改进蚁群算法求解最短路径问题

针对蚁群算法在求解最短路径问题时存在容易陷入局部最优解的问题,对经典蚁群算法提出三方面改进。首先,在初始化信息素浓度时加入方向引导,加快初始搜索速度;其次,在局部信息素浓度更新过程中采用信息素重分配思想,避免由路径信息素衰减过程导致的最优路径信息素浓度过分减少;最后,在全局信息素更新过程中引入动态因子,使其自适应地更新较优路径信息素浓度,以提高全局搜索能力。仿真实验结果表明,该改进算法可以保证收敛速度,并提高算法搜索到最优路径的几率。     

Propagation and self-healing properties of Lommel-Gaussian beam through atmospheric turbulence

The superposition of basic non-diffracting beams triggered new research hotspots lately, laying opportunities for long-distance wireless optical communication. The Lommel-Gaussian (LMG) beam formed by the superposition of Bessel-Gaussian light not only possesses non-diffraction feature, but also has tunable symmetry. With the help of Poynting vector analysis, we observed a smaller radial energy flow component during the propagation of the high order symmetrical LMG beam, which allows it to maintain the original beam profile over long distance. Thanks to the energy oscillation of the mainlobe and sidelobes, the mainlobe blocked by the symmetrical LMG beam can be restored. Also, the random phase screen with angular spectrum method is used to describe the beam behaviors in turbulence. The results show that the symmetry LMG is preferred in free space optical communication, and the asymmetric LMG performs poorly due to asymmetric energy transfer.     

基于记忆因子的连续相位调制信号最大似然调制识别

为解决有记忆非线性的连续相位调制(CPM)信号调制方式识别精度低的问题,该文提出一种基于记忆因子的CPM信号最大似然调制识别新方法。该方法定义具有时齐马尔科夫性的映射符号,通过计算其后验概率构造记忆因子,进一步结合CPM分解和EM算法,推导出时间可分离,信道参数可估计的CPM信号似然函数。该调制识别方法所需符号数目少,适用信噪比范围广,识别CPM信号种类多且精度高,对相位误差鲁棒性强。仿真结果证明,当符号数目为200,信噪比为0 dB,相位误差任意时,该方法对8种CPM信号的识别率可达95%以上。