Enhancing Spectral Efficiency with Over-the-Air Multiuser-MIMO Enabled NOFS Signals

We innovate a novel over-the-air signal waveform termed non-orthogonal frequency shaping (NOFS), aiming for artificial intelligence air interface for future 6G communications. By reshaping sub-carriers using irregular Sine(irSinc) patterns, the proposed NOFS waveform effectively reduces bandwidth and exhibits asymmetric shaping characteristics. The unique irSinc pattern empowers NOFS to utilize two-dimensional modulation schemes beyond the Mazo limit using simple detection techniques, where previously only maximum likelihood detection methods were possible. Moreover, NOFS demonstrates robustness in spectrally efficient communications even with high out-of-band power leakage. Additionally, the explainable nature of the neural network used in the NOFS waveform design provides insights into the signal patterns, enhancing transparency and trust in the system. Finally, an over-the-air signal transmission is operated in hardware validating the effectiveness and robustness of NOFS waveforms in real-world.