Research On Transaction Assignment And Multi-Replication Block Placement In Blockchain

In recent years,blockchain has attracted increasingly more attention.Sharding greatly improves the throughput and the storage scalability of blockchain.As another technology to solve the problem of blockchain storage scalability,the multi-replication block solution effectively reduces the storage burden of nodes.This thesis studies the problems of profit-oriented transaction assignment and query cost-oriented multireplication block placement in blockchain.The main contents of the thesis are as follows:(1)Profit-oriented transaction assignment in sharding blockchain.Most existing sharding solutions randomly assign transactions into shards,while ignoring the impact of transaction assignment on the profit of transaction fees in the blockchain system and the limited cross-shard transaction verification capability of each shard.This thesis proposes a Levy flight and Elite mechanism based Binary Evolutionary algorithm(LEBE)for the problem.LEBE uses a greedy algorithm to get the initial solution,and gets the initial population from this initial solution with neighborhood search.LEBE iteratively runs the evolutionary process based on the elite mechanism to generate trial solutions.In order to avoid falling into the local optimum,algorithm LEBE uses the Levy flight strategy in the process of getting the trial solutions to the global search ability.Furthermore,when the trial solution is infeasible,LEBE uses a greedy algorithm to repair it into a feasible solution and optimize it.Finally,we conduct experiments using real blockchain datasets.Experimental results show that algorithm LEBE effectively improves the profit of the blockchain system and the resource utilization of shards.(2)Query cost-oriented multi-replication block placement in blockchain.Each node in the blockchain network stores a complete ledger data,which brings a heavy storage burden to every node in the blockchain network.The multi-replication block mechanism can reduce the storage redundancy inblockchain.However,most of the current multireplication mechanisms focus on the scalability or the availability of blockchain storage scalability,ignoring the query cost of the multi-replication mechanism compared with the full-replication mechanism.This thesis studies the block placement problem in blockchain based on the multi-replication mechanism with the objective of minimizing the query cost.This thesis proposes an Elite opposition based learning and Cauchy mutation strategy based Binary Sparrow Search algorithm(ECBSS)for the block placement problem.First,ECBSS uses the strategy of elite opposition based learning to optimize and improve the randomly generated initial population to generate an opposition solution corresponding to the initial population.Second,ECBSS selects the better solution from the initial population and the opposition population as the improved population.Third,ECBSS enters the iterative process.In each iteration,ECBSS calculates the fitness value of all solutions,find the current best solution and the worst solution,updates the positions of the producer,follower and sparrows who perceive the danger in the population in turn,and converts the latest solution into binary 0-1 variable.At the end of the iteration,ECBSS repairs this 0-1 solution to ensure it to be a feasible solution,and update the solution if the obtained feasible solution is better than the current best solution.Finally,ECBSS outputs the optimal solution.The experimental results show that the ECBSS can effectively reduce the total block query cost.