Low-Complexity Design of Massive Device Detection via Riemannian Pursuit

Active device detection is a precondition of realizing grant-free random access in beyond fifth-generation (B5G) cellular Internet-of-Things (IoT). However, due to the deployment of a large antennas array and the existence of a huge number of IoT devices, activity detection usually has high computational complexity and needs long pilot sequences. To overcome these challenges, we first propose a dimension deduction method by projecting the original device state matrix to a much lower dimension space. Then, we develop an optimized design framework with a logarithmic smoothing objective function and a coupled full column rank constraint. Under that framework, we transform the original interested matrix to a positive semidefinite matrix, followed by proposing a Riemannian trust-region algorithm to solve the problem in complex field. Simulation results show that the proposed algorithm outperforms the state-of-art algorithms in terms of device detection performance.