中文摘要：

为了满足移动数据的爆炸性增长,在无线边缘部署缓存已经成为下一代移动通信的研究热点之一。业务模型与实际相符的程度直接决定了无线缓存性能评估的准确性,然而目前几乎所有关于无线缓存的文献所采用的文件请求模型并未考虑内容流行分布的动态变化。最近已有文献考虑时变流行分布,但是均采用了被动缓存策略,而主动缓存策略对无线缓存十分重要。本文分别基于一个能够描述内容流行度动态变化的文件请求模型——散粒噪声模型、和现有文献中广泛采用的独立参考模型,分析了采用被动和基于预测的主动基站缓存策略时内容流行分布动态变化对命中率和无线网络吞吐量的影响。仿真结果表明,采用散粒噪声模型对评估无线缓存系统吞吐量的影响很大,特别是当网络中干扰较小时;另外,已有文献基于被动缓存策略得到的结论并不适于主动缓存策略。

英文摘要：

To satisfy the explosive growth of mobile data,caching at wireless edge has attracted significant research efforts for the next generation of the wireless communication. To which extent the traffic model matches the real traffic largely affects the accuracy of performance evaluation on wireless local caching. However,the content request model adopted in almost all existing studies does not take the content popularity dynamics into consideration. Although there were several works that considered dynamic popularity,only reactive caching policy was evaluated. Yet proactive caching is important for wireless caching networks. This paper analyzes the impact of dynamic popularity on the cache hit ratio and average network throughput with proactive and reactive caching policies,based on Shot Noise Model（ SNM） that can capture the content popularity dynamics,and Independent Reference Model（ IRM） that is widely used in the literatures. Simulation results show that whether or not using SNM has large impact on the evaluated throughput of wireless networks,especially when the interference level is low. Moreover,the conclusions obtained from reactive caching policies are not applicable to proactive caching policies.