Neural Precoding for ISAC
Unsupervised neural precoding for integrated sensing and communication systems.
This project was my Bachelor’s thesis at the Signal and Information Processing Lab, Indian Institute of Science, under the supervision of Prof. Sundeep Chepuri.
We studied integrated sensing and communication systems, where a wireless transmitter must serve communication users while also illuminating sensing targets. I co-developed an unsupervised neural precoder that learns to optimize sensing and communication objectives end-to-end without labelled data.
The resulting model improved sensing target illumination while maintaining user SINR constraints, and generalized to user distributions significantly larger than those seen during training.
Topics: wireless communications, integrated sensing and communication, unsupervised learning, neural precoding, optimization.