In many engineering applications, physical systems are controlled by a set of external inputs and their behavior is monitored by the system states. The system states may not be measured or observed directly. However, the system states and the external inputs influence the measurements of the system. Hence, in theory, the unknown system states can be characterized by the given inputs and the observed measurements under the assumptions that the system is observable and initial conditions <re known a priori. In general, the measurements are corrupted by noise thus leading to the problem of having to estimate the state of the system. The problem of estimating the states of a physical system on the basis of noisy measurements has been a research issue for decades. However, modern engineering applications such as wireless channel estimation, visual object tracking, radar, appliance detection and so on, motivate current researchers to work further on the estimation problem.

In our work, the Bayesian estimation methodologies are used to address the issues of robustness and efficient implementations in multi-sensor linear/non-linear state estimation algorithms. They can overcome the following constraints: lack of accurate knowledge about initial conditions of the system, lack of accurate process and measurement model, lack of statistical knowledge of process and measurement noise, ad hoc sensor networks with limited energy reservoirs.

We are working on many fields of application in the area of statistical signal processing :

• Object tracking in visual sensor networks:  Object tracking is an extensively studied topic in visual sensor networks (VSN). A VSN is a network composed of smart cameras; VSN capture, process, analyze the image data locally, and exchange extracted information with each other.  The cameras in the VSN monitor the given environment and identify and track an object. The main applications of VSN are indoor and/or outdoor surveillance, e.g., airports, forests, deserts, inaccessible locations, and natural environments.
• Battery internal state estimation: In the modern age, portable and uninterrupted energy sources have become an integral part of human life. Using batteries as a mobile energy source permits applications such as smart phones, notebooks and electric vehicles. Battery management plays a critical role in the ever-increasing demand for battery powered devices. Battery management systems (BMS) use many internal parameters such as the state-of-charge (SoC), the state-of-health (SoH), or the internal impedance to determine the health or the power delivery capability of a battery. In general, these quantities cannot be measured directly, but have to be estimated based on noisy measurements.
• Appliance detection in energy distribution grids: The integration of smart meters improves the current power grid in terms of efficiency, reliability, and energy awareness. Modern smart meters providing fine-grained real-time load demand information allow the consumer not only to know but also to forecast and optimize his/her overall energy consumption. This is often referred to as Appliance Load Disaggregation (ALD) or Appliance Load Monitoring (ALM).
• Localization and navigation exploiting radio signals: Radio signals can be utilized to determine the position of transmitting nodes and implement navigation solutions.