Edge-to-Cloud Adaptive Offloading for Next-Generation Services

Abstract

Due to the continuous growth of user traffic demands, there is a need to cope with the increased processing and storage requirements. The increased number of connected end devices causes a problem with carrying the generated loads efficiently. Edge and cloud computing and storage are real solutions to overcome the limitations of end devices on many prospective including computation capabilities, storage capabilities, and power consumption. Terminal devices offload their overflow tasks to the cloud for processing, analysis, and storage. This paper aims to improve computation offloading from edge nodes to the cloud in Internet of Things networks by making efficient decisions using an adaptive offloading algorithm. Offloading is controlled by a processing time offloading threshold value, which is determined automatically by edge nodes based on their traffic intensity and adaptively increased or decreased in loads. The proposed algorithm had been programmed and simulated; experimental evaluations show that the proposed adaptive offloading algorithm minimizes the edge mean response time by up to 58% and the cloud mean response time by up to 25% compared to the existing fixed, pre-defined offloading threshold value.