Abstract
A great amount of endeavor has recently been devoted to activity detection for massive machine-type communications in cell-free massive MIMO. However, in practice, as the number of antennas at the access points (APs) increases, the Rayleigh distance that separates the near-field and far-field regions also expands, rendering the conventional assumption of far-field propagation alone impractical. To address this challenge, this paper considers a hybrid near-far field activity detection in cell-free massive MIMO, and establishes a covariance-based formulation, which facilitates the development of a distributed algorithm to alleviate the computational burden at the central processing unit (CPU). Specifically, each AP performs local activity detection for the devices and then transmits the detection result to the CPU for further processing. In particular, a novel coordinate descent algorithm based on the Sherman-Morrison-Woodbury update with Taylor expansion is proposed to handle the local detection problem at each AP. Moreover, we theoretically analyze how the hybrid near-far field channels affect the detection performance. Simulation results validate the theoretical analysis and demonstrate the superior performance of the proposed approach compared with existing approaches.
Introduction
The evolution toward 6G networks necessitates support for massive machine-type communications (mMTC), characterized by millions of IoT devices requiring sporadic network access. Cell-Free Massive MIMO has emerged as a promising architecture to meet these demands by deploying numerous Access Points (APs) that cooperatively serve devices without traditional cellular boundaries. However, the increasing physical size of antenna arrays at APs extends the Rayleigh distance—the boundary where spherical near-field waves transition into planar far-field waves. Consequently, practical deployments must account for a hybrid near-far field propagation environment. Traditional far-field models fail in these scenarios, and centralized processing of high-dimensional near-field data creates significant computational bottlenecks at the Central Processing Unit (CPU).
System Model and Distributed Architecture
To address these challenges, we propose a distributed covariance-based activity detection scheme. The architecture, illustrated in Figure 1, decomposes the global detection problem into local tasks performed at each AP.
The framework utilizes the Alternating Direction Method of Multipliers (ADMM) to achieve consensus between local AP detections and the global state maintained by the CPU. This distribution of labor avoids the $O((LK)^3)$ complexity bottleneck inherent in centralized processing, where $L$ is the signature sequence length and $K$ is the number of antennas per AP.
Algorithmic Innovation: CD-SMW-Taylor
A core contribution of this work is the development of a novel Coordinate Descent (CD) algorithm to solve the local detection problem at each AP. Near-field covariance matrices are inherently complex and high-dimensional, making direct matrix inversions computationally prohibitive. Our "CD-SMW-Taylor" approach leverages the Sherman-Morrison-Woodbury (SMW) identity and a first-order Taylor expansion to approximate these complex operations. This allows for efficient updates that maintain high accuracy while significantly reducing the computational load, enabling real-time operation in massive MIMO settings.
Results and Performance Analysis
The proposed method was evaluated through extensive numerical simulations in a 200m x 200m area with 100 IoT devices and a 10% activity ratio. Channel models followed standard path loss equations ($128.1 + 37.6 \log_{10}(\tau)$). As shown in Figure 2, the distributed algorithm demonstrates remarkable efficiency, typically converging to the performance level of a centralized coordinate descent (CD) approach within just two iterations.
Furthermore, the proposed approach consistently outperforms established benchmarks, including Approximate Message Passing (AMP), Orthogonal AMP (OAMP), and Vanilla Maximum Likelihood Estimation (MLE). A key insight from our analysis is that near-field effects, often viewed as a complication, actually serve as a feature. We theoretically prove that near-field channel vectors exhibit higher orthogonality (lower cosine similarity) than far-field vectors, which enhances the system's ability to distinguish between active devices and reduces the probability of error.
Conclusion
This paper presents a robust distributed framework for activity detection in hybrid near-far field cell-free networks. By leveraging a novel coordinate descent algorithm and the inherent spatial advantages of near-field propagation, we achieve superior detection accuracy with significantly reduced central computational requirements. Future work will focus on relaxing assumptions regarding channel state information (CSI) and exploring the impact of diverse antenna geometries on near-field orthogonality benefits.
Summary updated