【作者】 陈存香；

【导师】 王俊峰；

【作者基本信息】 北京交通大学 ， 交通信息工程与控制， 2009， 硕士

【摘要】 SCDMA(Synchronous Code Division Multiple Access--同步码分多址)接入技术采用了智能天线、软件无线电等先进技术,是一个全新的我国拥有完整自主知识产权的无线通信技术标准。SCDMA系统中基站分布规划的研究对于后期网络建设和系统优化有着重要的理论意义和实用价值。SCDMA系统基站分布规划属于多目标组合优化问题,需要在满足目标区域的最小场强达到通信要求下,还必须满足业务质量要求,业务覆盖要求,以及降低经济成本。然而这些目标之间往往存在着一定的冲突矛盾。因此在求解过程中,要设法平衡各个目标或找到解决冲突的方法。本文采用了子微粒群的概念,加强了共享信息,降低了计算代价。实验证明该方法更加灵活有效的捕捉到了最优解。微粒群优化算法的诸多优势适于解决多目标优化问题。但是种群的多样性随着时间的增加而下降过快,容易陷入局部最优解。为了降低在初期算法陷入局部最优解的可能性,本文融合了遗传算法中的轮盘赌选择算子。实验证明该方法能在早期抑制微粒的“早熟”,得到较快的收敛速度。当算法即将收敛时,种群中的个体大部分都集中在局部最优解或全局最优解附近,即发生早熟现象。为了使个体跳出该局限区域,本文应用了爬坡算子,加强其爬坡能力,加快其跳出局部最优解的能力。实验证明爬坡算子的引入在后期最大程度上避免了“早熟”的发生。更多还原

【Abstract】 SCDMA(Synchronous Code Division Multiple Access) technology is a completely new technical standard for wireless communications.Because it uses smart antenna,software radio and other advanced technology With independent intellectual property rights.It attaches theoretical and practical value to post-network construction and system optimization by the research of SCDMA base stations’placement.The aim of SCDMA base stations’placement is not only to meet minimum field strength in the target region,but also to meet the quality requirements,business coverage,as well as the reduction of the economic cost.However,these objectives are often conflictive.Thus in the solving process,we should try to balance the various objectives or to find a solution to them.By the concept of sub-particle swarm,it enhances information sharing and reduces the computational cost.Experiment shows that it is more flexible and effective to capture the optimal solution.Particle Swarm Optimization wins great advantage over multi-objective optimization problem.Nevertheless,the diversity of the population is prone to decrease with the passage of time,thus it is easy to fall into local optimal solution.In order to reduce the possibility into the local optimal solution at an early stage,the roulette selection operator often used in the genetic algorithm is introduced in this paper. Experiment shows that through this method "prematurity" is inhibited and convergence speed become faster.When the algorithm is about to converge,most individuals are centralized in the area of local optimal solution or near global optimal solution.In other words," prematurity " has occurred.In order to escape the limitation of individual regions, climbing operator is adopted in this paper,which could strengthen its climbing ability and accelerate the possibility of moving away local optimal solution.Experiment proves that at later stage the introduction of climbing operator avoids the " prematurity" to maximum extent.更多还原