AcoFuzz: Adaptive Energy Allocation for Greybox Fuzzing

Abstract

In recent years, coverage-based greybox fuzzing (CGF) has become one of the most important techniques to discover security bugs. The existing fuzzers score the seeds, and then allocate the energy to the seeds according to the scoring results, but most seeds obtain the same energy, and then repeatedly select the same seeds for fuzzing. These strategies have been proved to be inefficient. Our experimental observations show that various seeds have diverse efficiency, and the efficiency of test cases changes increase with execution time. In this paper, we propose a novel yet lightweight energy allocation strategy, called AcoFuzz, to improve fuzzing efficiency. AcoFuzz has one following distinct advantage: Dynamically allocate energy for seeds to cope with their efficiency variation. Extensive experiments based on real-world programs and the LAVA-M dataset have been conducted to evaluate the path discovery and vulnerability detection ability of AcoFuzz, which substantially outperforms 3 state-of-the-art fuzzers.