A PBFT Consensus Algorithm for Reputation Based on Online Social Behavior

In the blockchain-based social network platforms, consensus algorithms still cannot take into account both security and performance. In addressing the aforementioned challenges, we introduce SNPBFT, a pragmatic Byzantine fault-tolerant algorithm tailored for social networks. This algorithm leverages a reputation model to optimize both the selection of master nodes and the consensus mechanism among nodes. It can reduce request latencies and improve system throughput while enhancing the algorithm’s resilience against Byzantine attacks. Specifically, we designed a reputation model based on user social behavior for the first time. For the four social behaviors of browsing, liking, commenting and sharing, we use a linear weighting method to calculate the reputation value of nodes, dynamically updating and determining the final reputation value in a distributed manner. This approach enhances the reliability of the reputation value and reduces the probability of malicious nodes dominating the consensus process. Meanwhile, considering the large volume of social media communication, we simplify the PBFT consensus process and remove the submission stage to improve the efficiency of the consensus communication. The experimental results demonstrate that, in comparison to the other five benchmark algorithms, the SNPBFT algorithm reduces the average latency by 17.44% and increases the average throughput by 23.76%. Meanwhile, the SNPBFT algorithm has good scalability and stability for different sizes of node numbers and request loads. Furthermore, the SNPBFT algorithm maintains a high consensus success rate, exhibits commendable latency and throughput performance under malicious node attacks and reputation attacks, thereby fulfilling the consensus requirements of blockchains in social networks.