Over-the-Air Deep Learning Based Radio Signal Classification

We conduct an in depth study on the performance of deep learning based radio\nsignal classification for radio communications signals. We consider a rigorous\nbaseline method using higher order moments and strong boosted gradient tree\nclassification and compare performance between the two approaches across a\nrange of configurations and channel impairments. We consider the effects of\ncarrier frequency offset, symbol rate, and multi-path fading in simulation and\nconduct over-the-air measurement of radio classification performance in the lab\nusing software radios and compare performance and training strategies for both.\nFinally we conclude with a discussion of remaining problems, and design\nconsiderations for using such techniques.\n