Towards Optimising Fog Caching Using Deep Reinforcement Learning

The Internet of Things (IoT) has gained increasing popularity in recent years. However, these systems have limitations related to transient data creation and limited energy reserves. Furthermore, compliance with quality of service (QoS) measures, such as an adequate response time, remains a critical factor in such environments. In this context, caching IoT data at fog nodes can help to meet QoS standards while avoiding the inherent limitations of IoT networks. In fact, fog caching has emerged as a promising method for the Internet of Things to provide proximity services and reduce latency and network traffic. This paper proposes a deep reinforcement learning (DRL) solution to the problem of IoT data caching at fog nodes. The objective is to increase the caching success rate of IoT networks in order to reduce energy consumption as well as latency in fog computing by taking into account freshness and limited lifespan of data.