Service aware resource management into cloudlets for data offloading towards IoT

Abstract

Task offloading in edge cloud and self-organized distributed cloudlet design are two significant research areas to explore internet of things (IoT) setup in mobile cloud computing diligence through network function virtualization. The cloud service providers already set up metropolitan level cloudlet connectivity. In proposed senario by applying a virtual network function to execute various user demands with different IoT concerned service-related difficulties have presented. The objective is to exploit the number of self-proclaimed requests via load harmonizing to reduce the cohesive cost of entries within a period of time. A problem function is designed to offload a set of tasks in a metropolitan area network, where a particular network related function with a maximum bearable delay is requested by each sensor data computation through different network services according to different offloading requests. An algorithm is designed to make a reduced bipartite graph with maximum matching and minimum weight. To save cost, an effective prediction offloading algorithm has been developed where instances of network functions get created and/or released in different cloudlets considering that the task offloading request patterns dynamically changes. Later an analysis concludes that the possible outcome of the proposed algorithm is efficient in term of delay reduced while distributing the load among the cloudlets into the edge-cloud. The effective cost has been reduced up to approx. one-third and execution time has reduced up to one-tenth with the help of proposed queue management system design which helps to utilize the cloudlet instances ineffective way.