Adaptive Multi-paradigm Metaheuristic Optimization for Blockchain Scalability: A Comparative Evaluation of Transaction Latency and Throughput

Abstract Blockchain technology faces significant scalability challenges, primarily high transaction latency and limited throughput, which limiting its widespread adoption. Existing and recent optimization techniques, such as the Snake Optimization Algorithm (SOA), often lack adaptability under increasing workloads. This paper proposes the Multi-paradigm Adaptive Metaheuristic Optimization (MAMO), that integrates adaptive metaheuristic search, reinforcement-learning guidance, sparse optimization, dynamic feedback control, and game-theoretic. designed to minimize transaction latency while maximizing throughput in scalable blockchain system. An experimental comparative research design was employed using a Python-based simulation and 30 independent replications were specified for each algorithm-workload condition, with identical decision-space bounds, population size, iteration budget, and simulator conditions. across six workloads (1 to 100,000 transactions), MAMO was evaluated against SOA. Results show that MAMO consistently achieves lower mean latency (0.221% to 10.378% improvement) and higher throughput (45.041% to 94.184% improvement) across all workloads. At 100,000 transactions, MAMO achieved 0.939850 ms latency and 327,509.84 TPS, compared to 1.048680 ms and 202,864.77 TPS for SOA. Statistical analysis using the paired Wilcoxon signed-rank test with Holm correction confirms that throughput improvements are significant across all workloads (p < 0.001), while latency improvements are significant at workload 1 (p = 4.696 × 10⁻⁵). These findings provide strong evidence that MAMO offers a more effective and scalable optimization approach for blockchain systems compared to SOA.