A preview of the PDF is not available
Optimized Energy Trading between Electric Vehicles in Smart Grids using Blockchain (MS Thesis without Source Code)
Abstract and Figures
In this thesis, we proposed three system models to resolve the different problems related to energy optimization and anomaly detection. The exponential increase of EVs in smart cities has led to some complex trading problems. Therefore, in proposed system model, we deal with some major energy trading issues related to the charging of Electric Vehicles (EVs) in a Vehicular Energy Network (VEN). The enormous increase in the development of EVs and Intelligent Vehicles (IVs) has led to a complex network. When the number of vehicles increases, the number of links is also increased and generates intensive data. This complexity leads to insecure communication, road congestion, security and privacy issues in vehicular networks. Moreover, the detection of malicious IVs, data integration and data validation are also major concerns in VENs that hinder the network performance. To deal with these problems, we propose a blockchain based model for secure communication and detection of malicious IVs. In general, there are two major issues related to EVs charging. First, it is difficult to �find the nearest charging station with less energy consumption. Second, it is difficult to calculate the exact amount of energy needed to reach the nearest charging station from the current location of an EV. In traditional energy trading systems, centralized grids are being used where energy trading between EVs and charging stations is not secured. To deal with these issues, we proposed a consortium blockchain-based secure energy trading system with moderate charging cost. In the proposed work, the distance of nearest charging station is calculated using k-Nearest Neighbor (KNN) technique. Furthermore, EVs also face various energy challenges such as imbalance load supply, fluctuations in voltage level and energy scarcity from charging station. Therefore, a Demand Response (DR) strategy is used, which enables the EV users to flatten out the load curves and adjust the usage of electricity e�ciently. Secondly, we addressed the problem of insecure communication in Vehicular Network (VN). The integration of blockchain with vehicular network makes the network secure and trustful. For the authentication of IVs, Certi�ficate Authority (CA) is used, and InterPlanetary File System (IPFS) is integrated with CA to resolve the issue of storage. A reputation mechanism is introduced to detect the malicious IVs in the network based on their ratings. A branching concept is involved where the validated transactional data and the malicious data of IVs is stored in two separate chains: Integrity Chain (I-Chain) and Fraud Chain (F-Chain). In third contribution of our thesis, we deal with some major problems related to the �financial sector such as fraud and anomalies. These are common problems in E-banking and online transactions. Anomaly detection is a well known method to �find frauds and misbehavior in the �financial sector. However, with the advancement of �financial sectors, the methods of fraud and anomalies are also getting advanced. Furthermore, blockchain technology is introduced as the most secure technology and integrated with �finance. However, besides these advanced technologies, there still exist many fraudulent cases every year. Therefore, we proposed a fraud detection prediction model based on machine learning and blockchain. Machine learning techniques train the dataset according to the fraudulent and integrated transaction patterns and predict the new incoming transactions. Blockchain technology is integrated with machine learning algorithms to detect fraudulent transactions in �financial sectors. There are two machine learning algorithms: XGboost and Random Forest (RF) are used for the classi�fication of transactions. These models provide 99% accurate results to �find the frauds and anomalies in the given dataset. The proposed models are also able to predict the transaction patterns. We also calculate the precision and AUC of models to measure the accuracy. Simulation results show that our proposed integrated models outperforms the traditional techniques in terms of accuracy and security. A security analysis of proposed smart contracts is also shown in simulation section. The security analysis is used to identify the vulnerabilities of smart contracts. Furthermore, four attacker models are introduced to resolve the issue of sel�fish mining attack, double spending attack, replay attack, and Sybil attack.
Content may be subject to copyright.