Optimized Automatic Frequency Control

Abstract

Detection and correction of the carrier frequency error, known as frequency synchronization, is a vital function which must be performed in the receiver in order to help improve receiver performance and relieve the stringent accuracy requirements on an oscillator. In its feedback form, frequency synchronization is implemented by an automatic frequency control (AFC) in the receiver. A challenge in designing AFC loops is to determine a trade-off between the acquisition time and tracking jitter. In fast applications, such as those in low-earth orbit (LEO) satellites, a rapid acquisition time is required. To this end, as an option, the loop gain factor, β, is increased. However, any increase in β with the aim of achieving a faster acquisition can only be useful if the selfnoise is brought under control. Otherwise, the tracking jitter increases and the loop may subsequently lose tracking.The AFC enhancement method presented herein is based on using an adaptive filter to detect the signal level and either increase or decrease its amplitude. In other words, the loop gain factor is dynamically changed; during acquisition the gain is high to increase the speed of the loop, while during tracking the gain is reduced to minimize the tracking jitter. To determine whether the loop is in acquisition or tracking mode in order to change the filtering characteristics of the loop, the output of the smoothing loop filter, which can be a simple integrator, is constantly monitored. Ideally the difference between the current and previous values of the output is zero. However, in practice, a small value is set as a threshold for the difference. Anything over this threshold means that the AFC loop is in acquisition mode. Otherwise, the loop is in tracking mode.To investigate the efficiency of the method presented, it was incorporated in the Gardner frequency discriminator and tracking loop employed in the baseband model of a quadrature phase shift keying (QPSK) modem. Fig. 1 depicts the performance at the signal-to-noise ratio (SNR) of 10 dB. With the original loop, it is not clear when the error has been acquired. Furthermore, there is a considerable amount of tracking jitter. In the same figure, the performance of the optimized AFC is shown. The tracking jitter has noticeably mitigated. The results in Fig. 1 were then magnified over the first 200 symbols. It was observed that the optimized AFC loop had acquired the error in just under 20 symbols. Such a fast acquisition is a highly desirable feature in fast applications. The BER counting simulation results are shown in Fig. 2. It is observed that at all values of signal-to-noise ratio, the BER of the optimized loop is close to the theoretical results while the BER performance of the original loop is excessively high. Such a poor performance renders the receiver impractical. Similar simulations have been performed with the baseband model of a binary phase The extra complexity required to implement the optimization is a model to take the first difference, a model to take the absolute va...