Joint Message Detection and User Position Estimation for Cell-Free Networks in Realistic Propagation Environments

Accurate user localization is a fundamental challenge in wireless networks, particularly in user-centric cell-free (CF) architectures. In this work, we propose a novel localization framework for realistic propagation environments, that does not require accurate synchronization, array calibration, and line-of-sight propagation, unlike methods based on time-of-arrival (ToA) and angle-of-arrival (AoA). Our approach employs a multisource Approximated Message Passing (AMP) algorithm to efficiently detect uplink pilot sequences transmitted by the users on the random access channel (RACH) slot. The multisource AMP makes use of a coarse knowledge of the large scale fading coefficients (LSFCs) of the active users, obtained by associating RACH pilot codebooks with regions of the coverage area with similar LSFC “profile” (i.e., the collection of LSFCs from a given position to the radio units (RUs) of the CF network. Such regions, referred to as “clusters”, are obtained by clustering the LSFC profiles generated (offline) via state-of-the-art ray-tracing. Then, the scheme builds an estimate of the a-posteriori probability of the user positions given the AMP output, which can be analytically calculated using the rigorous AMP asymptotic statistics. Finally, the (active) user positions can be estimated using the maximum a-posteriori criterion. In this way, fast and accurate localization of random access users with potentially very sporadic user activity can be achieved.