The Information, Learning, Optimization and Communication Sciences (ILOCOS) team performs fundamental and applied research on the design of future wireless telecommunication systems. The team members apply mathematical tools such as stochastic optimization, queuing theory, information theory and machine learning to the modeling, cross-layer design, optimization and control of these systems. While the focus of the team is on 5G and 6G, other types of systems such as smart grids and social networks are also addressed. As of application domains, the systems are designed for supporting mobile broadband services with high throughput demands, along with Internet of Things (IoT) services. For mobile broadband, massive and multiuser MIMO technologies as well as reconfigurable intelligent surfaces are designed and optimized with an objective to achieve a larger capacity and a ubiquitous coverage. For IoT applications, not only systems related to massive machine type communications are considered, but also those related to Industrial applications such as industry 4.0 and autonomous vehicles, with an effort to integrate control and communication aspects in the system design.
One of the objectives of the ILOCOS team is the design of the radio interface for future generations of wireless networks. Technologies such as massive MIMO, non-orthogonal multiple access and Reconfigurable Intelligent Surfaces (RIS) are perfected for different types of communication links (cellular, D2D, HetNets). New frequency carriers in the mmWave and the TeraHz bands introduce new challenges in the deployment, dimensioning and optimization of these networks. New channel models that account for the non-linearity and the hardware imperfections in general are needed for theses technologies and bands. Information theory, queueing theory, stochastic optimization and machine learning tools are used for the design and optimization of the radio interface resulting from the aggregation of these technologies.
This axis focuses on the convergence of the world of wireless telecommunications and that of the Internet of Things (IoT). The latter includes the massive connection of objects (for smart city applications for example), as well as critical IoT (industry 4.0, tactile Internet, etc.). The radio channel access mechanisms have been redesigned to ensure low latency and high reliability. The virtualization mechanisms of network functions are exploited to allow flexible implementation (concept of slicing) and to integrate processing and storage resources in the overall framework of resource allocation (concept of fog computing). Queuing, stochastic optimization and artificial intelligence tools are used to achieve these goals.
New industrial applications of IoT include the control of machines or robots (Industry 4.0) or distributed systems (swarms of drones, platoons of vehicles). The classic approach to designing a generic telecommunications network that meets the requirements of any application is not optimal in this context. Based on the understanding of the impact of the quality of the radio channel on the performance of the control system, we are working on a joint design of the communication and control systems, considering two levels of control (local and collaborative), these are interacting and depend on the quality and availability of the communication network. Optimal control and artificial intelligence tools are used in this research area.
Networked applications generating machine type data (monitoring applications, control of machines, etc.) are expected to generate a huge amount of traffic in the next few years. In this context, a new communications paradigm has emerged, that is semantic and goal-oriented communications. The activities include sampling theory to generate the right/significant piece of information on-time, the development of new metrics to measure the effectiveness of the communication (i.e to ensure the goal of the communication is achieved), and a new design of the network (channel access, scheduling, routing, etc.) to transport the sampled packets, while exploiting the novel defined metrics. Markov decision theory and machine learning tools allow this new network design.