Blind Detection in Coexistence of Human-Type and Machine-Type Communications

Abstract

In this paper, we study joint device activity identification, channel estimation, and signal detection for the uplink transmission of a human-type communication (HTC) and machine-type communication (MTC) coexisted massive MIMO system. We first establish a probability model to characterize the crucial system features including channel sparsity of massive MIMO, signal sparsity of MTC packets, and sporadic access of MTC. With the probability model, we formulate a blind detection problem and establish a factor graph representation of the problem. Based on that, we develop a turbo message passing (TMP) algorithm involving affine sparse matrix factorization and service type identification. We show that our proposed blind detection algorithm significantly outperform their counterpart algorithms including the training-based algorithm.