Non Data-Aided Carrier Tracking Techniques for Continuous-Phase Frequency-Shift Keyed Signals

Communications between ground stations and nanosats in low earth orbit (LEO) require acquisition and tracking of large Doppler frequency offsets due to the relative velocity between the transmitter and the receiver. The Doppler frequency shift varies with time, reaching its fastest rate of change as the small satellite reaches its closest approach to the ground station. For phase modulated signals, a frequency offset estimate extracted from the band edge filters is used to correct for the Doppler shift. This approach encounters difficulties when it’s used to correct for frequency offset in continuous phase frequency shift keyed (CPFSK) signals. Optimized band edge filters are designed to have a frequency response that is the derivative of the matched filter frequency response. However, CPFSK detection often requires different matched filters for different symbols, particularly when there are multiple possible magnitudes of frequency shifts to choose from in each symbol period. Having separate matched filters for different symbols complicates the effort to use optimized band edge filters as it imposes a requirement to demodulate the signal before correcting its frequency offset. Furthermore, the fact that the signal of interest is itself changing in frequency from one symbol to the next further complicates the issue. However, a modification to the classic band edge filter design can permit the detection and correction of CPFSK signals without the need to demodulate the signal. This paper presents techniques for non data-aided correction of CPFSK signals using modified matched filters in frequency locked loops (FLLs). Examples of the approaches are shown in MATLAB for varying signal-to-noise ratio (SNR), static Doppler, and dynamic Doppler.