The rapid growth of wireless data has caused cellular networks (3G/LTE/4G) to be overburdened, which affects the user’s quality of experience. In a mobile edge computing (MEC) architecture, caching the content at the base stations cooperatively is a prudent solution and this reduces the user-perceived latency as it brings the content closer to the user and minimizes the burden on the backhaul. However, to enhance the quality of service in delay-sensitive and time-critical applications, the requested content should be served within the deadline. Therefore, maximizing the storage utilization while maximizing the saved delay in the mobile edge network is a critical problem. To address these challenges, in this paper, we formulate a cache placement problem in mobile edge networks as placing the contents at different MECs (base stations) to maximize the saved delay with capacity and deadline constraints. The problem is modeled as an integer linear programming problem for content placement in mobile edge network. A relaxation and rounding method is presented to solve the integer linear programming problem. Further, we propose a fuzzy logic based caching algorithm that considers deadline, benefit and content request prediction in caching decisions. In the proposed algorithm, the echo state network is used to predict content request distribution. Extensive simulation results show that the proposed fuzzy caching scheme significantly improves the performance in terms of acceleration ratio, hit ratio and the number of files satisfying deadline on MovieLens dataset as compared with three existing caching techniques.

无线数据的快速增长导致蜂窝网络（3G / LTE / 4G）负担过重，从而影响了用户的体验质量。在移动边缘计算（MEC）架构中，谨慎地在基站上缓存内容是一种审慎的解决方案，这减少了用户感知的等待时间，因为它使内容更接近用户，并最大程度地减少了回程负担。但是，为了提高对时延敏感和对时间要求严格的应用程序中的服务质量，应在截止日期之前提供所请求的内容。因此，最大化存储利用率同时最大化移动边缘网络中节省的延迟是一个关键问题。为了应对这些挑战，在本文中，我们将移动边缘网络中的缓存放置问题公式化为将内容放置在不同的MEC（基站）上，以在容量和截止日期约束条件下最大程度地节省延迟。该问题被建模为用于移动边缘网络中的内容放置的整数线性规划问题。提出了一种松弛和舍入方法来解决整数线性规划问题。此外，我们提出了一种基于模糊逻辑的缓存算法，该算法考虑了缓存决策中的截止日期，收益和内容请求预测。在提出的算法中，回声状态网络用于预测内容请求的分布。大量的仿真结果表明，与现有的三种缓存技术相比，所提出的模糊缓存方案在加速比，命中率和满足MovieLens数据集的截止日期的文件数量方面显着提高了性能。