Signal Enhancement & Array Processing
Signal enhancement is a process to either restore a signal of interest or boost the relevant information embedded in the signal of interest and suppress less relevant information from the observation signals.
Today, there is almost no field of technical endeavor that is not impacted in some way by this process. Array signal processing manipulates the signals picked up by the sensors that form an array in order to estimate some specific parameters, enhance a signal of interest, or make a particular decision.
These topics lie at the heart of many fundamental applications, such as hands-free voice communications, sonar, radar, ultrasound, seismology, autonomous cars, robotics, and so on.
Array Design and Beamforming for Immersive Voice Communication
Immersive voice communication facilitates group collaborations and tele-conferencing efficiently at low costs. It is becoming an integral part of modern communication networks. A key technical challenge for an immersive voice communication system is the ability to acquire high-fidelity acoustic, audio, and speech signals while keeping sources spatial information intact so that it is possible for the remote listener to sense the acoustic environment and follow a panel of speakers and distinguish them by listening to the reproduction of the signals.
Learning-based Multimodal Signal Processing
In recent years, multimodal signal processing has attracted a growing interest in the signal processing community due to an extensive use of multiple sensors in a wide variety of fields ranging from smartphones and computers to medical and security applications. Processing of multimodal signals has many advantages since signals obtained from different sensors often contain complementary information.
A major challenge is fusion of signals that are captured in different types of sensors and often have different dimensions, units, and value ranges. The objective of this research is to develop data-driven algorithms that are based on learning structures of signals captured via multimodal sensors.
Manifold Learning for Anomaly and Target Detection
The goal of this research is developing new data-driven methods for anomaly and target detection in images. Specifically, we focus on a manifold learning approach, in both supervised and unsupervised settings.
A data-driven approach has many advantages to statistical based solutions, as the separation between the object and background is implicitly inferred from the local geometry and intrinsic parameters of the data, without relying on a model or training data. A data-driven approach also enables a robust solution which is not dependent on the imaging sensor or specific detection task and can be adapted to new applications.
Sensors Arrays for Autonomous Platforms
Autonomous unmanned platforms offer a vast range of possibilities and applications. In the aerial medium, drones are increasingly used for missions of data collection and surveillance. In this medium, drones can collect visual, electromagnetic, as well as acoustic data using several suitable sensors. In the underwater medium, acoustic waves can propagate to large distances, compared to electromagnetic waves whose propagation is very limited. Hence, acoustic sensors are necessary for underwater communication, active and passive SONAR systems, and localization in the underwater medium.