【作者】 卢金；

【导师】 刘国亮；

【作者基本信息】 吉林大学 ， 系统工程， 2007， 硕士

【摘要】 随着互联网的飞速发展,业务种类和业务量与日俱增,网络对传输质量提出了更高的要求。网络的生存性在提供业务质量保证方面的作用也越来越受到关注。为了改善现有网络的服务质量,也为了更科学地建设未来的网络,加大网络建设的投资是不可避免的。因而,对于通信运营者来说,如何在通信网质量和建设、维护和费用之间找到一个平衡点无疑是一个值得考虑的问题。正是基于此,本文将对空闲容量优化配置问题进行研究。网状结构光传送网络的保护容量优化问题一直是业界研究的焦点,目前人们基于不同的优化标准已经提出了多种解决方案。运用上述方案尽管可以获得可行的优化结果,但它们固有的缺陷限制了其应用环境。为此,本文提出了一种新的基于遗传算法的网状光传送网络保护容量需求优化问题的求解方案,它能够快速、有效地搜索到问题优化解和优化目标结果。本文首先比较全面的介绍了光通信技术、网络生存性理论,以及遗传算法的工作机理。并对光传送网络空闲资源配置优化设计的相关问题进行了研究,给出了网络发生故障后的恢复流程,为不同的恢复方法进行了分类,对预留资源和动态路由的方法做了简单分析。然后,以保护容量造价需求作为优化目标函数,对环形和网状光传送网络分别建立了各自的优化模型,尤其是在网络结构光传送网络中,一些对保护容量需求影响较大的生存选项得到了优化考虑,如残余释放、保护容量使用方法等。最后通过仿真实验结果表明遗传算法能够在较短时间内求得光传送网络保护容量需求优化问题的近似最优解,性能良好。更多还原

【Abstract】 As the fast development of internet and the increase of operation categories and quantity, we expect better and better internet transmission quality. Thus the network survivability technology which can guarantee the transmission quality drives more and more attention. In order to improve the current quality of the network and to construct the future network in a more scientific way, it is necessary to present more investment. Therefore, for the communication operators, how to get a good tradeoff between the improvement of the communication network and the cost is undoubted a problem worth consideration. The main job of this thesis is to study how to achieve optimal collection of the space capacity.The optimal collection of the protection capacity is always a focus, and on which researchers have proposed several methods based on different optimal standard. B. Van Caenegem and his coworkers proposed the lowest cost model of the whole WDM network including the distribution of the optical nodes, the optical cables arrangement and the distribution of the wavelength channel ect.; and based on the model, they performed the calculation using linear programming and simulating anneal algorithm. While S.Ramamurthy and B. Mukherjee proposed the optimal model according to the source of the wavelength channel required by the working optical path and the protection optical path; also they performed some simulation about the special path protection, the shared path protection and the shared chain route protection on the CPLEX platform using linear programming. Besides, other method emphasized particularly on converting the irregular network topology to regular topology or linear topology, so that the mature and easy-performed APS and loop network protection techniques can be adopted to capacity optimization. All of the above methods can obtain feasible optimal results, but their intrinsic disadvantages limit their application range. The first method requires a lot of initial parameter, some of which can only be obtained fromcomplex calculation rather than the current network topology. The second method can only applied to small-scale network, and it only considers the chain route constructed by single optical cable, i.e. for any wavelengthλ(λ= 1,2, ,λmax), the channel can only be unique. The topology converting method requires complicate algorithm designing and solving, so it is not easy to perform. This thesis proposed a new calculation method for the protection capacity requirement optimization problem based on genetic algorithm, which can find out the optimal solution in a fast and efficient way.First of all, we generally introduce the existed survivability and anti-damage technologies, present the recover scheme, category the recover methods, and give a simple analysis on prestored sources and dynamic routing. Secondly, we construct the optimal models for annular and reticulate optical transmission network, taking the cost of the protection capacity requirement as the optimized object function. Especially in the reticulate network, some options which have greater influence on the protection capacity requirement get optimally considered, for example remains release and the use of protection capacity. Also, we introduce the principle of generic algorithm and the function of it in solving combined optimization problem. We study how to use generic algorithm to solve reticulate network protection capacity optimization problem, and the main content including: the mathematic model of the optimal problem, the gene coding configuration of chromosome, inheritance operation regulations (selection operator, amphimixis operator and aberrance oprator) and the algorithm’s breaking condition. The simulation results imply that the generic algorithm can obtain the approximate solution of the optimization problem in a shorter time.更多还原