一种基于隐马尔可夫的盲自适应多用户检测算法

针对在概率统计为基础的信道估计收敛速度慢，状态估计存在非合理性的问题，提出了以隐马尔可夫模型为基础的盲自适应多用户检测算法。在后向方法中采用了固定延迟和锯齿延时相结合的办法，根据前后序列之间的相关性和最陡下降法对多用户上行复合信道进行盲估计，对于发射序列状态之间可能存在时序非继承性，搜索与前向序列对应的具有最大转移概率的后续序列，实现了最大后验多用户检测方法。此算法避免了由于相邻状态序列后向概率相差很大而造成的信道估计发散，同时将计算复杂度从o(N^3)降低到o(N^2)，计算机仿真实验表明，该算法将信道响应估计速度提高了10％，且具有全局收敛性和系统的稳定性。

Blind Adaptive Multiuser Detection Based on Hidden Markov Models

A new blind adaptive multiuser detection based on the hidden Markov models is presented for the slow convergence of channel estimation and the unreason of state estimation in the probabilistic algorithms. The backward algorithm adopts the scheme of joining the fixed lag to sawtooth lag. The mutli channel of multiuser is estimated by using the steepest descent algorithm and the correlation between the previous and the following sequence. For seeking the following sequence of maximal transition probability corresponding to previous sequence, a maximum posteriori multiuser detector has been constructed. The algorithm avoids the divergence in the estimation channel and decreases computational complexity. The computer simulation shows that the new algorithm increases the speed of channel estimation by 10%, and has the convergence of global and the stability of system.