Privacy Preserving Offloading

While much research has been conducted on privacy preservation in conventional cloud computing, these techniques may not be straightforwardly applied to offloading edge computing. The reason is that conventional cloud servers are typically employed by industry leaders like Alibaba and Amazon at a centralized level, having thorough and sophisticated security protocols with high integrity. Even, edge servers might be installed by various organizations within an open ecosystem, featuring comparatively less stringent security measures. This can render them less reliable than conventional cloud servers and more inclined to both cyber threats and physical breaches. The privacy preservation in edge computing during data offloading has gained significant attention recently. A variety of research initiatives have been undertaken to probe various privacy preservation techniques in offloading, including data transfer metrics, wireless transmission techniques, and offloading destination selection schemes. In this chapter, we address privacy and latency issues in the context of offloading computations to the edge-cloud computing environment. Our solution focuses on leveraging the power of inductive learning to train a feature extractor and a centralized neural network, all while preserving the integrity of sensitive data at the network’s edge. We leverage local differential policy approach to ensure that private data is retained locally and never sent to the cloud. Additionally, our approach factors in data transmission costs and the resources available on edge devices, with the extensive aim of optimizing privacy and efficiency within mobile edge intelligent systems.