<p>This paper presents a quantum-resilient autonomy stack for medical-drone delivery that elevates communications and cryptography to first-class, stateful variables within motion planning rather than downstream constraints. A three-layer Air–Ground–Communications architecture integrates beyond-6 G (FR3/THz) links with ultra-reliable low-latency communication (URLLC) fallback and couples BB84-style quantum key distribution (QKD) with principled post-quantum cryptography (PQC) switching to ensure cryptographic continuity under mobility and adverse weather. At the core, a hybrid AI–RKF45 controller fuses a lightweight neural policy with the integrator’s local error, co-adapting control aggressiveness and solver step size for stiffness-aware manoeuvres and rapid re-optimisation under disturbances. The planner directly ingests link SNR, URLLC queueing delay, QKD quantum bit error rate (QBER), secure key rate (SKR), key-buffer levels, meteorological risk, and battery state-of-health, shaping a multi-objective cost that jointly minimises energy and <i>control-loop end-to-end latency</i> (sensing–compute–communication round-trip), while enforcing geofencing and Beyond Visual Line of Sight (BVLOS) constraints. In large-scale regional simulations (200 km <InlineEquation ID="IEq1"> <EquationSource Format="TEX">$\times $</EquationSource> <EquationSource Format="MATHML"><math> <mo>×</mo> </math></EquationSource> </InlineEquation> 200 km; up to 200 UAVs), the framework achieves a mean control-loop latency of 2.34 s (distinct from door-to-door delivery time), an 18% reduction in energy per sortie, and a 98.2% mission success rate, outperforming static planners, deep reinforcement learning (PPO/DQN), and model predictive control under matched compute budgets. The QKD<InlineEquation ID="IEq2"> <EquationSource Format="TEX">$\leftrightarrow $</EquationSource> <EquationSource Format="MATHML"><math> <mo stretchy="false">↔</mo> </math></EquationSource> </InlineEquation>PQC state machine applies conservative thresholds (QBER <InlineEquation ID="IEq3"> <EquationSource Format="TEX">$\le $</EquationSource> <EquationSource Format="MATHML"><math> <mo>≤</mo> </math></EquationSource> </InlineEquation> 8%, SKR <InlineEquation ID="IEq4"> <EquationSource Format="TEX">$\ge $</EquationSource> <EquationSource Format="MATHML"><math> <mo>≥</mo> </math></EquationSource> </InlineEquation> 5 kbps with key-buffer hysteresis), yielding rare, millisecond-scale fallbacks that preserve latency guarantees. Complexity scales as <InlineEquation ID="IEq5"> <EquationSource Format="TEX">$\mathcal {O}\!\left( \frac{n}{\nu }T\right) $</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi mathvariant="script">O</mi> <mspace width="-0.166667em" /> <mfenced close=")" open="("> <mfrac> <mi>n</mi> <mi>ν</mi> </mfrac> <mi>T</mi> </mfenced> </mrow> </math></EquationSource> </InlineEquation>, maintaining 2–5 s update times for fleets exceeding 200 vehicles. Reproducible artefacts and a staged pathway from hardware-in-the-loop to BVLOS trials support near-term healthcare deployment.</p>

Quantum-Secured AI-Driven Drone Logistics for Real-Time Healthcare Delivery

期刊论文
出版模式：

开放获取
发表日期：

2026年3月16日

查看全文

Sana Hafeez,
Ghulam E Mustafa Abro,
Maran Marimuthu

摘要

This paper presents a quantum-resilient autonomy stack for medical-drone delivery that elevates communications and cryptography to first-class, stateful variables within motion planning rather than downstream constraints. A three-layer Air–Ground–Communications architecture integrates beyond-6 G (FR3/THz) links with ultra-reliable low-latency communication (URLLC) fallback and couples BB84-style quantum key distribution (QKD) with principled post-quantum cryptography (PQC) switching to ensure cryptographic continuity under mobility and adverse weather. At the core, a hybrid AI–RKF45 controller fuses a lightweight neural policy with the integrator’s local error, co-adapting control aggressiveness and solver step size for stiffness-aware manoeuvres and rapid re-optimisation under disturbances. The planner directly ingests link SNR, URLLC queueing delay, QKD quantum bit error rate (QBER), secure key rate (SKR), key-buffer levels, meteorological risk, and battery state-of-health, shaping a multi-objective cost that jointly minimises energy and control-loop end-to-end latency (sensing–compute–communication round-trip), while enforcing geofencing and Beyond Visual Line of Sight (BVLOS) constraints. In large-scale regional simulations (200 km $\times $ $\times$ 200 km; up to 200 UAVs), the framework achieves a mean control-loop latency of 2.34 s (distinct from door-to-door delivery time), an 18% reduction in energy per sortie, and a 98.2% mission success rate, outperforming static planners, deep reinforcement learning (PPO/DQN), and model predictive control under matched compute budgets. The QKD $\leftrightarrow $ $\leftrightarrow$ PQC state machine applies conservative thresholds (QBER $\le $ $\leq$ 8%, SKR $\ge $ $\geq$ 5 kbps with key-buffer hysteresis), yielding rare, millisecond-scale fallbacks that preserve latency guarantees. Complexity scales as $\mathcal {O}\!\left( \frac{n}{\nu }T\right) $ $O (\frac{n}{ν}, T)$ , maintaining 2–5 s update times for fleets exceeding 200 vehicles. Reproducible artefacts and a staged pathway from hardware-in-the-loop to BVLOS trials support near-term healthcare deployment.