Blind despreading and deconvolution of asynchronous multiuser direct sequence spread spectrum signals under multipath channels

• Published in: IET signal processing 2023-05, Vol.17 (5), p.n/a
• Main Authors: Li, Liangliang, Zhang, Huaguo, Du, Songmao, Liang, Tao, Gao, Lin
• Format: Article
• Language: English
• Subjects: blind equalisers; blind source separation; channel estimation; code division multiple access; Gaussian distribution; least squares approximations; matrix decomposition; signal processing; spread spectrum communication; statistical distributions
• ISSN: 1751-9675; 1751-9683
• DOI: 10.1049/sil2.12220

In non‐cooperative scenarios, the spreading sequences or waveforms of the direct sequence spread spectrum (DSSS) signals is unknown to the receiver. This paper focuses on addressing the problem of blind estimation of the spreading waveform under multipath channels. In the scenario of direct signal path transmission, the spreading sequences can be directly obtained based on the estimated spreading waveforms. However, in the presence of multipath channels, the spreading waveform becomes the convolution of the spreading sequence and channel response, thus deconvolution should also be performed after estimating the spreading waveforms. In order to perform blind despreading and deconvolution of asynchronous multiuser DSSS signals under multipath channels, the authors propose to exploit the finite symbol characteristics of information and spreading sequences and then the iterative least square with projection method is adopted. Besides, the Cramer‐Rao bound of spreading waveforms is derived in such a circumstance as a performance benchmark. The effectiveness of the proposed method is verified via simulation experiments. An iterative least square with projection (ILSP)‐based approach is proposed to blind despreading of multiuser direct sequence spread spectrum (DSSS) signals. The channel response of the pseudo‐noise (PN) waveform is estimated first to despread DSSS signals. Then, the PN sequences and channels are estimated by deconvolution using subspace and ILSP methods, respectively. An optimisation method is given for the actual channel to achieve a lower channel response estimation error. The Cramer–Rao bound of the channel response of the spreading sequence is derived as a lower bound for the theoretical error of the estimated quantity to evaluate the performance of the algorithm.