Dynamic Unsourced Random Access With Massive MIMO

Abstract

In this paper, we propose a novel dynamic unsourced random access (URA) framework for massive multiple-input multiple-output (MIMO) uplink access. Unlike the existing studies, where the quasi-static channel models and the unchanged user states (active or idle) are assumed, we take the dynamics in both the channels and the states of user devices into consideration. Such a framework supports the high mobility of user devices, and facilitates their abrupt terminates and accesses during the whole transmission process. To model the dynamics, we adopt steady-state Gaussian Markov processes for all the channel coefficients of user devices, and introduce a series of latent variables to indicate the user states. We design a two-step algorithm, including the approximate message passing (AMP)-based inner decoding algorithm and the variational message passing (VMP)-based outer decoding algorithm, to decode the information sequences for all the user devices that have accessed the network. Simulation results show that our proposed method outperforms all the baselines when there are dynamics in the channels of user devices, and our proposed method has robustness to deal with the abrupt changes of user states by equipping the large number of antennas at the base station.