QoE-aware Edge Computing for Complex IoT Event Processing

Computing offloading is a promising technique for Internet of Things (IoT) to handle complex IoT events processing, e.g., voice recognition. Existing work mainly focuses on offloading in mobile networks, while little addresses the problems of heterogeneity and quality of experience (QoE) on resource constrained devices in edge networks. In this paper, we present a framework of QoE-aware edge computing for complex IoT event processing. Developers can offload their existing methods in edge networks with little modifications. We design a scheduling algorithm to minimize the offloading latency given user specified QoE. In addition, we present a scheduler selection algorithm to select the most suitable scheduler at runtime. We implement the above algorithms and a real-world RPC function on heterogeneous devices. We evaluate our framework and results show that our framework can save up to 91.64% of the execution time in a small edge network.