Research On Key Techniques In Blind Demodulation For Non-Cooperative Digital Communication Systems

During the improving modern communication techniques and communication systems, more and more requirements that connecting the systems will not interfere with the regular ones appear in the fields of point-to-multipoint networks, radio management and surveillance, malfunction detection in communication systems, and in the fields of communication reconnaissance and counterwork. This form of unauthorized access is often called patterns of non-cooperative communications.Conventional blind demodulators often base on the particular channels or communication systems, and they rarely make a comprehensive discussion on the blind demodulation techniques. For the case of indeterminacy in non-cooperative communication mode, rigorous requirements both on the demodulator architecture and on the key techniques are presented, which limit the application fields of the conventional blind demodulators. Especially, when a low signal-to noise, time-varying and deep fading multipath channel is presented, the conventional ones can hardly get satisfactory demodulation performance. This dissertation mainly focuses on the signal demodulation in a blind way under the patterns of non-cooperative communications. Some key techniques as symbol rate estimation, carrier recovery, timing recovery, modulation classification, blind equalization and also the architecture and implementation of the joint blind demodulator are studied and discussed in depth.As the major methodology for the study of non-cooperative communication systems, a brief description of the blind signal processing is made in this article for this special communication mode. Techniques of the blind signal processing are proved to be the basis way for blind signal parameters estimation, channel compensation and sequence estimation. After taking a review of the blind demodulator architecture and its key techniques, the dissertation makes a further study and particular discussion from the aspects of symbol rate estimation, carrier frequency offset estimation, timing phase estimation, modulation classification, linear blind equalization, joint blind demodulation and reduce complexity joint blind demodulation.Signal modulation parameters are the basis and crucial to a successful blind demodulation. For the purpose of extracting symbol rate information from the received signal waveform, the dissertation utilizes a kind of wavelet transform based algorithm. The classical methods usually depict the signal and system model in an ideal or simple way, which is unsuitable for the real environments. To overcome this situation, another wavelet transform based algorithm is proposed for communication channels with intersymbol interference. The choice of wavelet functions in the multipath channel model is studied from the singularity and vanishing moment viewpoint. Mean while, following the normalized spectrum maximum criteria, an optimal scale searching method is derived to solve the important factors in wavelet transform. Actually, theoretical analysis and simulation results demonstrate that, the proposed algorithm outperforms the classical Haar wavelet based method with lower signal-to-noise estimating threshold and fewer input samples requirements as well as the much better steadiness in frequency selected fading channels.Another important technical parameter is the modulation scheme, which impacts the blind modulation all the time. Conventional modulation schemes have been widely investigated and many modulation classification methods are derived mainly for flat fading channels. So few did they take into consideration the multipath channels because of the complexity. In fact, developing and intensifying the study in this situation will inevitably help the receiver improving both the architecture design and algorithm exploration. This dissertation first using the cyclic correlation matching principle to estimate the multipath channel coefficients, and then design a channel influence factor and statistical feature to help for the modulation recognition. The modulation set is limited for PSK/QAM types in this article. It is proved by simulation results that the proposed algorithm can successfully and accurately get the modulation scheme of a received signal in severe fading channels, which shows great immunes to the channel model mismatching to some degree.Both the timing synchronization and carrier synchronization are basic components of the correlated blind receiver. In this article, a timing recovery algorithm based on the band edge combination maximum criteria is studied. Under the circumstance of existing frequency offset in the received signal, there must be some extra way to guarantee the timing recovery correctly accomplished. A conjugate auto-correlation frequency offset estimating method is jointly used in the course of timing phase recursion. Performance analysis and simulation results state that, the proposed method helps to improve the convergence performance and robustness of timing recovery algorithm, fastening the convergence speed while having no effect on the estimation accuracy. Furthermore, when finishing the symbol timing, it can provide a frequency offset estimate for the later, which shall help the phase locked logic for tracking and elimination during or after blind equalization.Blind equalization is the most important and most complexity part of the blind demodulation architecture. The algorithm selections and implements of blind equalization play an important role in the blind demodulation. With the frequency selected fading channel model, this dissertation proposes an adaptive weighted multimodulus blind equalization algorithm for high-order QAM signals, implemented with linear FIR structure. Computer simulation results and analysis state that the proposed algorithm have better convergence performance over the conventional impliedly using high-order statistics methods, for it can adequately exploit the extra constellation information to make the modulus a better matching of the corresponding signal constellation.It is well-known that linear blind equalization algorithms are nearly ineffective under the time-varying, multipath channel with deep fading. Accordingly, the FIR linear structure or DFE nonlinear structure can not open the signal eye diagram. The principle of per-survivor processing is introduced with the blind maximum likelihood sequence estimation in this article. Based on the PSP joint channel and data estimation, we generally propose a framework architecture for blind demodulation. It jointly puts all the data-aided parameter estimations into the Veterbi algorithm, such as timing phase, frequency offset and initial phase, and also the channel coefficients. Theoretically, it can get asymptotically optima! estimation performance in spite of the complexity both in implementation and control. A simplified realization architecture is also given in this article and the symbol rate estimation, joint timing and frequency offset estimation, blind channel estimation, modulation classification are all involved. Simulation results demonstrate that the PSP-based blind demodulator has the ability to fit the time-varying, multipath channel with deep fading, which shows extensive application prospects.In order to decrease the computational complexity of the PSP-based joint blind demodulator, a reduced complexity blind maximum likelihood sequence estimation method is alternative. In this article, an adaptive reduced complexity blind MLSE algorithm is proposed after a synoptic description of reduced state MLSE. The proposed algorithm exploits the minimum Euclidean distance in the trellis as a threshold to select and keep some most likelihood survivors while discard the others. In this way, the states that participate in the trellis searching as well as the parameter estimators are fewer than PSP-MLSE. Theoretically, the proposed algorithm has the natural logarithm formal approximation to PSP-MLSE, but its complexity is enormously reduced. It helps improving more practicality of the PSP-based joint blind demodulator.