PrivateEdge: Gizliliği koruyan otomatik kod taşıma sistemi

By providing an enriched computation environment, Edge Computing enables many infeasible applications requiring low latency to be deployed at computationally constrained devices but connected to a network. However, as the computation moves away from devices towards the edge, data belonging to the user also moves outside the device, raising privacy concerns. Besides privacy concerns, it is a fact that moving computation from devices to edge requires support from applications, which inherently incurs development costs. Considering that these two challenges set a barrier to adopting the Edge Computing paradigm, this thesis proposes a framework named PrivateEdge that automatically analyzes and transforms applications to offload their subset of tasks to the edge without compromising user privacy by bringing advances in other concepts, technologies, and tools together. Internally, PrivateEdge uses LLVM to analyze and transform a given application at compile time. Additionally, it introduces the Confidential Computing paradigm for hosting a Trusted Execution Environment (TEE) on edge to secure and ensure the integrity of user data and computation. As a last feature, PrivateEdge employs WebAssembly technology on edge to provide a unified execution environment across different edge devices and secure them by sandboxing the execution of offloaded tasks. When the implementation of the framework is tested with a benchmark suite, a 15% reduction in overall execution time and also a considerable decrease in CPU utilization are observed. PrivateEdge can help many applications adapt to the Edge Computing paradigm without spending additional effort and compromising their users' privacy.