DESC0024986

C58-01a-281801 We propose to focus on the problem of providing resilient operations to Electric Vehicle Charging Systems (EVCSs) in the face of persistent malicious software threats. Our approach leverages state-of-the- art System-on-Chip (SoC) architectures to monitor, repair, and dynamically alter EVCS binaries from the tightly coupled Field Programmable Logic Array (FPGA) fabric inside the chip-boundary of an SoC. Building from our experience on the DARPA HARDEN and HASH programs, we will develop a heterogenous embedded ecosystem that allows vendorsĺ EVCS software to coexist with innovative security-oriented hardware inference engines. These engines will validate the temporal behavior of EVCS subsystems, detecting potentially malicious implants and intrusions, then restoring the system to a ôgold- standardö image and safe operational state, to mitigate potential attacks. The gold-standard will be constructed such that it does not share an exploitable address with the existing binary, thereby preventing re-infection. The inference engines, and the system behaviors they enforce, are expressed as high-level specifications and behavioral models that are subsequently automatically transformed into FPGA circuit blocks leveraging our enhanced High-Level Synthesis (HLS) technology. The approach is unique in that it leverages the FPGAĺs inherent standing as a bus-master to inspect and modify binaries from a hardware base-of-trust that is invisible to and immutable by malicious implants. This approach has only recently become viable due to maturity of the associated tool chains and the general acceptance of SoCĺs as the core building block for systems integration. The metrics we will use to assess the approach will be based on its ability to operate on any EVCS binary, the time to detect a potential malicious implant, and the time to repair and restructure a designated system binary. Our Phase I effort will focus on the design of a Minimal Viable Product (MVP) for a Resilient Ecosystem that could be utilized in any EVCS. Since our ecosystem rests upon its ability to operate on any arbitrary system binary, irrespective of the vendor, it is the feasibility of this aspect of the system that will be the central focus of the Phase I effort. We plan to explore operating on multiple EVCS software stacks, such as EVerest, in an initial prototype and develop exemplars of models to reflect EVCS behavior. The models will emanate from EVCS software call-graph execution and network interfaces. Web Sensing has, for more than a decade, been a pioneering advocate of resilient systems leveraging novel SoC technologies; our approach has seen continuous evolution at the forefront of several DARPA cyber security efforts, including the CRASH, MRC, AMP, HARDEN, and HASH programs. With experts in automated tools for building FPGA and SoC systems designs, we are uniquely qualified to conduct this application of our existing tools, techniques, and procedures to the domain of EVCS resilience.