In modern sensing systems, it is often impossible to transmit all the raw data captured by edge sensors (antennas, cameras, microphones, and more), due to bandwidth limitations, energy constraints, and sometimes privacy considerations. In such settings, the classical paradigm can be challenged: instead of compressing data to optimize signal reconstruction, compression can be tailored specifically to the task at hand—for example, detecting a desired signal or estimating an unknown parameter.

In this project, the practical scientific challenge is to jointly design encoding and estimation schemes under constraints imposed by the limited resources of the compression algorithm. On the theoretical side, we seek to better understand what is and is not achievable—namely, the fundamental performance limits—by establishing a close connection between estimation theory, signal processing, statistics, and information theory.