Channel Estimation Methods For Time-varying Ultra Wideband Wireless Communication Systems

Since ultra-wideband (UWB) signal will be affected and distorted unavoidablydue to the uncertain factors of wireless channel, channel estimation is indispensable toensure the coherent accurate reception of received signal. Hence, the overallperformance of system will be influenced by channel estimation performance directly.Generally, the traditional UWB channel estimation methods are mainly studied in theenvironments of indoor static channel. However, the movement of transceiver orscatterer will lead to the time-varying fading characteristic of UWB channel, andtherefore, the research on time-varying UWB channel estimation method is moreimportant. This dissertation focuses on the time-varying channel estimation method inmulti-band orthogonal frequency-division multiplexing Ultra-wideband (MB-OFDMUWB) systems. Firstly, introduced are the UWB channel models, where two kinds oftime-varying UWB channel model are given and a theoretical basis of the modelingautoregressive model is also analyzed. Secondly, the existing estimation methods arecompared and discussed. Finally, proposed are two Kalman filtering time-varyingchannel estimation methods based on the pilots.Kalman filtering is an unbiased estimation algorithm for tracking the time-varying channel. However, the traditional Kalman filtering algorithm only makes useof the channel time-domain correlation and ignores the frequency domain correlation.Thus, the modified Kalman filter algorithm in frequency-domain is proposed to gain abetter performance based on the minimum mean-square error (MMSE) criterion, but ata cost of high computational complexity. In order to overcome these drawbacks, aKalman channel estimation based on MMSE algorithm frequency domain correctionmethod is proposed where time-varying channel attenuation factor is tracked by pilots,and Kalman filtering and frequency-domain block MMSE algorithm are also used totrack both channel time-domain correlation and frequency domain correlationsimutaneously. The simulation experiments show that the proposed method reduces the computational complexity greatly with a small amount of estimation precision loss andis helpful to practical applications.On the other hand, Kalman filtering divergence suppression is an importantproblem of channel estimation. This dissertation analyzes the cause of filteringdivergence and inhibition of the traditional methods, and in order to solve the problemof filtering divergence caused by the channel transfer coefficient estimation inaccuracy,this dissertation proposes a channel state transfer coefficient (time-varying channelattenuation factor) threshold correction channel estimation method. This method trackschannel by comb pilots, estimates the filtering initial values and parameters based onthe least squares algorithm. And then, the estimated value is further modified bysetting threshold method. Thus, the Kalman filter can effectively track thechannel. Simulation experiments show that the method is not only simple but also caneffectively improve the channel estimation performance.