Fog Computing enabled Smart Grid Blockchain Architecture and Performance Optimization with DRL Approach

Abstract

Smart grid is willing to make full advantage of distributed clean energy to alleviate energy crisis and environmental problems. However, distributed renewable energy is usually invisible and uncontrollable for the current power system, and there are intermittent problems in its generation. Therefore, how to achieve the power balance, maintain safe operation, and ensure the reliability and quality of power supply when the distributed energy reaches a high penetration in the grid is a huge challenge. Blockchain as one of the research hotspots brings about new solution approach to the dilemma. The two fields have many commons on decentralization, autonomy, marketization and intelligence. In this paper, we discuss the feasible scheme of the integrated system and add fog computing to reduce costs. Considering the realization of system, we choose Hyper Fabric as the basic structure and add verifiable random function to the consensus aimed to improve randomness and security in the encrypted election. Meanwhile, in order to satisfy the business requirements, a flexible adjustment method based on Deep Q Learning algorithm is designed to realize the joint optimization of throughput, latency and storage cost. The proposed scheme provides the advantages including privacy, flexibility, extensibility and implantation simplicity.