Cloud-Edge Computational Offloading Techniques and Verification Methods: A Survey

Computation offloading could improve performance, energy efficiency, and resource utilization by transferring computation tasks from a local device to a remote server or cloud. With the rise of the Internet of Things (IoT) and latency-sensitive applications (AR/VR, autonomous driving, etc.), cloud-edge offloading has become a crucial paradigm. This paper presents a survey of recent cloud-edge offloading research. We classify state-of-the-art offloading techniques from heuristic algorithms and fuzzy logic to game-theoretic methods and deep reinforcement learning, and link them to application domains. We analyze the quality objectives these techniques optimize (e.g., latency, energy, throughput, fairness) and detail the evaluation methodologies used. We specify current technological challenges, outline best practices for verification, and propose standardized evaluation frameworks. The paper concludes with open challenges (such as improving reproducibility, standardizing benchmarks, and enhancing industry adoption) and highlights promising research directions to advance a more robust and practical cloud–edge offloading ecosystem.