The growing demand for ultra-high data rates and spectrum efficiency is driving the exploration of Terahertz (THz) communication as a promising candidate for future wireless networks. With the potential to deliver data rates of up to several terabits per second (Tbps), THz communication opens new avenues for next-generation applications such as ultra-high-definition video streaming, virtual reality, and Industry 4.0. However, because of high pathloss, molecular absorption, and limited transmission range in the THz frequency band, the practical deployment of THz communication faces significant challenges. To overcome these limitations, Intelligent Reconfigurable Surfaces (IRSs) have emerged as a solution to dynamically manipulate the wireless propagation environment, improving coverage and system performance.
In this context, the present thesis investigates the integration of IRSs in THz communication networks to mitigate the propagation challenges and enhance system efficiency. More specifically, we propose novel algorithms and models to optimize the performance of IRS-assisted THz networks, with a particular focus on resource management and user association under realistic conditions, including imperfect channel state information (CSI). By leveraging the IRS’s ability to adjust reflection properties and direct signals, the proposed framework is found to significantly improve data rates and reliability.
Chapter 2 presents the application of IRSs in THz networks with imperfect CSI. Specifically, we present an angle-based trigonometric channel model to facilitate the performance evaluation of IRS-aided THz networks. In addition, seeking to maximize the sum rate, we formulate the transmitter (Tx)-IRS-receiver (Rx) matching problem, which is a mixed-integer nonlinear programming (MINLP) problem. To address this non-deterministic polynomial-time hard (NP-hard) problem, we propose a Gale-Shapley algorithm-based solutions to obtain stable matching between transmitters and IRSs, and between receivers and IRSs, in the first and second sub-problems, respectively. The impact of the transmission power, the number of IRS elements, and the network area on the sum rate are then investigated. Furthermore, the proposed algorithm is compared to an exhaustive search (ES), nearest association, greedy search, and random allocation to validate the proposed solution. The results of our complexity and convergence analysis demonstrate that the computational complexity of our algorithm is lower than that of the ES method.
Chapter 3 presents the use of IRSs to enhance THz communication for the Industrial Internet of Things (IIoT), a key enabler for building large-scale interconnected industrial systems. In this chapter, we propose an IRS-aided THz communication system for IIoT networks. A power allocation and joint IIoT device and IRS association (JIIA) problem is formulated as a MINLP problem aimed at maximizing the sum rate under imperfect CSI. Due to the NP-hard nature of the problem, it is decomposed into sub-problems that are then solved iteratively. A Gale-Shapley algorithm-based solution is proposed to achieve stable matching between uplink and downlink IRSs, and the performance is validated against exhaustive search, greedy search, and random association methods. The results of our complexity analysis show that the proposed approach significantly reduces computational overhead as compared to exhaustive search methods while maintaining high efficiency.
In Chapter 4, we explore the coexistence of enhanced mobile broadband (eMBB) and ultrareliable low-latency communication (URLLC) services within IRS-assisted THz networks. The simultaneous support of eMBB and URLLC services in the same network introduces complex resource management (RM) challenges. To address this concern, we formulate a joint power and user allocation (JPUA) problem aimed at maximizing the achievable data rate for eMBB users. Specifically, we develop a one-to-many matching game model to optimally allocate IRSs to eMBB users, ensuring efficient resource utilization. Our simulation results demonstrate that the proposed scheme outperforms baseline methods in terms of eMBB sum data rates.
Chapter 5 examines the coexistence of eMBB and URLLC services within an IRS-assisted THz multi-cell network, which presents a more complex RM challenge. In this chapter, we propose a joint power, user, and service allocation (JPUSA) framework formulated as a multiobjective optimization problem aimed at maximizing the eMBB data rate while ensuring URLLC reliability. The NP-hard MINLP problem is tackled by converting it into a single-objective optimization problem using the weighted sum method. The problem is further decomposed into eMBB and URLLC RM sub-problems. A one-to-many matching game is then used to allocate IRSs to eMBB users, while a puncturing technique assigns eMBB resources to URLLC users. Our simulation results demonstrate that the proposed scheme significantly outperforms baseline methods, particularly in terms of eMBB sum data rate and URLLC reliability. The proposed algorithm’s sum rate for eMBB users closely approximates that of the ES method, while maintaining lower computational complexity as compared to ES, making it highly efficient for multi-cell network deployments.
Finally, in Chapter 6, we expand upon the coexistence of further eMBB (FeMBB) and extreme URLLC (eURLLC) services within an IRS-assisted hybrid millimeter-wave (mmWave) and THz network. To address the line-of-sight (LOS) blockage issue prevalent in such high-frequency bands, we propose a hybrid deployment scheme incorporating both terrestrial and aerial IRSs. A novel concept of sub-IRSs, where large IRS surfaces are divided into smaller units to cater to multiple users simultaneously, is then introduced. The coexistence of FeMBB and eURLLC services in the same network further complicates resource allocation (RA). To address this concern, a multi-objective optimization problem is formulated to jointly optimize power, user, and service allocations, aiming to maximize the FeMBB data rate while ensuring eURLLC reliability. Due to the NP-hard nature of the problem, it is converted into a single-objective optimization problem using the weighted sum method and then decomposed into two subproblems for FeMBB and eURLLC RA, respectively. A many-to-many matching game is used to allocate IRSs to FeMBB users, optimizing resource utilization. Our simulation results indicate that the proposed scheme not only surpasses traditional baseline methods but also achieves a FeMBB sum rate close to that of exhaustive search methods, with a much lower computational complexity.
| Date | 30 May 2025 |
|---|
| Original language | American English |
|---|
| Awarding Institution | - École de technologie supérieure
|
|---|
| Supervisor | Georges Kaddoum (Supervisor) |
|---|