AI-based Reconstruction of Impaired Signals (AIRIS)

Modern wide-band communication systems and RF sensors face challenges from rapidly changing multi-path radio propagation channels, interference, and jammers, causing signal degradation of receiver performance. Contributing factors to rapid variation in received signal power are carrier frequency, increase of BW of signals and systems, the advent of 5G systems that will extend well into the mm-wave bands, bursty nature of modern wideband systems, and bursty transmissions of adjacent channel signals and radar. These factors put additional strain on the automatic gain control (AGC) and present other serious challenges that form the basis of this research proposal.Our team will address the problem of signal distortion with signal reconstruction and fast/predictive AGC. We will develop signal processing architectures and techniques that leverage artificial intelligence, machine learning and deep learning to improve signal recognition and reception in the presence of clipping and strong fading. In Phase I, the team will utilize computer simulations and ML capability to model a complete OFDM communication link (e.g., LTE) with high-fidelity modeling of an RF frontend and ADC converters. In Phase II, a prototype RF platform will be developed and used to demonstrate improved signal recognition and reception under the stated requirements.