一种基于实数编码遗传算法的常模盲均衡

Constant Modulus Blind Equalization Based on Real Coded Genetic Algorithm

盲均衡可以看作代价函数优化问题。为了改进经典常模算法的性能,研究了利用实数编码遗传算法的常模盲均衡,把均衡器系数向量作为遗传算法的决策变量,采用轮盘赌选择和精英保留策略相结合的混合选择算子、算术交叉算子和非均匀变异方式,经过一系列的遗传操作,搜索到适应度值最高的个体,即均衡器的最优系数。计算机仿真结果证明了算法具有收敛速率快、能够搜索到全局最优解等特点。

The blind equalization can be seen as lus blind equalization using a real coded genetic an optimization task of cost function. A new constant modualgorithm （GA） is presented for improving the performance of the standard constant modulus algorithm （CMA）. The coefficient vector of a blind equalizer is regarded as the decision variables, and the optimal solutions can he obtained by hybrid selection operator of roulette wheel method and elitist model, arithmetic crossover and non -uniform mutation. Computer simulations show that the proposed algorithm has the advantage of fast convergence and the capability of reaching globally optimal solutions.