AQRB: A Crypto-Agile Quantum-Resilient Authentication Layer for Permissioned IoRT Blockchains
The Internet of Robotic Things (IoRT) increasingly operates as a multi-agent edge network where robots, sensors, and gateways cooperate autonomously under tight latency and resource constraints. Blockchain can strengthen trust and access control in such deployments, but most edge implementations rely on classical ECC-based signatures, which are vulnerable to quantum-capable adversaries. This paper proposes Adaptive Quantum-Resistant Blockchain (AQRB), a crypto-agile framework that switches among classical, hybrid, and post-quantum (PQ) transaction-authentication modes for multi-agent edge IoRT. AQRB applies ML-DSA-44 (FIPS 204) at the device-tovalidator boundary; the permissioned Ethereum consensus substrate retains classical primitives — a deliberate scope boundary addressed explicitly. The framework pairs an edge permissioned blockchain with oneM2M off-chain storage, selects cryptographic modes via normalised resource and security scores, and coordinates mode changes through a lightweight on-chain governance protocol. We prototype AQRB on a three-node ZedBoard (ARM Cortex-A9, 666 MHz) testbed. ML-DSA-44 achieves sign 1.80 ms and verify 0.75ms — under 0.3% of the 780 ms block-finality budget, confirming the primitive is not the system bottleneck. Under nominal conditions, the adaptive baseline matches classical performance exactly (794 ms E2E, 45 TPS) while fixed-PQC wastes 53% more CPU; under high-criticality conditions, the emergency override commits quantum-resistant M3 on-chain within 830 ms. A simulation-based ablation shows the hysteresis mechanism cuts unnecessary mode-switch churn by 86% without delaying emergency escalation to M3. These results demonstrate that transaction-layer PQC migration is feasible on constrained IoRT edge hardware without sacrificing real-time performance; AQRB is deliberately scoped to transaction authentication and identity records, not full consensus-layer quantum resistance, which is left as future work.