Node Criticality Assessment in a Blockchain Network

Blockchain systems are being rapidly integrated in various technologies, with limited work on the effect of the underlying network topology on the blockchain performance. In this work, we investigate the significance of each network node on the overall blockchain performance. This is assessed by selecting critical nodes according to different criticality metrics, and investigating, using simulations, the degradation in performance incurred upon removing these nodes. The most critical nodes are the ones that incur the greatest degradation in performance. The considered performance metrics are the blockchain size and the packet drop rate. Criticality metrics such as Betweennes Centrality, Closeness Centrality and Degree Centrality are compared. It is found that the Sign Change Spectral Partitioning approach, enhanced with Blockchain Specific traffic flow information, is able to identify critical nodes better in the sense that higher degradation in performance is reported upon their removal.