| In order to meet the demand of high speed data transmission, the HF high speed communication systems are developed. Currently, block transmission and burst mode are the most important two directions of the researches on HF high speed communication. Trainning sequence is needed in each block in the block transmission mode for the estimation of the channel coefficients and equalization, such as OFDM (Orthogonal Frequency Division Multiplexing) system and SC-FDE (single carrier-frequency domain equalization) system. The burst mode, such as system adopts 3G-ALE protocols, selects the optimal communication channel and then the transmitter sends out signals immediately after shaking hands with the receivers.Channel estimation and SNR estimation are the key techniques of communication signal demodulation at receivers, their application to subsequent equalization and decoding can dramatically raise the reliability of the system. This dissertation mainly studies the channel estimation algorithms in OFDM system, SC-FDE system and system adopts 3G-ALE protocols. And a robust blind SNR estimation method is studied based on the summarization of the SNR estimation methods in recent years. The work in this thesis relies on a large scale project undertaken by the laboratory the author works in, the main contents and contributions are summarized as follows:1. Pilot-aided channel estimation in OFDM is studied. The comb-pilots can not effectively estimate the channels with a fast varyingly frequency response, and block-pilots can not keep up with the fast time-varying channel. In order to overcome these problems, an iterative decision feedback channel estimation method is studied. This method uses the decision of the estimation of transmitted symbols to update the channel coefficient, with which the received signal is equalized. The iteration continues until the channel estimation converges. The method greatly improves the performance of OFDM system under complicated shortwave channels.2. Channel estimation in SC-FDE system under doubly selected channel is studied. An improved channel estimation method is proposed. Firstly, the pilots of two adjacent data blocks are combined to acquire a new pilot, then based on the cyclic shift properties of circular convolution in time domain, the new pilot is used to derive several groups of channel coefficients under different shift valuation, then the sparse characteristics of wireless channel are considered, and the disturbance of the noise is eliminated for the estimation of each group of the channel coefficients, and finally, the estimation results are averaged, so that the performance of the channel estimation and the frequency equalization is greatly improved.3. Channel estimation in HF high-speed burst communication system is studied. A blind channel identification and blind equalization method is proposed. Firstly, it estimates the posterior probability density of transmitted symbols with particle filtering, then the estimation of transmitted sequence is acquired, and the channel coefficients and noise variance are updated according to ML criterion, after which the the posterior probability density is re-estimated. The estimation of the channel coefficients converges after several iterations. Trainning sequences are not needed in this method, and it exhibits a performance near to the methods with known channel and noise variance.4. SNR estimation algorithms are researched. Several SNR estimation methods are studied and simulated under different conditions, and some important conclusions are derived, based on which, this dissertation studies a robust blind SNR estimation method thoroughly. Proposed a new method to estimation the correlate matrix, which meet a compromise of performance and complexity. This method can directly process IF (Intermediate Frequency) signals, and it is independent on the modulation method and roll-off coefficient. It exhibits good SNR estimation performance under shortwave channel. |
PDF Full Text Request