We can understand control theory as a theory of the designing complex systems by the interconnection of smaller components. It has been historically developed mainly for macroscopic systems, where the thermal fluctuations have little impact. When the variables have a physical meaning (speed, forces, voltages, etc.) and take small values comparable in magnitude to the thermal noise, one has to rewrite or complement the rules of control theory so as to make them abide by stochastic thermodynamics. Even the most basic rules, such as the way to interconnect systems meaningfully, are expected to change at this scale. As a consequence one hopes to quantify the minimum energy consumption required by thermodynamics to produce a certain desired behaviour for an engineered system. For instance, it is obvious that an amplifier (delivering a signal that contains more power than the input signal) must consume energy. This internal power source is less obvious to quantify with systems aimed at processing information, such as denoising (e.g. the celebrated Kalman filter) or communication (a channel transferring information from one place to another). Therefore the energetic and information flows are strongly related. These relations have been studied in the recent years in the field of stochastic thermodynamics, either in general terms or for simple examples such as the Kalman filter or bit erasure. It will be our goal in this project to explore the trade-off between bit rate, power, and reliability (error rate), not only in general, but for specific classes of engineering devices, chosen according to their function (filtering, communication, digital computation), technology (overdamped vs underdamped implementation for instance) or complexity (linear vs nonlinear behaviour), leading to bounds that can be compared usefully to existing or projected technologies, e.g. low-energy processors, energy harvesting or 5G communication technologies. Prospectively, it may also lead to interesting comparisons with devices of similar functions in neurosciences (neurons, axons, photoreceptors, etc.). The team on this project will be composed of Massimiliano Esposito (main Host at University of Luxemburg, at the Complex Systems and Statistical Mechanics group of the Physics and Materials Science Research Unit), Matteo Polettini (permanent researcher in the same group) and Jean-Charles Delvenne (Visitor in the same group, on sabbatical from University of Louvain, Belgium, at ICTEAM: Institute of Information and Communication Technologies, Electronics and Applied Mathematics). Their high-profile and complementary expertise in Statistical Physics, Control Theory and Engineering will ensure a fruitful dialogue on these topics. Interactions with other members of ICTEAM (for computation and communication engineering) or international universities (KTH, Caltech, MIT for control engineering) are expected to take place, for which JCD will act as catalyst.