Joint active device detection and channel tracking of high-rate IoT devices in uncoordinated scenario with uncertain access probabilities over massive MIMO system

Abstract

Massive MIMO systems can serve a large number of devices simultaneously and support these devices' traffic requirements, which is a prime requisite for the high-rate Internet of Things (IoT). All IoT devices send pilots to BS without coordination, allowing pilot access to be uncoordinated. This paper aims to significantly increase the flexibility of uncoordinated scenarios by allowing devices to freely select their individual activity possibility as well as pilot access patterns. In each access slot, the BS is unaware of both the devices' activity probabilities and the Devices Pilot Access Pattern (DPAP). Thus, the detection concept of DPAPs converts uncoordinated access to coordinated access. This paper addresses the problem of active devices detection by simultaneously estimating and tracking all devices' channels, regardless of whether they transmit pilots or not. The Minimum Mean Square Error (MMSE) tracker for each access slot and its suboptimal version based on the Variational Bayesian method are derived. Additionally, a novel heuristic tracker is developed to reduce complexity. The convergence and stability conditions for these three trackers are then investigated. A tight bound on the performance limit is derived by adaptively estimating the uncertain devices' activity probabilities and DPAP. Theoretically, it is established that the performance of uncoordinated access is an ascending function of the devices' activity probabilities and that the introduced uncoordinated trackers are both effective and robust in the presence of channel state information uncertainty. Furthermore, this paper analyses the lower bound on the error covariance of the DPAP detection. The designed trackers can effectively resolve the channel aging issue. Moreover, there is no requirement to uplink access request control channels and downlink paging. Finally, simulation results demonstrate the designed uncoordinated trackers' performance, complexity, convergence, and error detection probability.