Design of an incremental learning model for shard management in performance-aware blockchains: GA-TLEHO approach

Abstract

Performance-Aware Blockchains (PABs) are the highly optimized version of regular blockchains, capable of mining blocks with low delay and low energy. To incorporate performance awareness, researchers have proposed various lightweight consensus models that depend on single-chained blockchains, which limits their scalability performance after several blocks. To overcome this limitation, sharded blockchains were introduced, but very few of these sharded chains use lightweight consensus models. Hence, in this work, we propose a novel incremental learning model with light-weighted consensus for shard management in performance-aware blockchains. By using incremental learning, the model could continuously learn from the changing characteristics of the blockchain and make informed decisions about shard size adjustments to optimize performance. Further, to optimize shard size and ensure mining operates with lower delay and lower energy consumption, we use a hybrid combination of Genetic Algorithm and the Learner-based Elephant Herd Optimizer (GA-TLEHO) method. The reason for this hybrid combination of optimization methods is to enhance convergence by leveraging both exploration and exploitation. It is evident from the simulation analysis that the proposed method potentially mitigates Masquerading, Distributed Denial-of-Service (DDoS), and Finney attacks with higher efficiency and lower delay when compared with other state-of-the-art blockchain methods.