一种基于连续蚁群算法的快速最大似然DOA估计

针对传统最大似然估计计算量大的问题,将连续空间蚁群算法与最大似然估计算法相结合,在ACOML算法的基础上,提出了一种用混沌序列初始化状态空间的改进蚁群算法MACOML(Muddleheaded ACO)。该方法使用混沌映射产生的初始状态空间来代替ACOML算法中的随机序列产生的初始状态空间,增加了初始解的遍历性,同时在寻优过程中增加了局部搜索。仿真结果表明:MACOML能保持最大似然估计方法的高分辨性能,而计算复杂度只是最大似然方法的1/20。

Fast Estimation of Maximum Likelihood DOA Based on Continuous Ant Colony Algorithm

Aiming at problem of great amount of computation for maximum likelihood estimation, an improvement on ant colony algorithm（ Muddleheaded ACO） by using chaotic sequence to initialize state space is presented on basis of ACOML algorihtm with combination of continuous space ant colony algorithm with maximum likelihood estimation algorithm. The method uses initial state space generated by chaotic map to substitute the initial state space generated by random sequence in ACOML algorithm, and ergodicity of initial solution is added, and local search is added during looking for optimization. The simulation results indicate that using MACOML can retain high resolution capability of maximum likelihood estimating method, and the computing complexity is only 1/20 of that using maximum likelihood method.