Markov Chain-based Data Privacy Prioritizing for Task Scheduling in Digital Twin Edge Networks

Edge computing plays a significant role in shaping the future of Internet of Things (IoT) networks by processing high volumes of data at the network edge and meeting the desired latency demand. Although edge computing is regarded as a promising technology for these features as well as its decentralization and supportive functionalities, there is a need to develop systems to protect data security and privacy since users can face the risk of data leakage when they offload data to the edge/cloud server. In order to solve this problem, in this paper, we propose a three-layer digital twin edge network architecture for task scheduling based on edge-cloud computing. The digital twin has been used to map the IoT network to the virtual space, monitor physical assets in real-time, and perform comprehensive network analysis. On this basis, a task scheduling model that prioritizes data privacy has been introduced to prevent users’ privacy-sensitive information from being collected by attackers. The proposed model can both take full advantage of edge and cloud computing and protect data privacy. We classify computation tasks based on data privacy sensitivity and model them using the M/M/1 Markov model in local, edge server, and cloud computations. Then, we propose a data privacy prioritizing task scheduling algorithm combining digital twin and edge-cloud collaboration. This genetic optimization based algorithm significantly increases data privacy by 40% compared to the greedy approach, while keeping the time under an acceptable level. In this way, each task is distributed between the local, edge server, and cloud server. Thanks to the intelligence behavior and repetitive growth of the proposed algorithm, the optimal task scheduling is performed in a digital twin system. Our analysis reveals that data leakage attacks can be significantly decreased with the proposed task scheduling algorithm by minimizing task processing time.