at low forward SINR (e.g., 0 dB), where interference limits spectral efficiency.
• The derived pre-conditioning matrix is applicable to arbitrary user weights, forward and reverse SINRs, and involves a computationally simple optimization process suitable for real-time implementation.