As part of these activities, we are looking for a research engineer to strengthen the team and carry out research and development work in the area of resource allocation, orchestration protocols and optimization for future wireless communication systems. These systems will include communication, computing and storage resources, as part of edge computing networks, enabling new applications such as artificial intelligence deployed at the edge (edge intelligence). The aim is to develop holistic systems where heterogeneous services and resources coexist on the same infrastructure, with different objectives in terms of reliability, sustainability, energy efficiency and exposure to electromagnetic fields, coherently with international recommendations. Beyond energy efficiency, the use of renewable energy sources integrated into networks represents an opportunity to reduce their carbon footprint, but also a challenge in terms of managing these resources (which are by definition less reliable and more intermittent than carbon-based resources) jointly with communication and computing.

The goal  is to develop novel orchestration algorithms (based on, e.g., stochastic optimization, Markov Decision Process, etc.) and test them through numerical simulations, to study typical trade-offs between energy, latency, reliability and exposure to electromagnetic fields, as well as the performance of new applications such as edge intelligence in radio access networks (RANs). Modeling these networks and identify gaps with respect to the state  of the art will be part of the candidate's activities.

These models and algorithms will take into account service heterogeneity, service coexistence, dynamically evolving requirements (e.g. traffic, number of users, QoS), the changing environment and several constraints on energy, latency, exposure to electromagnetic fields and performance. As the network may be confronted with a myriad of predictable and unpredictable events impacting the overall performance, the integration of machine learning-based optimization processes will be useful to manage and reconfigure the entire network (radio and computation adaptation, resource allocation and energy resource management, data representation and compression to achieve required performance, etc.). For example, in the context of edge intelligence, orchestrators will be asked to optimize resources to achieve the best energy-latency compromise under reliability constraints for a machine learning model deployed in edge computing resources. This means choosing, among others, the association of users to access and computing nodes, as well as beamforming configuration, bandwidth allocation (e.g., between wireless access and backhaul), the level of compression of data transmitted in uplink by users, and the portion of computations performed locally.

As part of this study, the candidate will interact with people in the department who will provide inputs to enrich the study.

The mission requires knowledge of wireless communications, in particular detailed knowledge of MAC and RRM protocols for cellular networks (e.g., 5G NR type) in order to define and evaluate resource allocation strategies for 5G networks and beyond. One of the key criteria is the knowledge of optimization techniques including, among others: convex, combinatorial, stochastic. Solid experience with Matlab and/or knowledge of the Python language (with focus on Machine Learning and Artificial Intelligence, reinforcement learning) is required.

Excellent spoken and written English is required.

Experience on an NS-3 network simulator and mastery of C and C++ languages is a plus.