Robust secure task offloading in multi-tier aerial edge computing under

doi:10.1016/j.comnet.2026.112571

Robust secure task offloading in multi-tier aerial edge computing under

2026-07-01

0

PRE

AI

M

Mohammed A. AlAqil

H

Hend Khalid Alkahtani

N

Nasser Allheeib

M

Muhammad Kashif Saeed

DOI：10.1016/j.comnet.2026.112571

Original

Original request for help

Save

Abstract

En 中文

The 6G era's computation-intensive applications drive the deployment of multi-tier aerial-ground edge clusters, but wireless offloading exposes these networks to severe eavesdropping. Guaranteeing physical-layer security (PLS) under spatial uncertainty degrades worst-case secrecy capacities, forcing latency-critical tasks to local execution and bottlenecking scalability. In addition, co-channel interference among concurrent aerial users and the limited onboard energy of mobile platforms-shared between communication and propulsion introduce fundamental performance trade-offs in practical deployments. We propose a secure collaborative offloading framework featuring a mobile protective aerial jammer (PAJ) to actively suppress eavesdroppers under bounded location uncertainty. Formulating the joint secrecy-latency minimization problem as a highly coupled mixed-integer nonlinear program (MINLP), we develop the provably convergent Alternating Spatio-Combinatorial Optimization (ASCO) framework. ASCO employs convex-analytic transformations to map semi-infinite spatial uncertainty into deterministic robust constraints, enabling an energy-aware PAJ trajectory design. Concurrently, Karush-Kuhn-Tucker (KKT)-derived shadow prices resolve the NP-hard combinatorial resource allocation, ensuring efficient utilization of edge computing resources under interference and capacity constraints. Extensive simulations under interference-aware channels and propulsion-constrained energy budgets demonstrate that ASCO achieves up to 39.6% latency reduction compared to fully local computation, 23.9% improvement over advanced SCA-based baselines, and up to 47.6% energy savings relative to heuristic approaches. Moreover, ASCO maintains robust performance under severe spatial uncertainty (k = 25 m), achieving approximately 3.3 s latency while outperforming non-robust schemes by nearly 30%. This unified optimization framework provides a practical and scalable solution for secure, low-latency offloading in next-generation aerial edge networks, supporting mission-critical applications such as autonomous coordination and emergency response systems.

Keywords:

Mobile edge computing (MEC)

Physical-layer security (PLS)

Unmanned aerial vehicles (UAVs)

Robust optimization

Task offloading

Mixed-integer nonlinear programming (MINLP)

Robust secure task offloading in multi-tier aerial edge computing under

Abstract

Journal

Organization

Cited Papers