Joint Estimation and Detection for Massive Access in Low-Altitude IoT Networks

This paper investigates the estimation of sparse channel state information (CSI) and the detection of active information in a massive machine-type communication (mMTC) system utilizing an uncrewed aerial vehicle (UAV) base station (BS). Both rotary-wing and fixed-wing UAVs are considered in the context of large-scale connectivity. In the first scenario, the rotary-wing UAV is equipped with a small-scale multiple-input multiple-output (MIMO) system. A single measurement vector approach is employed to estimate CSI across the time, frequency, and spatial domains, referred to as 3D-domain estimation. In the second scenario, the study focuses on angular-domain-based CSI estimation and active information detection for an mMTC system utilizing a fixed-wing UAV BS with massive MIMO. The multiple measurement vector Bayesian posteriori minimum mean square error method is implemented to enhance performance. To support these estimations, non-orthogonal discrete Fourier transform sequences are designed for the codebook matrix and transmitted via orthogonal frequency-division multiplexing subcarriers. Additionally, during each iteration, channel parameter learning facilitates the transfer of extrinsic expectation and extrinsic variance, ensuring convergence to the final results. Numerical simulations demonstrate the proposed approach's superior estimation and detection performance compared to existing methods, such as approximate message-passing algorithms and the orthogonal matching pursuit method, in terms of normalized mean square error, error detection probability, and bit error rate.