内容中心网络的缓存放置策略研究

发布时间：2018-06-01 08:39

本文选题：内容中心网络 + 缓存放置策略　；参考：《哈尔滨工程大学》2014年硕士论文

【摘要】：内容中心网络是未来互联网的一种新型体系结构。内容中心网络缓存机制的特点是网络中任何节点都具有缓存空间,且节点对经过其进行转发的内容不加区分全部进行缓存。内容中心网络的缓存机制容易产生内容冗余度高、缓存利用率低等问题。缓存放置策略是解决此类问题的有效方法。针对内容中心网络缓存机制存在的不足,本文对现有的缓存放置策略进行详细分析,提出基于预测的最优化缓存放置策略。首先,将缓存放置问题转化为最优化问题,综合考虑影响缓存性能的几个因素,构建最大化收益缓存放置模型(Max-Benefit模型),针对Max-Benefit模型中对象被访问频率不能表示对象将来的热度趋势这一问题,在Max-Benefit模型中引入预测机制,提出基于预测的最大化收益缓存放置模型(PB-Max-Benefit模型),通过对对象将来的热度趋势进行预测实现尽可能多得对热门内容的缓存,避免无效缓存的发生,提高缓存性能。然后,基于PB-Max-Benefit模型改进标准遗传算法的选择算子、交叉算子和变异算子,对其进行求解,通过对标准遗传算法进行改进来提高它的收敛速度,避免陷入所求问题的局部最优解,提高求解PB-Max-Benefit模型全局最优解的性能。最后,使用NS-3网络模拟器在虚拟机中进行仿真实验。以缓存命中率、无效缓存率和网络平均跳数作为验证指标,在不同缓存大小、数据访问模式以及网络规模的仿真环境下,通过与现有的最优化缓存放置策略以及CCN缓存机制进行仿真对比,得到仿真结果,将仿真结果以图表的形式表示出来并对其进行分析,最终验证本文提出的基于预测的最优化缓存放置策略的有效性。
[Abstract]:The content center network is a new architecture of the Internet in the future. The feature of content-centric network caching mechanism is that any node in the network has cache space and nodes cache all content transmitted through it without distinction. The cache mechanism of content-centric network is prone to the problems of high content redundancy and low cache utilization. Cache placement strategy is an effective way to solve this problem. In view of the shortcomings of the content center network caching mechanism, this paper analyzes the existing cache placement strategies in detail, and proposes an optimized cache placement strategy based on prediction. First of all, the cache placement problem is transformed into an optimization problem, and several factors affecting cache performance are considered synthetically. The Max-Benefit model is constructed to maximize revenue cache placement. Aiming at the problem that the object access frequency in the Max-Benefit model can not represent the future heat trend of the object, a prediction mechanism is introduced into the Max-Benefit model. This paper proposes a PB-Max-Benefit model based on prediction to maximize revenue cache placement. By predicting the future heat trend of objects, we can cache as many popular content as possible, avoid the occurrence of invalid cache, and improve the cache performance. Then, based on the PB-Max-Benefit model, the selection operator, crossover operator and mutation operator of the standard genetic algorithm are improved, and the convergence rate of the standard genetic algorithm is improved to avoid falling into the local optimal solution of the problem. The performance of solving the global optimal solution of PB-Max-Benefit model is improved. Finally, the NS-3 network simulator is used to simulate the virtual machine. With cache hit ratio, invalid cache rate and average hop number of network as the verification index, under different cache size, data access mode and network scale simulation environment, By comparing with the existing optimal cache placement strategy and CCN cache mechanism, the simulation results are obtained, and the simulation results are represented in the form of charts and analyzed. Finally, the effectiveness of the proposed optimal cache placement strategy based on prediction is verified.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TP393.02

【引证文献】