A Delay- and Power-optimized Task Offloading using Genetic Algorithm.

Internet of Things (IoT) introduces the Big Data era as the IoT devices produce massive amounts of data daily. Since IoT devices contain limited computational and processing capabilities, processing the data at the edge is challenging. For example, power consumption becomes problematic if data is processed on the IoT device itself. Thus, there is a need to feed this massive data into the cloud platform for analysis. However, uploading the data from IoT devices to the cloud platform causes a delay which is a significant issue for delay-sensitive applications. This tradeoff between delay and power needs a favorable policy to decide where it should allocate the task from edge to cloud processing platform. Research on this subject addresses this issue quite frequently, and various methods have been proposed to mitigate the problem. The previous studies usually focus on the edge-to-cloud computing platform, i.e., they efficiently offload the computational tasks onto the IoT devices and cloud. This thesis proposes a balanced task allocation between edge and cloud computing regarding power consumption and delay. We accomplish our idea by comparing the different task allocation methods, benchmarking in different scenarios, and evaluating by proposing mathematical modeling.