Multiscale Modeling in Beamed Energy Harnessing Applications  AFOSR

A large number of space-based applications exist for the transmission of electromagnetic energy to a system some distance away. For example,work is being done to use a satellite to collect and convert solar energy into microwaves that will then be beamed to earth to help contribute to US energy production. From the other direction, energy generated on earth could be transmitted to satellites to add to the power being generated by solar panels. Along those lines, beamed energy propulsion has also been proposed as a method of improving current rocket propulsion capabilities. A common thread in these technologies is the conversion of electromagnetic energy into a mechanically useful form. The overall goal of this work is to determine the viability of a microwave heat exchanger, where microwave energy is harnessed as the power source by an absorbing material through which a coolant is heated, through computational and mathematical modeling.

Prof. B.S. Tilley, Prof. V.V. Yakovlev (Mathematical Sciences, WPI)

 This project is focused on the physics of  interactions of ceramic materials and the millimeter wave electromagnetic field. 3D temperature fields are simulated by numerically solving corresponding electromagnetic-thermal coupled problem. Optimization of the material parameters is achieved with respect to efficiency of conversion of electromagnetic energy into heat and uniformity of the induced temperature field.

Prof. Vadim V. Yakovlev (Mathematical Sciences, WPI)

This project focuses on the use of robust dimensionality reduction techniques in cyber defense applications.   The algorithms of interest include Principal Component Analysis (PCA), Robust PCA, and Robust Deep Autoencoders.  In particular, we study streaming cyber data of various sorts, including Domain Name System data, to detect anomalies, misconfigurations, and attacks

Prof. Randy Paffenroth (Mathematical Sciences, WPI)

The ARMED project is developing techniques to protect computer systems from distributed denial of service (DDOS) attacks that use non-volumetric attack modalities. These attacks leverage flaws in network protocol stacks to exhaust systems resources using innocuous messages and normal levels of network traffic.  Much of the measured data in unlabeled, so unsupervised and semi-supervised machine learning techniques form the focus of the research.

Prof. Randy Paffenroth (Mathematical Sciences, WPI)

This project revolves around the use of machine learning to improve the performance of nanomaterial production processes.   Such processes involved numerous parameters that control the quality of the constructed materials, and we leverage machine learning to uncover important process parameters as well as improve the underlying manufacturing processes.

Prof. Randy Paffenroth (Mathematical Sciences, WPI)

Lightweight portable chemical sensors are an important component of protecting people in dangerous chemical environment. In this project we combine multiplexed chemical sensors with advanced machine learning and signal processing to maximize detection accuracy and minimize false alarms in complicated chemical environments.

Prof. B.S. Tilley, Prof. V.V. Yakovlev (Mathematical Sciences, WPI)