Weighted-Sum-Rate-Maximizing Linear Transceiver Filters for the K-User MIMO Interference Channel

Joonwoo Shin, Jaekyun Moon

Abstract

This paper is concerned with transmit and receive filter optimization for the K-user MIMO interference channel. Specifically, linear transmit and receive filter sets are designed which maximize the weighted sum rate while allowing each transmitter to utilize only the local channel state information. Our approach is based on extending the existing method of minimizing the weighted mean squared error (WMSE) for the MIMO broadcast channel to the K-user interference channel at hand. For the case of the individual transmitter power constraint, however, a straightforward generalization of the existing method does not reveal a viable solution. It is in fact shown that there exists no closed-form solution for the transmit filter, but a simple one-dimensional parameter search yields the desired solution. Compared to the direct filter optimization using gradient-based search, our solution requires considerably less computational complexity and a smaller amount of feedback resources while achieving essentially the same level of weighted sum rate. A modified filter design is also presented which provides desired robustness in the presence of channel uncertainty.

Introduction

In a $K$-user MIMO interference channel, multiple transmitter-receiver pairs operate simultaneously within the same frequency band, creating mutual interference that can significantly degrade system performance. The primary objective in managing such networks is often the maximization of the Weighted Sum Rate (WSR). This metric is critical because it allows network operators to balance total throughput with user fairness by assigning different weights to different users.

Despite its importance, maximizing WSR is a non-convex and computationally difficult problem. Existing approaches, such as Interference Alignment (IA), often target the high signal-to-noise ratio (SNR) regime and may perform poorly at finite SNRs. Furthermore, many high-performance precoding designs rely on Global Channel State Information (CSI), requiring every transmitter to possess knowledge of every link in the network. As the number of users ($K$) increases, the overhead required for global CSI feedback grows quadratically, leading to a "feedback explosion." This paper proposes a computationally efficient transceiver design that circumvents these issues by utilizing local CSI and a simplified optimization framework.

Methodology: The WMSE-WSR Framework

The core of the proposed method lies in the established relationship between WSR maximization and Weighted Mean Squared Error (WMSE) minimization. While this equivalence was previously leveraged for MIMO broadcast channels (single transmitter, multiple receivers), this work extends the framework to the more complex $K$-user interference channel (multiple transmitters, multiple receivers).

Iterative Optimization

The authors propose an alternating optimization algorithm that iteratively updates three sets of variables:

  1. Receive Filters: Optimized to minimize the MSE for a given set of transmit filters.
  2. MSE Weights: Updated based on the current error levels to maintain the link to the WSR objective.
  3. Transmit Filters: Optimized to maximize the WSR (or minimize WMSE) subject to power constraints.

Solving the Power Constraint Challenge

A significant contribution of this work is the handling of power constraints. The paper examines two distinct scenarios:

  • Sum-Power Constraint: Where a total power limit is shared across the network. In this case, a closed-form solution for the transmit filters is derived.
  • Individual-Power Constraint: Where each transmitter has its own power limit. The authors demonstrate that a closed-form solution does not exist for this more realistic scenario. However, they prove that the optimal transmit filter can be found via a simple one-dimensional (1D) numerical search. This is possible because the transmit power is shown to be a monotonically decreasing function of the Lagrange multiplier ($\lambda_k$), allowing for a rapid bisection search.

Decentralized Information and Feedback Efficiency

One of the most practical advantages of this approach is its reliance on Local CSI. In the proposed framework, transmitters do not need to know the cross-channels between other users. Instead, the iterative process allows the necessary information to be captured through the exchange of filter settings and effective channel gains.

This shift from global to local requirements changes the scaling of feedback resources. Traditional gradient-based methods typically require feedback that scales quadratically ($K^2$) with the number of users. In contrast, the proposed method scales linearly ($K$), making it far more viable for large-scale wireless networks.

Performance and Robustness

The proposed algorithm was evaluated using Monte Carlo simulations across 1,000 independent trials with i.i.d. complex Gaussian channels. The results highlight three key areas of success:

  1. Computational Complexity: For a 20-user system, the proposed method requires significantly fewer operations (below $10^7$ complex multiplications) compared to gradient-based searches (nearly $10^8$), representing an order-of-magnitude improvement in speed.
  2. Rate Performance: Despite the reduced complexity and feedback, the proposed scheme achieves a WSR essentially identical to that of the more intensive gradient-based optimization.
  3. Robustness: Recognizing that perfect CSI is rarely available, the authors introduced a modified "robust" design. Simulations show that this design maintains stable performance even with 10% channel uncertainty, whereas standard designs experience significant rate degradation at high SNR levels.

Conclusion

This paper provides a robust and efficient solution for the $K$-user MIMO interference channel. By exploiting the monotonicity of the power function and the relationship between WMSE and WSR, the authors developed a transceiver design that significantly reduces computational overhead and feedback requirements. The transition from global to local CSI requirements, combined with the 1D parameter search, offers a scalable path forward for managing interference in dense, multi-user wireless environments.

Summary updated

Abstract

This letter is concerned with transmit and receive filter optimization for the K-user MIMO interference channel. Specifically, linear transmit and receive filter sets are designed which maximize the weighted sum rate while allowing each transmitter to utilize only the local channel state information. Our approach is based on extending the existing method of minimizing the weighted mean squared error (MSE) for the MIMO broadcast channel to the K-user interference channel at hand. For the case of the individual transmitter power constraint, however, a straightforward generalization of the existing method does not reveal a viable solution. It is in fact shown that there exists no closed-form solution for the transmit filter but simple one-dimensional parameter search yields the desired solution. Compared to the direct filter optimization using gradient-based search, our solution requires considerably less computational complexity and a smaller amount of feedback resources while achieving essentially the same level of weighted sum rate. A modified filter design is also presented which provides desired robustness in the presence of channel uncertainty
BibTeX
@article{shin2012weighted,
  author = {Joonwoo Shin and Jaekyun Moon},
  title = {{Weighted-Sum-Rate-Maximizing Linear Transceiver Filters for the K-User MIMO Interference Channel}},
  journal = {IEEE Trans. Communications},
  year = {2012},
  url = {https://openreview.net/forum?id=ztdRMnlu_I},
}
Figure