Intelligent phase detection of signals on varying SNR using machine learning algorithms

The phase detection as a part of phased-locked loop (PLL) circuit is a critical component in any telecommunication systems. With the advancement of modern-era technologies, the researchers are trying to replace the complicated hardware circuitry with the machine learning algorithms with high degree of accuracy thereby making the system more compact. This paper introduces efficient techniques for machine learning (ML) classifiers that are able to classify a time-domain input signal for the accurate detection of phase shift w.r.to a reference signal. The incoming input signal is synchronised with the reference signal by using a pseudo-random pilot sequence present in the signal itself. The proposed schemes replicate the phase detection circuit with ML based software model without compromising the efficacy in terms of the receiving system functionality. The ML model designed is able to detect the phase of an incoming signal upto 93% accuracy in Additive white Gaussian noise (AWGN) channel.