Diversity-Based Evolutionary Fuzzing for Smart Contract Security Analysis

The immutability of smart contracts makes flaws critical, reinforcing the need for automated security testing. This work proposes DivFuzz (Diversity Fuzzer), a diversity-driven evolutionary fuzzing model based on CONFUZZIUS. The strategy incorporates phenotypic distance in the selection of individuals to broaden the exploration of the execution space. The evaluation used a benchmark of smart contracts and focused on code coverage as the primary effectiveness metric. The results show higher code coverage over time, indicating greater potential to reach unexplored execution regions. The findings confirm the importance of diversity in evolutionary fuzzing for smart contracts.