Optimal Probabilistic Caching in Heterogeneous IoT Networks

Abstract

With the development of hardware and communication technology, Internet-of-Things (IoT) devices have become more powerful in computing, caching, and communication. The enhancement of equipment makes it possible for IoT devices to act as caching helpers in cache-enabled networks to address the conflict between the limited caching resources and the sharply increasing data traffic. However, how to efficiently utilize the caching resources of IoT devices and edge servers is a challenging problem especially when contents are of different sizes. In this article, we focus on the probabilistic caching for contents of different sizes in heterogeneous IoT networks, aiming at improving the offloading rate for backhaul links. We present a mathematical framework to formulate the hit probability of cache-enabled IoT networks based on stochastic geometry and propose an improved caching probability conversion (CPC) algorithm to derive the closed-form solutions of optimal caching probabilities. Moreover, we further extend the network model with serving capacity constraint, i.e., caching devices can only serve a limited number of users simultaneously and solve the corresponding nonconvex problem by the difference of convex (DC) programming. We validate our analytical framework by the Monte Carlo method and give extensive numerical results to compare the performance of the proposed strategy with that of two existing caching strategies.