Skip to content

Most Significant Contributions

1. Radio Propagation Model for LPWAN

Illustration of LPWAN propagation model In [el-chall3A2019qy], we conducted an extensive study of radio propagation characteristics within the sub-GHz spectrum. We performed a series of measurement campaigns in diverse environments, including indoor and outdoor scenarios, as well as urban and rural settings. Based on our empirical findings, we developed new path loss models tailored for LoRaWAN communications and made our measurement results accessible in an open data format on https://zenodo.org/record/1560654. Our path loss models have since become widely used, with several researchers employing them in key applications. For example, they have been used to estimate coverage in dense Amazon vegetation [O25], build multi-floor models [O26], and design cognitive-based LoRaWAN solutions [O27]. They have also been validated through large-scale measurements in German urban environments [O28]. This work was part of a funded multidisciplinary project focused on LPWANs for smart agriculture. I served as a co-applicant for this project and was the scientific leader for path loss modeling activities. I also supervised the postdoctoral fellow involved in these activities.

2. Algorithms for Enhancing Performance and Reliability of LPWAN

Illustration of LoRaWAN optimization Over the past six years, our research has been central to advancing LPWAN technologies. We began with [Loubany%3A2020gd], providing unique insights into the challenges of LoRaWAN in dense deployments. While densification is crucial for improved coverage, it often leads to congestion on lower spreading factors (SFs), affecting overall performance. To address this trade-off, we introduced original optimization models for multi-gateway networks in [Loubany%3A2020gd] and enabled autonomous decision-making through ML in [Ta%3A2019eu]. Our work also focused on energy efficiency in [Loubany%3A2023bq] and [Loubany%3A2023rg], and included battery-less devices with energy-harvesting capabilities in [Loubany%3A2022if]. Notably, we developed a LoRaWAN simulator featuring RL algorithms for SF allocation. The open-source code for the simulator is available at https://github.com/tuyenta/IoT-MAB and documented in [Ta%3A2019yg]. This simulator has enabled many researchers to replicate and build upon our work. For example, authors in [O29] and [O30] used it to address RL convergence problems or to devise deep RL approaches. Finally, we tackled the challenge of multi-operator co-location within unlicensed spectrum, as documented in [Khawam%3A2022wd], [Fawaz%3A2020la], and [Fawaz%3A2021pi]. We employed advanced methodologies such as long short-term memory networks and game theory to optimize network performance without compromising the privacy of each operator. I served as the scientific leader of this research stream, co-supervising 10 HQP, including three post-doctoral researchers, three PhD candidates, and four master's students. This work was conducted through international collaborations between major institutions in France and Lebanon, fostering diverse research profiles and facilitating multiple student mobility opportunities.

3. Full-Duplex Communications

Illustration of full-duplex communication We have made substantial contributions to full-duplex communications, a technology that promises to nearly double spectral efficiency in cellular networks. When we began our research, most work in full-duplex communications focused primarily on self-interference cancellation. Recognizing that other factors could significantly impact system performance, we demonstrated that radio resource management plays an equally important role, encouraging the research community to consider these aspects as well. We introduced optimal scheduling models for full-duplex OFDMA wireless networks in [fawaz%3A2018vn] and applied these models with different scheduling objectives, considering a non-full buffer traffic model [fawaz%3A2018ly]. We showed in [Fawaz%3A2020uo] and [Fawaz%3A2019wd] that with optimal scheduling, full-duplex communications can nearly double user throughput and reduce waiting delays by half compared to half-duplex networks. Next, we presented an RL-based scheduling approach in full-duplex networks [Fawaz%3A2021jt], eliminating the unrealistic assumption of perfect channel state information [fawaz%3A2018rt]. Finally, we developed joint scheduling and power allocation algorithms [Fawaz%3A2019fv] for multi-cell full-duplex wireless networks [Fawaz%3A2023wu]. Our results indicated that the gains from full-duplex communications are closely tied to interference mitigation achieved through cell deployment and highlighted the importance of cell cooperation in fully realizing these gains. I served as the scientific leader of this research stream, supervising a student who completed both their master's and Ph.D. degrees on this subject. Our research comprehensively explored various aspects of full-duplex communications, leading to significant contributions published in four journals and six conferences, including high-profile venues such as the IEEE Journal on Selected Areas in Communications (JSAC) and Elsevier’s Computer Networks. Additionally, we developed an open-source simulator (https://github.com/Hassan-Fawaz/FDOFDMA-Simulator) that replicates scheduling in FD-OFDMA networks [fawaz%3A2018ve].

