Impact of Clustering technique in enhancing the Blockchain network performance

Abstract

Blockchain technology facilitates transparency, de-centralization, immutability, and security for each transaction. The decentralization characteristics enable the block chain network to have different nodes as the Certificate Authority (CA) for a different transactions. Choosing a different validator for every individual transaction may increase the security perspective of the transaction, but it can lead to one major concern such as the network overhead increase. There can be several numbers of transactions for a single network and selecting a validator every time can cause a block propagation delay which also decreases the network efficiency. To avoid such kind of issue clustering of network nodes can be an emerging solution. In the current research work, the entire network participating nodes are considered for making different clusters based on the response time factor. Every cluster has the average response time as the threshold value. The node that wants to initiate a transaction needs to raise a minimum response time requirement and the cluster selection procedure will be executed based on the received response time value. In the current research work, the K-Means clustering with the Elbow method as an internal validation method is considered to decide the number of clusters. The computational time for both cases is compared to measure the effectiveness of the clustering process on the blockchain network.