Abstract:

To deal with the blind dispreading of long code direct-sequence code division multiple access (DS-CDMA) signals, this paper introduces an algorithm for joint spreading codes and information sequences estimation based on reversible jump Markov chain Monte Carlo (RJ-MCMC) by analyzing a signal model. The proposed algorithm analyzes and processes the signal models separately, and obtains the samples of distribution to be estimated through iterative sampling. It is able to construct a reversible Markov chain sampler that jumps between parameter subspaces of different dimensionality, so that the posterior distribution of parameters to be estimated is obtained. Finally, it estimates the entire spreading code and information sequence of each user by splicing. The simulation shows that the iteration converges after some twenty steps. Regardless of whether the power is equal or unequal, when the signal-to-noise ratio (SNR) is greater than -9 dB, the similarity degree between the estimated sequence and true sequence exceeds 0.95, with the bit error rate of information sequence less than 0.01. In addition, the algorithm has good adaptability for different number of users and different modulation styles. Compared with the fast-ICA algorithm, this algorithm improves the estimation performance by an average of about 3 dB.

Key words: spread spectrum communication, code division multiple access, Bayesian model, Gibbs sampler, Markov chain Monte Carlo, reversible jump