Loading...

Physical Layer Security Utilizing Reconfigurable Intelligent Surfaces

Rahimi, Masoud | 2024

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58003 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Behroozi, Hamid; Mirmohseni, Mahtab
  7. Abstract:
  8. With the increasing use of wireless communications, the need for optimal utilization of bandwidth and energy has become undeniable. Reconfigurable Intelligent Surfaces (RIS) have emerged as a promising technology for the efficient use of telecommunications resources. One proposed model of these surfaces is the Simultaneously Transmitting and Reflecting RIS (STAR-RIS), which, unlike conventional reflective surfaces, can simultaneously reflect part of the incoming waves and transmit part of them. This study examines the potential use of these surfaces in simultaneous wireless information and power transfer networks. In the defined problem, energy receivers are not reliable, and there is a possibility that they might attempt to eavesdrop on the information users' signals. Additionally, Rate-Splitting Multiple Access (RSMA), an innovative technique in wireless communications, is utilized. Given the complexity of the problem and the need for real-time solutions, efforts have been made to solve the issue using Deep Reinforcement Learning (DRL) algorithms. One of the challenges of conventional reinforcement learning algorithms is their suboptimal use of data, and they generally face challenges when learning in complex problems. In this study, by integrating unsupervised learning with reinforcement learning, the speed and performance of learning were significantly improved. The defined problem aims to find parameters that maximize the secure transmission rate for information users (IR) while simultaneously allowing untrusted energy receivers (UER) to access the required power without being able to eavesdrop on the information users' signals. Simulation results showed that the proposed algorithm significantly outperformed traditional optimization algorithms and conventional reinforcement learning algorithms
  9. Keywords:
  10. Intelligent Reflective Surfaces ; Deep Reinforcement Learning ; Deep Deterministic Policy Gradient ; Rate Splitting Multiple Access ; Pre-Trained Deep Deterministic Policy Gradient ; Reconfigurable Intelligent Surface (RIS) ; Physical Layer Security

 Digital Object List

 Bookmark

No TOC