A Hybrid PSO–GA Algorithm for Adaptive Decentralized Sequencer Selection and DAG-Aware Workload Coordination in Layer-2 Blockchain Networks

Abstract

Decentralized Layer-2 sequencing requires fixed-size committees to balance performance, fairness, and resilience while downstream transaction tasks remain dependency-constrained. This study formulates committee selection as a bounded seven-objective discrete optimization problem and combines a feasibility-preserving hybrid particle swarm optimization–genetic algorithm (PSO–GA) search with directed acyclic graph scheduling and post-selection Byzantine fault tolerance (BFT) audits. Across 72 matched model-based runs, hybrid PSO–GA produced a 16.1% higher score, 21.9% lower latency, 47.0% higher throughput, 9.6% lower energy, and 6.2% lower effective gas than the controlled stake-only BFT baseline. It also exceeded random eligible selection, PSO-only, MOPSO, and NSGA-II; however, GA-only attained the highest scalar score (0.8299 versus 0.8266). The hybrid is therefore interpreted as a competitive feasibility-preserving trade-off, not a universally superior optimizer. Under high dependency and task heterogeneity, the scheduler reduced mean makespan to 32.75 from 34.38 for FIFO-ready and 49.59 for round-robin scheduling. Targeted attack simulations could exceed the BFT budget. The results support the framework as a reproducible decision model, while testbed deployment, authenticated telemetry, and incentive mechanisms remain necessary for protocol-level validation.