Improving cost-effectiveness of mutation-based fault localization using higher-order mutants

Abstract

Fault localization is a technique to deduce the exact source of a failure from a set of failure indications. It plays a crucial role in improving software quality, especially software debugging. Mutation-Based Fault Localization (MBFL) techniques are proposed to locate faults at a finer granularity and with higher accuracy than traditional spectrum-based fault localization techniques. Despite its effectiveness, the immense cost required in mutation analysis precludes MBFL from being practically utilized. Various mutation alternative strategies have been suggested to reduce the cost of MBFL but at the expense of localization accuracy. This study proposes a novel and cost-effective MBFL technique called HOTFUZ, Higher-Order muTation-based FaUlt localiZation, that employs Higher-Order Mutants (HOMs) to reduce the cost of MBFL while minimizing the accuracy degradation. HOTFUZ combines the mutants of a program under test into HOMs to decrease the number of mutants to be executed. This is one of the earliest research to apply higher-order mutations to MBFL and the first to use HOMs to improve the cost-effectiveness of MBFL. An extensive experimental study is conducted using 460 real-world faults of CoREBench and Defects4J to evaluate the cost-effectiveness of HOTFUZ. The experimental results demonstrate that the proposed approach outperforms extant mutation alternative strategies in fault localization accuracy while executing the same number of mutants. HOTFUZ has three major benefits over existing cost reduction techniques for MBFL: (1) it retains the advantage of using the entire set of mutation operators

(2) it does not discard generated mutants randomly for the sake of efficiency

and (3) it significantly decreases the proportion of equivalent mutants. In conclusion, the proposed approach, HOTFUZ, can effectively localize faults at a low cost, thus practically supporting the software debugging process.