ARMM: Sandwich-Attack Resilient Automated Market Maker

A popular decentralized exchange (DEX) model is the automated market maker (AMM), supporting billions of USD of daily trading volume. Notably, a sandwich attack is a predatory trading strategy upon DEX, where adversaries extract on a monthly basis over 10M USD from the benign traders. Previous solutions provided sandwich-attack resilience primarily through complex trader operations or costly order matching, which are user-unfriendly and inefficient on blockchain. In this work, we propose a new AMM design-the Autonomous Reordered Market Maker (ARMM) with integrated sandwich-attack resilience. Specifically, we provide a prediction resistant on-chain reordering algorithm to randomly shuffle and reassemble transactions, reducing the probability of a successful sandwich attack. Our empirical evaluation suggests that ARMM can prevent a profitable single-block sandwich attack with 90.16%, and increase the sandwich attack victim's revenue with a probability of 96%. In addition, we simulate the profit of adversaries in cross-block sandwich attacks and find that ARMM can even keep the adversary from profiting and increase the victim's revenue with a probability of 49.60% and 94.60%, respectively. Besides, our analysis shows that ARMM can avoid 100% of the perfect single-block attacks and has a lower limit probability of 33.50% in avoiding imperfect single-block attacks from adversary miners.