An interleaved iterative-structured joint detection algorithm for active users and frequency offsets in grant-free massive access

Great-free user access is an efficient access method for massive machine-type communications (mMTC). In the massive grant-free access, frequency offsets between users and base stations lead to the degradation of active user detection and channel estimation performance. Although traditional methods achieve joint estimation by extending the perceptual matrix using the grid method, they all ignore the effect of channel fading on joint detection, which can seriously degrade the accuracy of detection. In this paper, we propose an interleaved iterative-structured-vector approximation message passing (VAMP) algorithm, which makes use of the structuring of the extended perceptual matrix, and designs a minimum mean square error (MMSE) nonlinear vector noise reducer based on the Bayesian principle to eliminate the effect of channel fading on detection. In addition, in order to improve the detection accuracy, a two-layer alternating iterative search method is proposed, which effectively overcomes the performance loss caused by the frequency offset of the grid method estimation. Simulation results show that the proposed scheme is superior in active user detection and frequency offset detection accuracy compared with the traditional schemes.