Incentivising Resource Sharing in Edge Computing Applications

Abstract

There is increasing realisation that edge devices, which are closer to a user, can play an important part in supporting latency and privacy sensitive applications. Such devices have also continued to increase in capability over recent years, ranging in complexity from embedded resources (e.g. Raspberry Pi, Arduino boards) placed alongside data capture devices to more complex “micro data centres”. Using such resources, a user is able to carry out task execution and data storage in proximity to their location, often making use of computing resources that can have varying ownership and access rights. Increasing performance requirements for stream processing applications (for instance), which incur delays between the client and the cloud have led to newer models of computation, which requires an application workflow to be split across data centre and edge resource capabilities. With recent emergence of edge/fog computing it has become possible to migrate services to micro-data centres and to address the performance limitations of traditional (centralised data centre) cloud based applications. Such migration can be represented as a cost function that involves incentives for micro-data centres to host services with associated quality of services and experience. Business models need to be developed for creating an open edge cloud environment where micro-data centres have the right incentives to support service hosting, and for large scale data centre operators to outsource service execution to such micro data centres. We describe potential revenue models for micro-data centers to support service migration and serve incoming requests for edge based applications. We present several cost models which involve combined use of edge devices and centralised data centres.