Gia-Huy Nguyen

This research investigates the secrecy performance and energy efficiency of a dual-unmanned aerial vehicle (UAV) computation offloading system that leverages uplink non-orthogonal multiple access (NOMA) designed for Internet of Things (IoT) networks operating under Nakagami- $m$ fading propagation. Specifically, a multiantenna UAV-enabled mobile-edge computing (MEC) server offers proximate computation services to clusters of resource-constrained edge devices (EDs), while utilizing selection combining (SC) and maximal ratio combining (MRC) schemes to enhance the offloading reliability. Simultaneously, a secondary UAV is deployed as a dedicated aerial friendly jammer (FJ) to fortify data confidentiality against eavesdropping threats. Accordingly, the system’s secrecy computation offloading performance is evaluated through a novel closed-form formulation termed secrecy successful computation probability (SSCP). This metric uniquely incorporates the detrimental effects of imperfect channel state information (iCSI) and imperfect successive interference cancellation (iSIC), both of which are critical real-world factors that reduce the effective signal-to-interference-plus-noise ratio (SINR) of legitimate links at UAV-MEC. This reduction in SINR directly leads to a significant decrease in the SSCP and overall reliability of the secure offloading process, due to the impaired ability to successfully decode information at the legitimate UAV. Building upon the SSCP derivation, the secrecy energy efficiency (SEE) metric is adopted to analyze the fundamental trade-off between security and energy. Furthermore, the problem of jointly enhancing computation offloading security and energy efficiency is formulated as a SEE maximization problem under the constraints of dual-UAV altitudes and positions, the task-offloading ratio (TOR), the power-allocation ratio (PAR), and the number of UAV-MEC antennas. Hence, an optimization approach grounded on the Ananya algorithm is employed, as its superiority is validated against the benchmark particle swarm optimization (PSO) algorithm and baseline arbitrary configurations, achieving a significantly higher SEE solution while converging over 35% faster. Ultimately, the accuracy and feasibility of the proposed system model are consistently verified through extensive numerical simulations of sixth-generation (6G)-envisioned conditions, accounting for a variety of key system parameters.