Distributed Joint User Activity Detection, Channel Estimation, and Data Detection via Expectation Propagation in Cell-Free Massive MIMO

2024-05-16

International Workshop on Signal Processing Advances in Wireless Communications

AI Summary

To address the challenges of severe coupling among user activity detection, channel estimation, and data detection; pronounced pilot contamination; and poor scalability of centralized processing in uplink cell-free massive MIMO (CF-mMIMO) systems, this paper proposes a distributed joint detection framework based on Expectation Propagation (EP). By constructing a factor graph model, the joint posterior distribution is decomposed into localized message-passing tasks, enabling decentralized iterative inference at base stations. This work is the first to introduce EP into distributed joint detection for CF-mMIMO, effectively mitigating pilot contamination caused by pilot reuse. Experiments demonstrate that the proposed method maintains high detection accuracy and strong robustness even in overloaded scenarios where the number of users far exceeds the pilot sequence length—thereby overcoming the fundamental bottleneck in conventional grant-free CF-mMIMO, wherein user capacity is strictly limited by pilot resource constraints.

Technology Category

• Constraint Satisfaction and Optimization: Distributed CSP/Optimization
• Machine Learning: Distributed Machine Learning & Federated Learning
• Search and Optimization: Distributed Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Security and Privacy: Large-scale security measurements
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing

Abstract

We consider the uplink of a grant-free cell-free massive multiple-input multiple-output (GF-CF-MaMIMO) system. We propose an algorithm for distributed joint activity detection, channel estimation, and data detection (JACD) based on expectation propagation (EP) called JACD-EP. We develop the algorithm by factorizing the a posteriori probability (APP) of activities, channels, and transmitted data, then, mapping functions and variables onto a factor graph, and finally, performing a message passing on the resulting factor graph. If users with the same pilot sequence are sufficiently distant from each other, the JACD-EP algorithm is able to mitigate the effects of pilot contamination which naturally occurs in grant-free systems due to the large number of potential users and limited signaling resources. Furthermore, it outperforms state-of-the-art algorithms for JACD in GF-CF-MaMIMO systems.

Problem

• User Activity Detection
• Channel State Estimation
• Data Detection

Innovation

• JACD-EP Algorithm
• Expectation Propagation
• GF-CF-MaMIMO System