Multi-objective Optimization of Block Size Based on CPU Power and Network Bandwidth for Blockchain Applications

Abstract

Distributed ledger technology (DLT) is fast growing as a solution that could address pertinacious challenges in the financial sector offering a more efficient and resilient approach to transaction record keeping. Its popularity is principally determined by its potential to transmit data in bulk securely over a peer-to-peer network. Block size optimization is an important issue in blockchain network as scalability bottlenecks prevent higher throughput and minimized latencies. Increasing adoption has raised concerns about its ability to scale and serve as a real-world usable network. An increased block size may cause a higher transmission time as the rate of transactions made will increase and may also cause the system to reach its maximum capacity to clear transactions. As compared to larger block size, small block size is more efficient, but creating too small a block might cause higher block building/creation time. An efficient blockchain-based application requires an optimal block size so that a justified performance is achieved. In the proposed approach, multi-objective optimization is performed, and 40 different solutions are obtained based on transaction selection time and block building time. The selection of a particular block size as an optimal solution is based purely on processing power of CPU, throughput and the available network bandwidth and latencies, where the application is to be deployed. The results establish that for various systems having 1.1 GHz CPU–3.0 GHz CPU, a 3.8 MB of block size selection optimizes the transaction selection and block building time.