R.: Tamper-tolerant software: Modeling and implementation

Abstract. Common software-protection systems attempt to detect ma-licious observation and modification of protected applications. Upon tam-per detection, anti-hacking code may produce a crash or gradual failure, rendering the application unusable or troublesome. Such a response is designed to complicate attacks, but has also caused problems for devel-opers and end users, particularly when bugs or other problems invoke anti-tampering measures accidentally. To address these issues, an al-ternative approach is to detect and fix malicious changes. This paper presents a scheme to transform programs into tamper-tolerant versions that use self-correcting operation as a response against attacks. Combin-ing techniques from the fields of fault tolerance and software security, the approach transforms programs via code individualization and redun-dancy. We also describe security enhancements through error correction, delayed responses and checkpointing. For security analysis, we adapt a graph-based model of attacks and defenses in the context of software tamper-resistance. This helps to estimate the difficulty of breaking our scheme in practical scenarios.