EnergyTradingRank Algorithm for Truthful Auctions among EVs via Blockchain Analytics of Large Scale Transaction Graphs

In a decentralized vehicle-to-vehicle (V2V) energy trading among electric vehicles (EVs), there is a need of a highly secure and privacy protected transaction environment. With the advent of digital currencies like Bitcoin, most of the blockchain applications incorporate a similar cryptocurrency system for secure processing. In this work, we propose a permissioned blockchain based energy trading among EVs to enhance security and privacy of the EV user's information. We introduce, a new cryptocurrency "ETcoin" for energy trading among EVs. All the matched bids comprises of traded energy unit and ETcoins transferred are stored in blockchain. The transaction graph formed by energy trade can be analyzed as a large scale graph processing using Big Data analytics. Existing works have not addressed blockchain analytics for processing large transaction graphs of energy trading. All the blockchain transactions submitted in a pre-defined period are considered for analytics. We propose an EnergyTradeRank (ETR) algorithm for Blockchain analytics of the large scale transaction graphs. ETR algorithm inherently applies the weighted ranking approach, which considers both the edge as well as vertex properites of graph to calculate the ETR Score for each EV during a traded period. Transaction graph analytics on energy trading acts as a proof-of-activity and it also help in incentivizing the EVs with maximum satisfiable participation in the blockchain environment. We implemented the blockchain based proof-of-concept (POC) using IBMs hyperledger fabric and composer with minimum transaction latency and in-time modeling of each participant to the system. Transaction graph analytics is implemented via an open source Apache Spark's GraphX library for a realtime distributed dataflow processing. Transactions are modelled as edge triplets on several machines and ETR score for every vertex is computed parallelly to converge at a set tolerance value. Simulated results show that the algorithm converges faster and scales better as the number of parallel machines increases. We also analyzed that the overall incentive gain for EVs with maximum, truthful and committed participation increases and thus promotes active participation in the system.