Edge-centric resource allocation for heterogeneous IoT applications using a CoAP-based broker

Edge/fog computing is an emerging paradigm that exploits computing and management at the edge of the network to improve QoS of traditional cloud-centric IoT applications. In large-scale IoT deployments, multiple applications are expected to access the same constrained resource (e.g., a sensor) with heterogeneous QoS requirements. However, IoT applications may be unable to adapt their access patterns to cope with bandwidth limitations. To address this issue, we propose a fog-based IoT broker that determines suitable notification rates to access IoT resources while maximising applications' QoS and avoiding network congestion. We first formalise the rate selection as an optimisation problem, and then we propose a practical algorithm that leverages transmission reliability to dynamically select notification periods. We developed a CoAP-based prototype and validated our proposal through simulations and experiments in a real IoT testbed. Our broker provides higher QoS satisfaction, throughput and energy efficiency compared to a conventional CoAP proxy.