A New Cache Placement Strategy for Wireless Internet of Things

Abstract

Caching has shown the success in performance improvement for many wireless communications and networking systems. However, the existing researches generally decide whether cache the data or replace it rely on local content popularity on each single node. It will cause different nodes caching the same data and result in unnecessary cache redundancy. In this paper, we investigate the global optimal problem of cache placement for IoT. We first prove that finding the optimal data cache location from the whole network is an NP-hard problem, and propose a centralized algorithm to obtain the approximate global optimal solution based on the Lagrangian Heuristic Algorithm. Then, inspired by the Lagrangian relaxation, we transform the iteration procedure of finding the optimal cache location into local decisions of cache location selection and cache replacement, and we propose a distributed cache placement algorithm. Besides, the cache replacement algorithm can also be used to adjust the best cache location when the user requirement changes. Finally, we implement the distributed cache placement strategy in NDN. The experimental results show that the distributed caching strategy approximates the global optimal solution very well, and can save the network traffic by about 12.6% on average comparing with other caching strategies.