4. Game-Theoretic Approaches for Wireless Communications

Over the past six years, our work has significantly advanced the application of game theory in wireless networks. This includes contributions from five HQP co-supervised with Professor Kinda Khawam from Paris-Saclay University. Our research addressed radio resource allocation across various challenges, including spectrum allocation, scheduling, and power management, with applications in cloud and heterogeneous networks [Yassine%3A2023sr], Taleb%3A2020it, full-duplex networks [Khawam%3A2022wd, Fawaz%3A2019wd, Fawaz%3A2019fv], LPWAN [Fawaz%3A2020la, Fawaz%3A2021pi], and network slicing [Awada%3A2022hc, Awada%3A2023le, Awada%3A2023vf]. We applied game theory to model the selfish behavior of network entities—such as devices, base stations, and controllers—aiming to optimize their own utilities. This approach provided fresh insights into resource allocation challenges. We developed tailored cost functions that balance rewards (e.g., throughput) with negative impacts (e.g., interference and energy consumption), analyzed Nash equilibria, and devised efficient distributed algorithms. Interestingly, these algorithms performed comparably to centralized approaches in terms of price of anarchy, convergence time, and computational complexity. Recently, we extended our research to explore the intersection of game theory and machine learning on constrained IoT devices [Durand%3A2024fr, Khawam%3A2024lq, Khawam%3A2023ng]. While conventional methods assume automatic participation in federated learning, we investigated the incentives for IoT devices to process tasks locally or collaborate. Using a federated learning game, we examined how devices balance precision and communication costs, offering new insights into when and how they should join learning coalitions. We calculated the optimal size of such coalitions, assessed their stability [Khawam%3A2023ng], and developed semi-distributed algorithms to facilitate their formation [Durand%3A2024fr].

Illustration of game-theoretic wireless communication

  • Software/Dataset: [None]

5. Dissemination of IoT Expertise

I have been dedicated to sharing my expertise in wireless IoT with the broader scientific community. In response to open calls, I was selected alongside Dr. Melhem El Helou to present tutorials at major international conferences, including the IEEE 5G World Forum 2019 [P3] and the International Conference on Telecommunications (ICT) 2018 [P4], as well as at specialized venues such as WPMC 2019 [P2] and SPECTS 2018 [P6]. Our tutorials addressed LPWAN challenges and IoT applications, attracting around 20 attendees per event, including academic researchers and industry experts. Additionally, I was invited by IEEE Region 8 to deliver a keynote on using ML in LPWAN [P1] and by the Order of Engineers in Lebanon to present on IoT and smart cities [P5], each event drawing around 50 attendees. In 2022, I led the establishment of a new master’s program in IoT through a collaboration between Saint Joseph University in Lebanon and Paris-Saclay University in France. This program contributed to training HQP in the IoT field, with most students securing apprenticeship contracts with leading industry players. Furthermore, I co-organized a public IoT event in Lebanon (https://usj.edu.lb/iot-leb19), which drew a diverse audience of around 100 industry professionals and academics. I also coordinated a concurrent hackathon (https://usj.edu.lb/iot-leb19/#Hackathon), providing 20 participants with hands-on experience in LPWAN technologies.

Illustration of IoT dissemination activities