自适应并行蚁群算法在TCP网络速率控制中的应用  

Application of Adaptive Parallel Ant Colony in Rate Control in TCP Network

作　　者：李向丽[1] 周林成[2] 

机构地区：[1]江苏信息职业技术学院,江苏无锡214153 [2]无锡机电高等职业技术学校,江苏无锡214028

出　　处：《机械制造与自动化》2009年第5期92-94,共3页Machine Building & Automation

摘　　要：针对原始蚁群算法搜索能力不强、易陷入局部最优的问题,提出了一种自适应并行蚁群算法。引入共享函数解决算法易陷入局部最优的问题。给出"聚度"的概念,使算法能充分利用学习机制、强化最优信息的反馈。将新算法应用到TCP网络速率控制中,去掉了以往研究中对效用函数的严格假设,最大化用户建立了新的效用函数。算法中链路从网络获知拥塞链路的条数,用户根据对应的效用函数和拥塞反馈信息调整自身速率。仿真表明了算法可以很快地收敛到最优速率。This paper proposes that the self-adaptive parallel ant colony algorithm is used to solve the local optimum problem of the original ant colony algorithm. This proposed algorithm makes full use of learning mechanism and intensifies the feedback of the optimal information. When the ant is stagnated, the shareware function is used to jump out from the local optimum. When this new algorithm is applied to rate control in TCP network, we remove restrictive assumptions on utility function and propose a simple distributed algorithm for achieving the optimal rates based on the network utility maximization framework. In our algorithm, the network communicates to the user the number of congested links on the user＇s path, and the user adjusts its rate accordingly, taking into account its utility function and the network congestion feedback. Numerical example shows that our algorithm converges to the optimum rates.

关 键 词：并行蚁群方法 共享函数 速率控制 效用函数 

分 类 号：TP393[自动化与计算机技术—计算机应用技术]