Networked Controlled System Team

innovative research is in control and estimation of networked systems, with a broad spectrum of applications. Topic description: In anonymous (or ‘privacy-preserving’) networks, agents are not able or not willing to provide an ID label uniquely describing their identity. This might be due to technological limitations (when sensor or robotic networks are self-configured), or to privacy concerns (e.g., when agents represent smart-phones or computers, with corresponding human users). In this anonymous setting, even simple tasks such as computing the number of nodes become very difficult: it has been proved that it is impossible to compute the size of the network with probability one by using bounded communication, memory and computation-time resources. The literature provides a variety of algorithms able to compute the number of nodes with some good (albeit non-zero) probability, and/or to compute it using resources growing with the network size. Algorithms are based: on Markov chains (e.g., first return time of a token sent around the network), see [1,2]; on ML estimation and order statistics, together with consensus iterative algorithm [3]; on a variation of consensus algorithm [4]; or on system identification to identify the order of a dynamical system run on the network [5]. The proposed work is to study and compare such algorithms. Implementation can be done in Matlab or C. The comparison should aim at understanding advantages and weaknesses of the algorithms, the restrictions to their applicability, and how performance changes with the network graph properties. An application-oriented student can exploit the opportunity of a real-life implementation on the sensor network FIT/IoT LA