Driving Trust at the Edge - MACsec 2.0: Securing Physical AI for SDVs and Robotics

We present a MACsec 2.0 approach that optimizes how Ethernet data is partitioned for the lightweight ASCON cipher, building on the ongoing IEEE P802.1AEef work. By streamlining per-frame processing, it reduces the number of cryptographic operations in confidentiality mode—and in selected integrity-only configurations—across MSDUs of any size, with the strongest benefits on small, sensor-scale payloads. Analytical and prototype results show up to 20 percent fewer crypto operations, yielding lower latency, reduced energy, and tighter timing budgets—key for real-time Physical AI in SDVs and robotics.