Implementing Encryption as an Architectural Approach to Securing Memory Systems

Ensuring secure memory systems is critical in mitigating hardware-level threats in modern computing. This work investigates the integration of encryption into memory architectures as a defensive mechanism against physical attacks such as RowHammer and Cold Boot. A custom encryption engine is implemented within Ramulator 2.0, an open-source DRAM simulator, to enable evaluation of performance-security trade-offs. The impact of encryption on memory latency is quantified, and system-level overheads are analyzed. Additionally, optimization strategies involving memory scheduling are explored to reduce timing penalties associated with encryption. The results demonstrate that encryption can enhance memory protection while introducing architectural challenges related to latency and integration.