Efficient Hosting of Robust IoT Applications on Edge Computing Platform

摘要

Demanding IoT application requirements such as high dependability and low latency, cannot be satisfied by centralized cloud computing when deploying these applications. Edge computing is emerging as an alternative to deploy demanding IoT applications closer to the edge of the network. However, with edge computing, available resources are distributed among different resource-constrained devices, which cannot host large monolithic applications. We propose a new hierarchical IoT application model, suitable for the distributed nature of edge computing. Thus, a task in the application is modeled using multiple configurations of smaller tasks, each with their own functionality level and resource requirements. For deployment we use a decentralized resource technical framework that finds a satisfiable task mapping on edge devices. Its functionality is inspired by an auction house, having the objectives of (i) deploying an application such that its requirements are met and (ii) empowers edge devices to be in control of their available resources. For the latter, we propose a new decision policy to help edge devices take better decisions regarding the use of local available resource. Our solution enables efficient device resource utilization when deploying IoT applications at the edge.