Joint Beamforming Design and Stream Allocation for Non-Coherent Joint Transmission in Cell-Free MIMO Networks

Xi Wang, Xiaotong Zhao, Juncheng Wang, Li You, Qingjiang ShiarXiv 2024 · DOI · arXiv

Abstract

We consider joint beamforming and stream allocation to maximize the weighted sum rate (WSR) for non-coherent joint transmission (NCJT) in user-centric cell-free MIMO networks, where distributed access points (APs) are organized in clusters to transmit different signals to serve each user equipment (UE). We for the first time consider the common limits of maximum number of receive streams at UEs in practical networks, and formulate a joint beamforming and transmit stream allocation problem for WSR maximization under per-AP transmit power constraints. Since the integer number of transmit streams determines the dimension of the beamformer, the joint optimization problem is mixed-integer and nonconvex with coupled decision variables that is inherently NP-hard. In this paper, we first propose a distributed low-interaction reduced weighted minimum mean square error (RWMMSE) beamforming algorithm for WSR maximization with fixed streams. Our proposed RWMMSE algorithm requires significantly less interaction across the network and has the current lowest computational complexity that scales linearly with the number of transmit antennas, without any compromise on WSR. We draw insights on the joint beamforming and stream allocation problem to decouple the decision variables and relax the mixed-integer constraints. We then propose a joint beamforming and linear stream allocation algorithm, termed as RWMMSE-LSA, which yields closed-form updates with linear stream allocation complexity and is guaranteed to converge to the stationary points of the original joint optimization problem. Simulation results demonstrate substantial performance gain of our proposed algorithms over the current best alternatives in both WSR performance and convergence time.

Introduction: The Challenge of Cell-Free MIMO

In the quest for 6G, the "cell-free" MIMO paradigm is a frontrunner. Instead of users connecting to a single base station at a cell edge, they are served by a distributed cluster of Access Points (APs). This paper focuses on Non-Coherent Joint Transmission (NCJT), where APs send different data signals to the same user. While NCJT is more practical than coherent transmission (it doesn't require nanosecond-level synchronization across the whole network), it introduces a messy optimization problem.

Cell-free MIMO network architecture featuring distributed Access Points (APs), User Equipments (UEs), and Central Units (CUs).

Specifically, researchers face three hurdles:

  1. Coupled Variables: The number of data streams a user receives determines the size of the beamforming matrix, making it hard to optimize both simultaneously.
  2. Computational Complexity: Standard algorithms scale cubically with the number of transmit antennas ($M^3$), which is a nightmare for "massive MIMO" setups.
  3. Fronthaul Bottlenecks: Centralized controllers usually need to exchange massive amounts of raw channel data with APs, clogging the network's internal links.

Methods: Breaking the Complexity Bottleneck

The authors propose a two-step solution to break these bottlenecks:

1. Proving WMMSE Applicability

Contrary to previous beliefs in the field, the authors proved that the classic Weighted Minimum Mean Square Error (WMMSE) approach can be adapted for NCJT. They reformulated the non-convex rate equations into a convex MSE minimization problem by introducing auxiliary variables.

2. The RWMMSE-LSA Algorithm

To solve the joint problem, they introduced two key innovations:

  • Low-Dimensional Substitution (RWMMSE): They observed that the optimal beamformer always lies in the subspace spanned by the channel vectors. This allows them to "reduce" the optimization space. Instead of working with large matrices based on the number of transmit antennas ($M$), they work with smaller ones based on the number of receive antennas ($N$). This shifts the complexity from $O(M^3)$ to $O(M)$, a massive win for large-scale arrays.
  • Linear Stream Allocation (LSA): To handle the "how many streams" question (a discrete integer problem), they used a "stream indicator matrix" and a quadratic-linear property. This allowed them to relax the integer constraints into a continuous problem with closed-form updates, ensuring the algorithm converges to a stationary point of the original problem.

Key Results: Efficiency and Performance

The proposed RWMMSE-LSA algorithm demonstrates significant improvements in both efficiency and performance.

Comparison of interaction overhead and computational complexity for Local EZF, WMMSE, and the proposed RWMMSE algorithms.
  • Speed: The proposed WMMSE-based approach is over 100x faster than the traditional SCA-CVX method while achieving the exact same Weighted Sum Rate (WSR).
  • Performance Gain: Compared to simple Local Zero-Forcing (EZF), the RWMMSE-LSA algorithm provides a 48% improvement in sum-rate performance at high Signal-to-Noise Ratios (SNR).
  • Efficiency: The algorithm achieves near-optimal performance in just 3 iterations, making it highly suitable for real-time implementation in dynamic wireless environments.
  • Complexity Reduction: As shown in the comparison table, RWMMSE reduces computational complexity from $O(M_i^3)$ to $O(M_i)$, enabling its use in massive MIMO systems with many antennas.

Conclusion and Future Work

By leveraging a novel low-dimensional substitution property, the authors developed a framework that jointly optimizes beamforming and data stream allocation with significantly lower communication overhead than previous state-of-the-art methods. While Coherent Joint Transmission (CJT) still offers higher rates, the reduced synchronization overhead of this NCJT approach makes it a more viable candidate for practical 6G deployments.

Future work includes extending this "linear stream allocation" logic to User Scheduling (RWMMSE-LUS), where the algorithm would automatically decide which APs should serve which users to maximize global efficiency.

Summary updated

Abstract

We consider joint beamforming and stream allocation to maximize the weighted sum rate (WSR) for non-coherent joint transmission (NCJT) in user-centric cell-free MIMO networks, where distributed access points (APs) are organized in clusters to transmit different signals to serve each user equipment (UE). We for the first time consider the common limits of maximum number of receive streams at UEs in practical networks, and formulate a joint beamforming and transmit stream allocation problem for WSR maximization under per-AP transmit power constraints. Since the integer number of transmit streams determines the dimension of the beamformer, the joint optimization problem is mixed-integer and nonconvex with coupled decision variables that is inherently NP-hard. In this paper, we first propose a distributed low-interaction reduced weighted minimum mean square error (RWMMSE) beamforming algorithm for WSR maximization with fixed streams. Our proposed RWMMSE algorithm requires significantly less interaction across the network and has the current lowest computational complexity that scales linearly with the number of transmit antennas, without any compromise on WSR. We draw insights on the joint beamforming and stream allocation problem to decouple the decision variables and relax the mixed-integer constraints. We then propose a joint beamforming and linear stream allocation algorithm, termed as RWMMSE-LSA, which yields closed-form updates with linear stream allocation complexity and is guaranteed to converge to the stationary points of the original joint optimization problem. Simulation results demonstrate substantial performance gain of our proposed algorithms over the current best alternatives in both WSR performance and convergence time.
BibTeX
@misc{wang2024joint,
  author = {Xi Wang and Xiaotong Zhao and Juncheng Wang and Li You and Qingjiang Shi},
  title = {{Joint Beamforming Design and Stream Allocation for Non-Coherent Joint Transmission in Cell-Free MIMO Networks}},
  year = {2024},
  eprint = {2402.18231},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2402.18231},
}
Figure