Robust Task Offloading for UAV-enabled Secure MEC Against Aerial Eavesdropper

TL;DR

This paper proposes a robust task offloading framework for UAV-enabled secure multi-access edge computing, addressing security threats from aerial eavesdroppers and considering uncertainties in task complexities.

Contribution

It introduces a distributionally robust optimization approach with chance constraints for secure UAV-based MEC, incorporating task uncertainty and optimizing deployment and offloading.

Findings

The proposed algorithm effectively handles uncertainties with only 5% additional energy consumption.

The approach improves security and robustness in UAV-enabled MEC systems.

Simulation results validate the effectiveness of the robust optimization method.

Abstract

Unmanned aerial vehicles (UAVs) are recognized as a promising candidate for the multi-access edge computing (MEC) in the future sixth generation communication networks. However, the aerial eavesdropping UAVs (EUAVs) pose a significant security threat to the data offloading. In this paper, we investigate a robust MEC scenario with multiple service UAVs (SUAVs) towards the potential eavesdropping from the EUAV, in which the random parameters such as task complexities are considered in the practical applications. In detail, the problem is formulated to optimize the deployment positions of SUAVs, the connection relationships between GUs and SUAVs, and the offloading ratios. With the uncertain task complexities, the corresponding chance constraints are constructed under the uncertainty set, which is tricky to deal with. Therefore, we first optimize the pre-deployment of SUAVs by the K-means algorithm. Then, the distributionally robust optimization method is employed, and the conditional value at risk is utilized to transform the chance constraints into convex forms, which can be solved via convex toolkits. Finally, the simulation results show that with the consideration of uncertainties, just 5% more energy is consumed compared with the ideal circumstance, which verifies the robustness of the proposed algorithms.