Novel Information Processing Methodologies for Intelligent Sensor Networks

Start Date: 2010 End Date: 2014
PIs: João Manuel de Freitas Xavier (IST/UTL) and Bruno Sinopoli (CMU)
Co-PIs: Miguel Rodrigues (University College London, UK) and Nuno Brás (LMIT)

Dual Degree Ph.D. Student: Dragana Bajovic (Electrical and Computer Engineering), and Vinay Prabhu (Electrical and Computer Engineering)

Teams: Instituto Superior Técnico/Unidade Técnica de Lisboa (IST/UTL), Instituto de Sistemas e Robótica (ISR), Faculdade de Ciências da Universidade do Porto, Instituto de Telecomunicações (IT), Carnegie Mellon University (CMU)

Company: wisemetering (LMIT)

Keywords: Most informative sensors selection; Cyber physical systems security; Information theory; Distributed sensing and actuation

Over the past few years, we have witnessed tremendous advances in the miniaturization of low cost electro-mechanical devices that exhibit several sensing capabilities (e.g. temperature, pressure, vision, etc) and possess significant embedded processing and low-range wireless communication. This triggered interest in wireless sensor networks (WSN) where large collections of such small wireless sensor nodes are scattered across critical geographical areas or infrastructures to execute a variety of monitoring and acting tasks. While the use of WSN is crucial to assess the state of a physical system, the ultimate engineering goal is to control such system to perform certain desired tasks. We refer to systems that close the loop around wireless sensor networks as cyber-physical systems (CPS). Cyber-physical systems are physical and engineered systems the operations of which are monitored, coordinated, controlled and integrated by a computing and communication infrastructure. Examples of CPS include medical devices and systems, aerospace systems, transportation vehicles and intelligent highways, defense systems, robotic systems, process control, factory automation, building and environmental control and smart spaces.
To realize the immense potential benefits of CPS in real-life applications, a plethora of novel problems lying at the intersection of statistical signal processing, control, communication theory, and distributed optimization, must be addressed. Indeed, it is clear that to achieve desired network-wide detection, estimation or control objectives, system designers should deal efficiently with simultaneous challenges, typically tackled in separate engineering disciplines. The challenges include securing energy-efficient operation as the nodes are usually battery-operated and constrained to small power budgets, developing distributed analogues of centralized signal processing and control algorithms, coping with the dynamic communication topology interconnecting the sensing nodes, dealing with the random finite-bandwidth wireless medium, etc.
The vision of this project is to advance CPS/WSN design by integrating expertise along several key dimensions.