Researches On The Application Of Signal's Cyclostationarity In OFDM Systems

This thesis is devoted to researches on the application of signal's cyclostationarity in OFDM systems. The objective of the research is how to estimate channel, transmission delay and frequency offset blindly based on cyclostationarity.OFDM is widely regarded as an attractive technique for the next generation communication systems. Synchronization and channel estimation are two key techniques for OFDM systems to ensure its good performance. In an appropriate transmission condition, blind synchronization and channel estimation will avoid degrading the information transmission efficiency. It has been found that many communication signals are cyclostationary in nature. Signal processing based on cyclostationarity are always blind. Furthermore, for a cyclostationary signal, there is a possible spectrum copy on each cyclic frequency. When these copies are used in the signal processing, the information of a signal is fully utilized, which will make signal processing easier or improve the performance.Theory of Cyclostationary Signal (CS) is introduced first in this thesis, including its definition, cyclic autocorrelation and cyclic spectrum, the way to produce a cyclostationary signal and its applications.Based on the theory of CS, the relationship of a CS and a stationary signal is discussed. And then the characteristic of a sampled CS is investigated. Furthermore, the cyclostationarity property of an OFDM signal is proved, which gives possibility to design cyclostationarity-based blind algorithms for channel estimation or delay and frequency offset estimation for OFDM systems.Transmission delay and frequency offset estimation is a key for system synchronization. In this thesis, flat-fading channel model and frequency selective fading channel model are both used to design a blind method for delay and frequency offset estimation in OFDM systems. It has been proved that by oversampling the received signal and choosing a suitable oversampling factor, a larger frequency offset and a finer delay can be estimated. This algorithm only needs the power and transmission delay of each channel path, while for a method which is based on a linear invariant channel model; it needs all the channel information. Meanwhile, in a system where pilots are needed, it has been found that by laying pilots reasonably, the cyclostationarity property of the received signal is hold, so that the proposed algorithm is still applicable with a little modification.To achieve equalization and demodulation, channel gains need to be estimated. An earlier channel estimation method based on solving linear equations is much operation consuming, so an improved method is introduced first in this thesis, which reduces mathematic operation greatly. Because methods mentioned above are sensitive to the error of the cyclic autocorrelation function estimation, a new method is proposed further, in which the order of the channel is estimated, and then a channel is achieved by estimating its zeros. It is proved by simulation and theoretical analysis that this new method can get a better performance and has a tolerance to the error of channel order estimation to some extent.Transmitted signals are affected by frequency offset and channel together. So an algorithm which can estimate frequency offset and channel jointly is necessary for an OFDM system. By combined with subspace technique, joint estimation of the channel and frequency offset can be done by a cyclostationarity-based method. Estimation is achieved in three steps. First, the channel containing a frequency offset is estimated by means of subspace technique, and then the result is used by a cyclostationarity-based method to obtain the frequency offset. Finally, the estimated channel is achieved by removing the frequency-offset component from the result of the first step. By using iterative method, more accurate channel estimation and...