Abstract:

The performance of multi-user detection in the low earth orbit satellite random access system deteriorates seriously over rapidly time-varying channels. To solve this problem, a multi-user detection algorithm is proposed to overcome the large Doppler frequency shift. The algorithm uses the basis expansion model for modeling the channel of each user and the basis coefficients of each user are jointly solved by the maximum likelihood method. Then, the channel of each user is fitted by the basis coefficients. After that, all interference packets are reconstructed and eliminated. Finally, the residual signals after interference cancellation are decoded. To improve the detection probability of the user's packets, the channel estimation and interference cancellation are repeated using the decoded bits, each user's training blocks and the detected user data. Simulation results show that, compared with the existing algorithm, the algorithm has significantly improved the performance of the packet error rate, channel estimation mean square error and throughput. The algorithm can work efficiently in low earth orbit satellite multi-user scenarios, which have a high signal noise ratio and a large Doppler frequency shift.

Key words: satellite, multi-user detection, maximum likelihood, interference removal