Optimum phase acquisition in charge-pump PLL and least squares based frequency estimation for wireless communication systems

Abstract

In this thesis, we considered two levels of synchronization in digital communication systems. One is a symbol timing recovery, which may be called a clock recovery since the symbol timing, i.e. sampling clock, should be extracted from the incoming data whose phase is usually disturbed with a random phase jitter due to noise. The other is a carrier frequency recovery where the residual frequency offset due to limited oscillator precision and/or the Doppler effect may degrade the performance of a receiver. In case of clock recovery, a new optimum phase-acquisition algorithm controlling the loop gain of a charge-pump PLL (CP-PLL) in the sense of the minimum mean-square error (MMSE) criterion is proposed. By using the algorithm, an optimum gear-shifting sequence is obtained with a zero-phase start (ZPS) assumption. It is shown that this gear-shifting sequence is independent of the variance of the input phase jitter. A procedure for applying this sequence to the design of CP-PLL circuits is described. Both behavioral simulation and HSPICE circuit-level simulation demonstrate that the proposed design leads to an efficient CP-PLL having both fast acquisition and significant jitter reduction characteristics. These methods can be used for clock recovery applications such as data communication receivers, disk drive read/write channels, and local area networks, as well as for other applications requiring very short initial preamble periods such as a fast frequency hopping system. For a carrier frequency recovery, a data-aided frequency estimator for frequency-selective channel is proposed. First, an estimator is developed on a least squares criterion, so called the LS-estimator. Next, the estimator is extended to the eigenvalue approach estimator which shows better performance than that of the LS-estimator. Simulation result verifies that the estimate is unbiased and almost achieves the Cram$é$r-Rao lower bound (CRLB). A recursive least squares (RLS) type estimat...