Robust Anti-Tampering via Cooperative Agents, Oblivious Computation, and Self-Modifying Code

The proposed effort shall design, implement, and investigate the effectiveness of improved techniques for anti-tampering application software. The anti-tamper techniques shall be designed to make it very difficult and costly to reverse engineer the software to discover how it operates. It must be presumed that the attacker is well funded (e.g. by a foreign government) and therefore sophisticated—having expertise of number theory, and other areas of computer science. The scope of the effort includes the design and implementation of cooperating software agents and sentinels that provide strong anti-tampering to an application. Their primary means of operation shall be live modification of application code during to provide dynamic obfuscation of application functionality. The effort shall produce: 1. A prototype that demonstrates the design, implementation, and operation of this AT methodology 2. Analysis of the techniques for robustness and execution overhead 3. Performance benchmarks The methodology to be presented focuses on the use of software agents / sentinels for implementing software applications that dynamically self-obfuscate themselves while simultaneously performing their intended functionality. The techniques are scalable (both upward and downward) in complexity so as to provide strong obfuscation even within the realm of real-time embedded applications.