An arXiv preprint reports higher simulated wireless secrecy when a UAV-mounted intelligent reflecting surface (IRS)—a passive array with controllable reflection settings—is optimized in position and orientation together with the base station’s beamformer. The proposed six-dimensional-plus-passive-beamforming designs outperformed orientation-only and fixed-IRS baselines in the paper’s numerical comparisons.
The study’s main measure, achievable secrecy rate, is the positive-part difference between the legitimate user’s and eavesdropper’s log-rate terms. Its modeled setup combines a base station, a single-antenna legitimate user, a sensing target called Eve that may act as an eavesdropper, and a passive UAV-mounted IRS with a uniform planar array.
How the virtual test was run
The work models a downlink secure integrated sensing and communications (ISAC) link—a system that combines data transmission with sensing—and evaluates it with numerical simulations. Its solver alternates three optimization blocks: linearized ADMM for active beamforming, warm-started particle swarm optimization for the IRS pose, and Riemannian gradient descent for the passive reflection vector.
The numerical setup used a base station with 32 transmit and 32 receive antennas, a 3.6 GHz carrier, 1 MHz bandwidth and maximum transmit power of 30 dBm. The UAV-mounted IRS altitude was fixed at 150 metres, and the comparison covered multiple schemes to assess six-dimensional mobility.
What the simulations showed
Across the schemes, the objective value rose rapidly during the first few alternating-optimization iterations and then stabilized. The paper gives this as a directional convergence pattern, not a numerical summary: exact iteration counts, repeated-run variability and uncertainty intervals were not reported.
The proposed six-dimensional-plus-passive-beamforming schemes were reported to have higher secrecy rates than orientation-only and fixed-IRS designs. The large-region configuration also had a higher secrecy rate than the small-region version, although the size of that difference was not given.
Secrecy rate increased as the number of IRS reflecting elements increased, then the gains gradually saturated beyond a threshold. The paper does not specify that threshold or provide exact rates, confidence intervals or statistical tests.
The authors interpret six-dimensional pose optimization as a key spatial-domain mechanism for physical-layer security when the legitimate user and Eve are angularly close from the base station’s perspective. That explanation is drawn from the model and simulations, not independently validated.
A result still inside the model
The evidence is limited to an analytical channel and system model with numerical simulations. All links are modeled as line of sight, so the study does not establish performance in obstructed or fading environments, and it offers no hardware or field validation.
The evaluation uses one legitimate user and one potential eavesdropper or sensing target. It does not address multi-user systems, imperfect channel-state information or dynamic UAV trajectories; the authors identify those areas as future work.
The paper is an arXiv preprint, version 1, dated 20 Aug 2026. Its findings are a comparative computational result, not a measured secrecy-rate gain from an operating network.
Paper data and sources
Original title: Secrecy Rate Maximization for UAV-Mounted Six-Dimensional Movable IRS-Assisted ISAC Systems
Authors: Chengye Hong, Botang Shi, Rongkun Zhu et al.
Journal/Repository: arXiv
Status: Preprint, not yet peer-reviewed
First online: 2026-08-20
DOI: Not available
Original paper · Full text