SP-PoR: Improve blockchain performance by semi-parallel processing transactions

Abstract

As is well known, performance is always an obstacle to blockchain’s further development and application. However, limited by information propagation latency and the existence of conflicted transactions, neither shortening block generation interval, expanding block capacity, nor designing noval parallel processing supported blockchain structures cannot actually improve the performance. To address the performance improvement dilemma, in this paper, starting with the transaction relation, we propose a semi-parallel transaction processing protocol called the “Semi-Parallel Proof-of-Relevance” (SP-PoR). By analyzing the relevance determination condition of different types of transactions, we found that splitting and processing normal transactions and contract transactions respectively are key factors in boosting the entire blockchain performance. Then, theoretically, the parallel processing of unrelated blocks has been proven to have no influence on final consensus achievement. Based on this, a semi-parallel processing method is proposed to parallel and serially process normal and contract transactions based on relevance determination. Furthermore, multi-way tree DAG structure and tangle structure are designed to support our research. Finally, extensive experiments based on real Ethereum transaction data prove that our proposed method can indeed improve blockchain performance in which normal transactions and contract transactions of our method have 8.575 times and 2.54 times improvement respectively. The entire blockchain can experience 2.117 times performance improvement on blockchain synchronization considering practical execution environment deductions.