Partial Offloading Schemes for Latency and Computation Sensitive Tasks

Computational Offloading is a revolutionary technology for enhancing mobile devices' performance. Computation offloading solves resource limitations of mobile devices by providing lower energy consumption, lower latency, and improved application performance. There are two types of offloading to transfer computational workload in the cloud: 1. Full offloading and 2. Partial offloading. This paper provides a complete overview of past developments and recent achievements of partial offloading modeling in edge computing and cloud computing. In particular, we differentiate how our survey work is unique from other offloading related surveys. We mainly concentrate on the most prominent models (Markov-based, graph-based, and deep learning-based) frequently used for partial offloading, and we compare algorithms that follow these models. Finally, we summarize each methodology with a concise explanation and highlight some crucial open research problems from our findings.