The current High-Speed Railway (HSR) communications increasingly fail to satisfy the massive access services of numerous user equipment brought by the increasing number of people traveling by HSRs. To this end, this paper investigates millimeter-Wave (mmWave) extra-large scale (XL)-MIMO-based massive Internet-of-Things (IoT) access in near-field HSR communications, and proposes a block simultaneous orthogonal matching pursuit (B-SOMP)-based Active User Detection (AUD) and Channel Estimation (CE) scheme by exploiting the spatial block sparsity of the XL-MIMO-based massive access channels. Specifically, we first model the uplink mmWave XL-MIMO channels, which exhibit the near-field propagation characteristics of electromagnetic signals and the spatial non-stationarity of mmWave XL-MIMO arrays. By exploiting the spatial block sparsity and common frequency-domain sparsity pattern of massive access channels, the joint AUD and CE problem can be then formulated as a Multiple Measurement Vectors Compressive Sensing (MMV-CS) problem. Based on the designed sensing matrix, a B-SOMP algorithm is proposed to achieve joint AUD and CE. Finally, simulation results show that the proposed solution can obtain a better AUD and CE performance than the conventional CS-based scheme for massive IoT access in near-field HSR communications.

Massive machine-type communications (mMTC) is envisioned to be one of the pivotal scenarios in the fifth-generation (5G) wireless communication, where the explosively emerging Internet-of-Things (IoT) applications have triggered the demand for services with low-latency and high-reliability. To this end, grant-free random access paradigm has been proposed as a promising enabler in simplifying the connection procedure and significantly reducing access latency. In this paper, we propose to leverage the burgeoning reconfigurable intelligent surface (RIS) for grant-free massive access working at millimeter-wave (mmWave) frequency to further boost access reliability. By attaching independently controllable phase shifts, reconfiguring, and refracting the propagation of incident electromagnetic waves, the deployed RISs could provide additional diversity gain and enhance the access channel conditions. On this basis, to address the challenging active device detection (ADD) and channel estimation (CE) problem, we develop a joint-ADDCE (JADDCE) method by resorting to the existing approximate message passing (AMP) algorithm with expectation maximization (EM) to extract the structured common sparsity in traffic behaviors and cascaded channel matrices. Finally, simulations are carried out to demonstrate the superiority of our proposed scheme.