HSelect: Effective Selection of Heterogeneous Software for Common Mode Failure Mitigation
Zhiyuan Yu, Xianglong Kong, Xinyuan Miao, Yi Cao, Wei Bo Huang, Jiayu Du · 2025
With the growing risk of Common Mode Failure (CMF), N -versioning has become a key strategy for improving fault tolerance and security in software systems. While studies often assume the availability of off-the-shelf heterogeneous variants, practical N -versioning systems are typically built using different versions of the same software. Selecting heterogeneous variants is a key challenge in building robust N -versioning systems. This work introduces HSelect, a novel approach for selecting N versioning variants by measuring software heterogeneity. HSelect detects code changes at file and block levels, generating a heterogeneity matrix to select maximally diverse variants. Evaluation on eight projects shows HSelect more effectively measures heterogeneity and mitigates CMF vulnerabilities than traditional methods.