Joint Activity and Data Detection for Asynchronous Grant-Free Access in NTN

Grant-free-based non-terrestrial networks (GF-NTN) hold immense potential for future networks due to the benefits of wider coverage and lower access latency. However, it is challenging to estimate active devices and data since synchronization cannot be guaranteed in GF-NTN. To address this issue, a multiple-slots approximate expectation propagation (MS-AEP) detector is proposed. Concretely, we first conceive an asynchronous detection model for GF-NTN. Based on the model, we propose a symbols-based approximate expectation propagation (AEP) detector, utilizing the property of discrete constellation symbols. Then, to fully exploit the structured sparsity for performance enhancement, we extend the AEP detector to the proposed MS-AEP detector. Furthermore, a power-assisted solution is developed to estimate unknown delay based on MS-AEP. Simulation results show that the proposed detector approaches the performance of Oracle least squares detector in asynchronous detection for GF-NTN scenario.