Abstract
This paper focuses on securing confidential communication in multiple intelligent reflecting surfaces (IRS) assisted terahertz (THz) systems, where a potential eavesdropper can intercept either the base station (BS)-IRS link or the IRS-user link. Notably, the secure transmission may be intercepted and blocked by the eavesdropper due to the blockage-prone nature in THz bands. To characterize the blocking effects of the eavesdropper, the blocking-based path loss is first investigated. With the imperfect eavesdropper channel state information (ECSI), the worst-case secrecy rate (WCSR) is derived, and a joint optimization problem of hybrid beamforming at the BS and reflecting beamforming at the IRS is formulated. For the BS-IRS link eavesdropping, the zero-forcing (ZF) principle-based hybrid beamforming and the closed-form phase shifts of multiple IRSs are respectively proposed. For the IRS-user link eavesdropping, an iterative algorithm is proposed to tackle the non-convex optimization problem with a given information leakage threshold. Finally, a robust secure transmission strategy for multi-eavesdropper systems is further investigated. Simulation results demonstrate that compared with blockage-unaware scenarios, our proposed scheme can resist the adverse effects of the blockage-prone nature of THz waves on information security, and significantly boost secrecy performance.
Introduction
Terahertz (THz) frequencies are considered the "final frontier" for 6G, promising terabit-per-second data rates. However, THz waves suffer from significant propagation loss and are easily obstructed by physical objects. To mitigate coverage issues, Intelligent Reflecting Surfaces (IRS)—passive element arrays that steer beams around obstacles—are employed.
A critical security challenge arises when an eavesdropper (Eve) intercepts these narrow THz beams. In the THz band, Eve’s physical presence can actually block part of the signal she is trying to intercept. This paper addresses the "blocking eavesdropper" scenario in a multi-IRS system, accounting for imperfect Eavesdropper Channel State Information (ECSI).
Methodology and Blockage Modeling
The authors utilize a Worst-Case Secrecy Rate (WCSR) objective to ensure security under unfavorable channel conditions. A key component is the METIS-based shadowing screen model, which uses knife-edge diffraction to quantify power loss when Eve obstructs the Line-of-Sight (LoS) path.
The system employs a partial-connected hybrid beamforming structure at the Base Station (BS). Optimization is split into two scenarios:
- BS-IRS Link Eavesdropping: A Zero-Forcing (ZF) principle nullifies signals leaked toward Eve, paired with closed-form IRS phase shifts.
- IRS-User Link Eavesdropping: A Semidefinite Programming (SDP)-based iterative algorithm manages the non-convex optimization of beamforming and reflecting patterns.
Key Insights: The Blocking Paradox
A counter-intuitive finding is that Eve’s attempt to eavesdrop can sometimes improve the secrecy rate. Because THz beams are extremely narrow, Eve's physical obstruction often degrades her own intercepted signal quality faster than it affects the legitimate user (Bob). By being "Blockage-Aware," the system can exploit this physical obstruction to enhance security.
Furthermore, using multiple distributed IRS units provides "spatial degrees of freedom." Spreading the signal across multiple surfaces makes it significantly harder for a single eavesdropper to intercept or block all paths simultaneously.
Results and Performance
Simulation results, utilizing the HITRAN database for molecular absorption and METIS for blockage, highlight several key findings:
- Secrecy Gains: The blockage-aware scheme recovered nearly 15% of the secrecy performance lost in blockage-unaware scenarios.
- IRS Scaling: Increasing IRS elements (e.g., from 36 to 64) creates sharper beams, reducing leakage.
- Multi-IRS Superiority: Three distributed IRS units achieved more than double the secrecy rate of a single large IRS with the same total elements.
- Efficiency: The iterative algorithm for the IRS-user link typically converges within 13 to 18 iterations, suggesting real-time feasibility.
Conclusion
This research demonstrates that accounting for the physical blocking effects of eavesdroppers is vital for securing THz communications. Future work will look into handling imperfect CSI for legitimate users and real-time beam-tracking in highly dynamic environments with moving blockers.
Summary updated