Transforming Wireless Communication With Deep Learning

ECE William Lincoln Smith Professor Tommaso Melodia and Assistant Professor Francesco Restuccia were awarded a patent for “Deep learning-based polymorphic platform.”

Abstract Source: USPTO

A polymorphic platform for wireless communication systems is provided that employs trained classification techniques to determine physical layer parameters from a transmitter at a receiver. The system includes a learning module to determine transmitted physical layer parameters of the signal using a trained classification module, such as a deep learning neural network. The trained classification module receives I/Q input samples from receiver circuitry and processes the I/Q input samples to determine transmitted physical layer parameters from the transmitter. The system includes a polymorphic processing unit that demodulates data from the signal based on the determined transmitted parameters.

Related Faculty: Tommaso Melodia, Francesco Restuccia

Related Departments:Electrical & Computer Engineering