A Survey on Computation Offloading and Current Trends

Computation offloading has proven to be an important technique in modern computing to improve performance by saving energy consumption and optimizing resource utilization by transferring computational tasks from devices constrained by resources, such as mobile devices, Internet of Things (IoT), nodes, and edge systems, to more powerful remote servers, cloud platforms, or edge computing nodes. This survey offers a critical view of computation offloading from architectural models to decision-making strategies in real-world applications. Offloading strategies may differ between full and partial offloading, highlighting significant key challenges such as network latency, task scheduling, security risk, and resource allocation. Additionally, this section provides an overview of various machine learning techniques, including supervised machine learning, unsupervised machine learning, deep learning, and reinforcement learning, which are utilized for real-time adaptive and secure task distribution. Furthermore, advancements in 5G networks and edge intelligence have significantly reduced computational delay, making offloading more efficient for applications that demand low latency, such as cloud gaming, autonomous vehicles, and unmanned aerial vehicles used for surveillance purposes. The survey enhances computation offloading strategies for future distributed computing environments by reviewing state-of-the-art techniques and methodologies. A part of this analysis is focused on the simulation model analyses used in various computation offloading schemes. Ultimately, we believe that there is still significant scope for enhancing computation offloading, optimizing energy consumption, and reducing latency to uphold the overall quality required. This survey paper explores the potential for future research directions in this area.