Classifier based Gateway for Edge Computing

Edge Computing technology aims to replace regular cloud IoT solutions on applications where data intensity and link latency plays critical role. Improvement is achieved by placing processing at the edge of the network, deploying service close to data source of user. Limited resources of Edge devices stipulate the need to smartly distribute over devices computational tasks, as well as to implement role switching techniques in order to guarantee smooth distribution when network conditions change. Gateway technique is proposed in this paper, providing experimental comparison of Edge Computing, Cloud Computing and Content Delivery Network (CDN) data flow scenarios where terms of network delay, service time and processing time are considered. Simulation results achieved by EdgeCloudSim software confirms the performance gain of Classifier based Edge gateway in particular balanced hardware to load ratios.

I. Introduction

In recent years, storing and processing all data in the cloud is associated with high expenses and high data cost as data volumes increase. Edge Computing is the opportunity for system architects to implement distibuted computing power from end to end by tapping into the capabilities of mobile field devices, gateways and cloud altogether [1]. It ensures near real time network response, allows operations offline or with intermittent connectivity to be implemented. Tasks pushed to edge of the network greatly impacts on service latency and response time. Network gateways are suitable location in communication path to perform these tasks.