Decentralized Resource Auctioning for Latency-Sensitive Edge Computing

The prevalence of Internet of Things (IoT) in contemporary settings has induced systems composed of heterogeneous devices, computing infrastructures, and cloud services. New paradigms have emerged where computational resources are managed closer to IoT end-devices, within a general theme of decoupling from the cloud. This is because meeting application demands must occur at runtime, in the face of uncertainty and in a decentralized manner. Taking advantage of available resources closer to devices calls for novel resource allocation techniques that comply with latency, privacy and decentralization demands of IoT applications. To this end, we propose a novel decentralized resource management technique and accompanying technical framework for the deployment of latency-sensitive IoT applications on edge devices. Our technique is inspired from the functionality of an auction house and has two objectives; (i) find a deployment mapping for an arbitrary application, compliant with its individual resource requirements and latency constraints, (ii) facilitate privacy, as each device participates at their own will, based on its own availability and privacy preferences. Our approach ensures seamless deployment at runtime, assuming no design-time knowledge of device resources or network topology.