Research And Implementation Of Consortium-chain Block Generating And Consensus Mechanism

Blockchain technology has developed rapidly in recent years.The early blockchain applications are all public-chain systems,which carry a large number of nodes that can freely enter and exit.Therefore,these systems generally use Proof of work and other consensus algorithms.Blockchain systems that use proof-of-consensus algorithms generally can only process a low number of transactions per second and its transaction confirmation delays are very high,making it difficult to adapt to financial blockchain applications that require high concurrency and low latency.On the other hand,in the public-chain system like Bitcoin,in order to motivate the consensus nodes to keep mining,the system will give the consensus nodes economic rewards for mining new blocks,which is conducive to the continuous extension of the blockchain.Economic rewards directly motivate consensus nodes to mine and extend blockchain.The transaction fee of the transaction to be selected depends on selection algorithm.The priority-based transaction selection algorithm adopted by Bitcoin is relatively simple,but generally can't get the maximum transaction fee.In this thesis,the process of selecting transactions by consensus nodes in the block generating process is modeled as a 0/1 knapsack problem,which is solved by branch-bound algorithm and genetic algorithm respectively.The simulation results show that the transaction selection algorithm based on the branch-bound method has a higher sum of transaction fees than the priority-based algorithm,which can increase the economic benefits of consensus nodes.As a heuristic algorithm,genetic algorithm is suitable for the selection process with huge transaction volume,and the parameters of genetic algorithm will affect the result of sum of transaction fee.Simulation illustrates the impact of genetic algorithm's crossover rate,mutation rate and population size on indicators such as the sum of transaction fees.This thesis then illustrates the process and shortcomings of common consensus algorithms,and the detailed process of PBFT(Practical Byzantine Fault Tolerance)algorithm adapted to the consortium-chain.The selection process of the master node of the PBFT algorithm is relatively random,which easily triggers the view-change protocol and causes the rise of consensus time.This thesis models the selection of the main node as a fuzzy comprehensive evaluation process to ensure the objectivity of the selection,and uses the coefficient of variation method and the CRITIC(Criteria Importance Through Intercriteria Correlation)method to distribute the weight of the fuzzy comprehensive evaluation which both reflects the volatility and confliction between different indicators.In that case,we can avoids the subjectivity of weight distribution that relies too much on human experience in the past.The simulation results show that the improved algorithm has a significant increase in consensus efficiency,which can effectively reduce transaction confirmation delays.The last test shows that the improved system still has higher block generating efficiency and higher system throughput when the number of Byzantine nodes reaches the theoretical upper limit than other blockchain systems such like Bitcoin,Ethereum,and Hyperledger,which can meet the needs of most consortium-chain applications that have higher throughput requirements.