【作者】 徐胜；

【导师】 王少尉；

【作者基本信息】 南京大学 ， 通信与信息系统， 2017， 硕士

【摘要】 接入节点规划对无线通信系统具有重要的作用。合理的接入节点规划可以在保证网络性能的前提下,降低系统的总部署成本,使系统运营商在竞争中更具优势。论文主要研究云无线接入网(Cloud Radio Access Networks,C-RAN)中基带处理池规划问题和智能电网中数据汇聚节点规划问题。我们引入近似算法求解以获得最坏性能保证。论文的主要工作如下:1.研究了C-RAN部署的成本最小化问题,并同时考虑了拉远射频头(Remote Radio Head,RRH)和基带处理池的时延要求、RRH处理需求、以及基带处理池的容量限制。我们通过三个步骤求解该问题。第一步,给定基带处理池初始可行解集;第二步,对基带处理池与RRH关联问题进行松弛;第三步,引入基于局部搜索的(8+ε)-近似算法迭代更新基带处理池当前解集。仿真结果表明,该算法能够稳定收敛,且相对于遗传算法和禁忌搜索算法能够显著降低C-RAN的部署成本。2.研究了智能电网中数据汇聚节点规划问题。我们首先构建了数据汇聚节点规划问题的成本模型,综合考虑了包括数据汇聚节点与智能测量器的连接成本,数据汇聚节点的安装成本、以及同上层控制中心的连接成本等因素。其次,我们考虑了电力线信道衰减和延迟特性。我们通过引入了基于原始对偶的5-近似算法对问题进行求解。仿真结果验证了该近似算法的有效性和可靠性。最后,通过比较不同场景下不同近似比算法的规划结果,近似比好的算法所得的规划结果不一定有优势,最终结果往往取决于具体问题的形式与参数。综上所述,论文通过引入近似算法,分别研究了 C-RAN部署的成本最小化问题和智能电网数据汇聚节点规划问题。仿真结果表明,近似算法不仅可以有效解决论文所研究的网络规划问题,而且还可以得到最坏性能保障,从而为解决通信系统中大量存在的网络规划问题提供了新的解决思路。更多还原

【Abstract】 Access point planning plays a vital role in communication systems.Reasonable access point planning can save the construction cost of the system without compro-mising performance,making it possible for the system operator to stay ahead of the competition.This thesis mainly researches the Baseband unit Pool(BBU Pool)plan-ning for Cloud Radio Access Networks(C-RAN)and aggregation point planning in Software Defined Networks(SDN)based smart grid communication system.Approxi-mation algorithms are introduced to solve these problems in order to provide worst-case performance-guaranteed solutions.The contributions of this thesis have been summarized as below:1.This paper analyzes the minimal cost of C-RAN deployment problem while considering the latency requirements of the fronthaul,the traffic demand of remote ra-dio heads(RRHs)and the processing capacity of BBU Pools.The procedure of solving this problem can be divided to three steps.Firstly,giving an initial feasible solution for opening the BBU Pool.Secondly,providing the relaxed form of BBU Pools and RRHs association problem.Thirdly,the local search based(8+ε)-approximation algorithm is used to update the current solution of BBU Pool.Numerical results illustrate that the proposed algorithm not only converge stably but also reduce the cost of deploying C-RAN significantly comparing with the genetic algorithm and tabu search algorithm.2.We give the solution for aggregation points planning problem in smart grid.To begin with,we build the cost model of aggregation points planning problem while considering the connection cost between aggregation points and smart meters,the de-ploying cost of aggregation points and the linking cost from aggregation points to the control center.Moreover,the optimization problem takes account of attenuation and delay characteristics of the power line channel.The problem is solved by introducing the primal-dual schema based 5-approximation algorithm.Numerical results verify the effectiveness and reliability of the proposed algorithm.Lastly,by comparing the plan-ning results of two approximation algorithms in different scenes,we can conclude that the results of smaller approximation ratio are not necessarily better than the bigger one,the final result always depends on the form and parameters of the specific problem.In this thesis,the problem of the minimal cost of C-RAN deployment and the aggregation points planning in smart grid are studied by introducing approximation algorithms.Numerical results show that approximate algorithm not only can effec-tively solve the networks planning problems,but it also be able to get worst-case performance-guaranteed solutions,so as to provide a new perspective for a large num-ber of networks planning problems in communication systems.更多还原