Blind Source Separation Based Pilot Collision Resolution for Grant-Free Random Access in Cell-free Massive MIMO Networks
Year: 2025
Type: conference-paper
Abstract: User access has become the main challenge for 5G and future networks, especially with the surge in the number of users. Compared with the traditional grant-based random access (GBRA), the grant-free random access (GFRA) allows users to directly transmit data, which avoid heavy signaling overhead and reduce delay. However, as the number of users grows, pilot collision interference resulting from uncoordinated resource selection in GFRA leads to serious transmission failure, especially in internet of things scenario. In this paper, we propose a blind source separation based pilot collision resolution algorithm for GFRA, making that limited preambles meet the needs of massive users. Utilizing single-channel blind source separation, conflicting users’ pilot signals are decomposed into several intrinsic mode functions (IMF) first, then the number of source signals is estimated. Next, the corresponding number of IMFs are selected to perform fast independent component analysis (FastICA). In this way, the pilot signals of conflicting user can be separated at the access point, which enables successful user access. We use similarity coefficient to assess the separation effect. In simulation experiments, the similarity between the separated signal and the original signal can be up to 0.8, which is a high degree of similarity, proving the feasibility of our approach.
Authors: Keying Li, Mingqi Gao, Changqing Li, Qi Su, Jianmei Dai
Institution: Space Engineering University
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7127049587