Annonce
3-year Postdoc in Network Information-Theoretic Sensor Management for Multi-Target Surveillance
15 Septembre 2023
Catégorie : Post-doctorant
We are looking for postdoctoral candidates with a strong background in statistical signal processing, information theory, and good programming skills. The successful candidate will develop the underpinning methods and algorithms required for autonomous distributed sensor management and fusion in challenging environments. The research work conducted by the successful candidate will deliver key advances in intelligent sensing to enable continuous and adaptive surveillance in dynamic environments. More specifically, the work will involve the proposal of metrics for quantifying the information perceived by different sensors on multiple stochastic processes. Subsequently, these metrics will guide the development of algorithms for sequentially estimating the state of these processes, fuse information from heterogeneous sensors, and allocate resources. These resulting algorithms will be applied to multiple target tracking with sensor networks.
Sequential Bayesian inference methods are commonly employed by sensor networks in order to detect and estimate (i.e., track) a multi-target process. Due to the advent of the Internet-of-Things and other extensive sensor networks, algorithms that judiciously manage the communication, sensing, and energy resources of such networks are crucial for efficient inference under constraints.
The project is conducted at Télécom SudParis and involves the development of algorithms that are able to track multiple targets, fuse information from heterogeneous sensors, and allocate resources based on information-theoretic criteria. Building on recent developments by the investigators in multi-target tracking and distributed sensor fusion, this work programme will develop methods based on point process theory, which is designed to accommodate uncertainty in the states of individual targets and the number of targets. Information-theoretic metrics tailored for point processes will be proposed as well as optimization methods that employ these metrics in order to allocate sensor resources and refine the knowledge of the scene.
Applications, consisting in a CV, a list of publications, and any other supporting document should be sent to augustin.saucan@telecom-sudparis.eu. For an informal discussion to find out more about the role please contact Augustin Saucan at augustin.saucan@telecom-sudparis.eu.