Study On Equalization Algorithm With Improved Convergence For Single-Carrier Terrestrial Digital Television Broadcasting

Adaptive Equalization is one of the key technologies of single-carrier digital television terrestrial broadcasting transmission system. It occupies the most of the computation used to implement the receiver, and determines the maximum quality that the receiver can arrive at. This paper focuses on the study of the equalizer with high performance and low computational complexity, for the purpose of the tradeoff between performance and computational complexity.In this paper, first, the definition and pertinent features of DTV, category of DTV transmission system are given. In the following, the history of development of the existing DTV transmission systems is presented. Then, we briefly introduce the frame structure, fundamentals and some key technologies taking ATSC and DVB-T as the examples, and compare the two kinds of systems. The effect of terrestrial channels on the transmitted signal is described. In the terrestrial DTV transmission, small-scale fading has more complicated effect on the performance of the receivers. Depending on the relation of the parameters of transmitted signal and channels, we give the categories of the channels of small-scale fading. Afterward, wireless communication systems are modeled, and then a set of DTV terrestrial channels which represent different kinds of channels are given to reflect the performance of the equalizer.In the following, we review the equalization theory, and analyze the characteristics and computational complexity of equalizers in the context of DTV transmission. We bring forward passband adaptive DFE, and get a clear understanding by the geometric explanation of Wiener filtering. For the well tradeoff between performance and computational complexity, fractionally-spaced DFE adapted by LMS is adopted as the basic whose operation is unified on symbol rate. This equivalent structure helps to understand how LMS FS-DFE works, following which, we give the guideline for engineering to choose the parameters. By detailed analysis of simulation in depth, we bring out the robust adaptive equalizer is improved by two aspects that data is feedback into the feedback filter of DFE as accurate as possible, and the error signal is generated as exact as possible. Furthermore, the guideline to choose parameters for control is presented. Simulation shows that the improved robust adaptive equalizer can smoothly converge to work converge to work stably.Then, we give the fundamentals of adaptive equalizer which include scheme for implementation, the process for development and the fixed-point theory for LMS theory. In the following, we give the implementations that one is how algorithm is mapped into the hardware architecture, the other is the fixed-point design of the equalizer. We assign appropriate the wordwidth to the signals according to significance and sensitivity to fixed-point, give the process to choose the wordwidth, and determine them by simulation. Also, the accumulation of quantization noise in the adaptive loop will degrade the performance of the equalizer, and even make the equalizer crash in some extremely severe cases. Consequently, we design a simply symmetrical leaky-LMS algorithm to prevent the tap coefficients from drifting to infinite. Simulation shows that this algorithm can effectively reduce the loss of the performance for the fixed-point design.Finally, for further improving the performance, we bring forward robust subband by combining the subband adaptive filtering with the robust algorithm. The pressure for the equalizer will decrease if the equalizer converges well during the training period. For high computational complexity of the equalizer, we must conform to the constraints that the adaptive algorithm converges faster with no more incensement of computation. We first review corresponding concepts and theory for adaptive subband filtering, which is basic of the following. Then, we apply subband adaptive filtering to the aforementioned robust equalizer. Subband filtering is applied to the feedforward-filter, and close-loop subband filtering is adopted. According to the equivalent FS-DFE, we get the subband adaptive FS-DFE adapted by LMS.Simulation shows that the new equalizer converges better and get into stable work faster.