论文部分内容阅读
为了避免传统最小均方/最小二乘(LMS/LS)盲均衡方法的系数矩阵病态问题、减少最小均方(LMS)和最小二乘(LS)方法对数据长度的严重依赖,提出一种基于脊回归(RR)方法的光通信系统电域盲均衡算法。通过RR盲均衡代价函数的构建,盲均衡问题转化为求解无约束优化问题并完成求取均衡器的数学推演。分析了该算法的复杂度和脊参数对于代价函数及其性能的影响,构造批处理形式迭代算法求解该优化问题。该方法可将多进制相移键控(MPSK)和正交幅度调制(QAM)系统盲均衡问题纳入统一的框架。最后,通过仿真验证了新方法在性能上优于传统LMS/LS类盲均衡算法。
In order to avoid the ill-posed coefficient matrix ill-posed problem of the traditional Least Squares / Least Squares (LMS / LS) blind equalization method and reduce the severe dependence on the data length of the least mean square (LMS) and least square (LS) methods, Sphere Regression (RR) Method for Optical Domain Blind Equalization Algorithm in Optical Communication Systems. Through the construction of RR blind equalization cost function, the problem of blind equalization transforms to solve the unconstrained optimization problem and completes the mathematical derivation of equalizer. The complexity of the algorithm and the influence of ridge parameters on the cost function and its performance are analyzed. An iterative algorithm in batches is constructed to solve the optimization problem. This method can bring the problem of blind equalization in multi-ary phase shift keying (MPSK) and quadrature amplitude modulation (QAM) systems into a unified framework. Finally, the simulation results show that the new method is superior to the traditional LMS / LS blind equalization algorithm.