Reconfigurable compute is a double-edged sword: FPGAs let you deploy custom datapaths for low-latency inference, networking, and acceleration, but the bitstream itself can become an attack surface in multi-tenant cloud and shared embedded platforms. This talk explains a practical “bitstream vetting” idea where users could upload a compiled .bit file, then an offline/on-device classifier screens it for hardware-Trojan style payloads before configuration happens. https://sc.edu/
—
HDMI® Technology is the foundation for the worldwide ecosystem of HDMI-connected devices; integrated with displays, set-top boxes, laptops, audio video receivers and other product types. Because of this global usage, manufacturers, resellers, integrators and consumers must be assured that their HDMI® products work seamlessly together and deliver the best possible performance by sourcing products from licensed HDMI Adopters or authorized resellers. For HDMI Cables, consumers can look for the official HDMI® Cable Certification Labels on packaging. Innovation continues with the latest HDMI 2.2 Specification that supports higher 96Gbps bandwidth and next-gen HDMI Fixed Rate Link technology to provide optimal audio and video for a wide range of device applications. Higher resolutions and refresh rates are supported, including up to 12K@120 and 16K@60. Additionally, more high-quality options are supported, including uncompressed full chroma formats such as 8K@60/4:4:4 and 4K@240/4:4:4 at 10-bit and 12-bit color.
—
The core method is intentionally lightweight and static: treat the FPGA bitstream as raw binary, extract byte-frequency features, then compress the feature space with Truncated SVD (TSVD) so inference stays cheap. Training data is built from Trust-Hub benchmark designs (AES-128 and RS232 variants) synthesized into labeled benign/malicious/empty bitstreams, with class imbalance handled via SMOTE and model selection done with k-fold cross validation. The best-performing model is a Random Forest, which is a good fit here because it handles noisy, high-dimensional distributions without needing deep learning on the target node or a big GPU.
A key point is deployment realism: the pipeline is demonstrated on the Digilent PYNQ-Z1 (AMD/Xilinx Zynq-7000), using the PYNQ Python stack and a Jupyter workflow rather than custom HDL changes. On-device results show about 3.35 seconds average latency per classification (feature extraction dominates, prediction is ~15–17 ms), while a hold-out test reports ~97.14% true-positive rate with ~0.8% false-positive rate, which matters when “false alarm” means re-running a long build or blocking a tenant. The interview is filmed at Supercomputing SC25 in St. Louis, which is a fitting venue for the cloud-to-edge security angle that shows up here.
What’s interesting going forward is interpretability and developer ergonomics: the next step described is pairing the detector with an NLP-style explanation layer so a misclassification can be translated into human-readable “why” signals, instead of a bare label. Combined with newer academic boards like AMD University Program AUP-ZU3 (Zynq UltraScale+ XCZU3EG class platforms), this kind of binary-level screening hints at a deployable security control that doesn’t require netlists, source RTL, or reverse engineering, and can sit right at the boundary between CI/CD and the FPGA fabric roadmap.
I’m publishing about 90+ videos from Embedded World North America 2025, I upload about 4 videos per day at 5AM/11AM/5PM/11PM CET/EST. Join https://www.youtube.com/charbax/join for Early Access to all 90 videos (once they’re all queued in next few days) Check out all my Embedded World North America videos in my Embedded World playlist here: https://www.youtube.com/playlist?list=PL7xXqJFxvYvjgUpdNMBkGzEWU6YVxR8Ga
This video was filmed using the DJI Pocket 3 ($669 at https://amzn.to/4aMpKIC using the dual wireless DJI Mic 2 microphones with the DJI lapel microphone https://amzn.to/3XIj3l8 ), watch all my DJI Pocket 3 videos here https://www.youtube.com/playlist?list=PL7xXqJFxvYvhDlWIAxm_pR9dp7ArSkhKK
Click the “Super Thanks” button below the video to send a highlighted comment under the video! Brands I film are welcome to support my work in this way 😁
Check out my video with Daylight Computer about their revolutionary Sunlight Readable Transflective LCD Display for Healthy Learning: https://www.youtube.com/watch?v=U98RuxkFDYY
