Providing secure boot of obfuscated opcodes for Cybershield using a trusted execution environment

Abstract

As Field Programmable Gate Arrays (FPGAs) become increasingly integrated into security-critical applications--particularly in space, IoT, and edge computing environments--the need for robust security mechanisms has grown paramount. Despite their versatility and performance, FPGAs were not initially designed with strong security in mind, leaving them vulnerable to attacks such as code injection and buffer overflows. This thesis introduces CyberShield, a Trusted Execution Environment (TEE) architecture integrated into the radiation-tolerant RadPC softcore processor to defend against such threats. CyberShield enables secure boot with obfuscated opcodes, leveraging RadPC's Quad Modular Redundancy (QMR) and opcode diversification to prevent unauthorized code execution. By assigning unique opcode offsets to each core and validating integrity through an anti-voter module, CyberShield can detect code and command injection attacks. This thesis details the architectural modifications necessary to achieve secure boot from non- volatile memory, the encryption and bootloading process, and how the system mitigates injection-based attacks. Experimental validation confirms CyberShield's ability to resist buffer overflow exploits while preserving system reliability and performance. This work lays the foundation for further development of other TEE related security features for RadPC and embedded systems deployed in adversarial environments.