一种移动内容分发网络的分层协同缓存机制

With the rapid growth of network traffic in recent years, edge computing has become an important method to reduce network load, increase network capacity and improve service quality. In this context, mobile content delivery network (mobile CDN) has been gradually adopted by ISPs. Within a mobile CDN, cache nodes are deployed in the mobile network, i.e., the radio access network (RAN). In the LTE network, the cache nodes in RAN sink to the base station. The main feature of RAN cache is that the single node only covers a small number of users; the storage capacity of a single node is limited, while there are many nodes in RAN. Therefore, it is necessary to perform cooperative caching. Besides, users share their cache space through wireless network by cooperative caching. Compared with traditional CDN system, the mobile CDN newly adds a cooperative caching layer in RAN and a cooperative layer on user side. So far, there is no exact solution on which caching allocation and replacement strategy should be used and how to coordinate between multi-layer caches in mobile CDN system. In this paper, we establish an optimization model for mobile CDN hierarchical cooperative cache system, aiming to minimize the overall link cost of user access to resources. Based on this model, we proposed a heuristic hierarchical utility-based cooperative caching (UCC) strategy based on real user demands. UCC makes cache decisions according to the global utility value of the cache resource to reduce the cost of resource access as much as possible. Simulation results and trace-based verification experiments show that the proposed strategy can significantly reduce the average link cost of resource access, effectively improve the cache hit rate. The main contributions of this paper are as follows: (1)For the first time, this paper builds a unified model of the multi-level cooperative caching system in mobile CDN scenario. In this model, the core network cache, the RAN cooperative cache and the user-assisted cooperative cache are united to solve the resource delivery and cache replacement problem of cache nodes in the same level and between different levels.(2) According to this model, the problem is closely related to a knapsack problem which is known as NP-Complete problem and hard to solve. Assuming that the optimal solution can be obtained with linear time, many operations such as resource transfer and cache replacement need to be performed between the optimal solutions at different moments, which would result in heavy transmission load and mobile data consumption. Thus a utility-based cooperative caching strategy UCC is proposed to make user access resources as close as possible. (3)Simulation experiments and trace-based verification experiments are designed to verify the performance of the proposed strategy, and other currently widely used caching strategies are implemented in mobile CDN system for comparison. Simulation results show that UCC can reduce the link cost by 10% and improve cache hit rate by 20% compared with the cooperative list-based strategy cLRU(m). Furthermore, we design trace-based verification experiment using real data from China Mobile. The experimental results show that the proposed strategy still has obvious advantages over other widely used caching strategies under the conditions of different load.