Smart Sensing and Information Processing (SSIP) Lab

Our research concerns mathematical and algorithmic aspects of signal processing and data science: we design innovative sampling and sensing techniques to acquire useful data in a resource-efficient manner, while obeying application-specific constraints. PI Pal pioneered a new research direction in sensor array and multichannel signal processing, by developing new sparse array geometries constructed as unions of uniform arrays with variable spacings, and demonstrating that they provably outperform the widely used uniform arrays, due to their ability to localize “more sources than sensors”. Inspired by this idea, our research group at UC San Diego has demonstrated that it is possible to attain similar benefits in a larger class of high dimensional inverse problems, such as Sparse Bayesian Learning. Our results have direct applications in target localization and tracking (radar, sonar, mobile and autonomous vehicles), ill-posed inverse problems in imaging, next generation massive MIMO and hybrid wireless communication, neural signal processing for deciphering brain activity, and radiology. More recently, our group is using ideas from signal processing to make machine learning more sample-efficient, adaptable (to new domains at test-time) and interpretable. Along with collaborators in coding theory, we are also actively investigating new fundamental questions at the intersection of coding and modrn sensing problems, including quantum sensing and coding.

See People and Publications for more details.

We gratefully acknowledge the support of NSF, ONR, DOE, NIH, CWC, Qualcomm, and Texas Instruments.