Study On Channel Estimation Algorithm In MB-OFDM UWB System

Regarding future ideal wireless communication system, there are five demands needed to solve, which are great data quantity, high data rate, far signal distance, quicker communication speed, bigger network number of users. However for the actual communications system, it is impossible to realize simultaneously the above five requests, we can only come to a compromise. The initial communications system makes every effort to lengthen the signal distance, but the present wireless communication's trend of development is to take sacrifices the signal distance as a price, greatly improve the performance of the other four aspects, and feed the request of communication distance through the network covering. The ultra-wide band communication is a very competitive technology in the high speed short distance wireless communication domain.The channel estimation is the key technologies for signal detection, so the study on it has the vital significance. There are usually three kinds of channel estimation methods: pilot based channel estimation, blind channel estimation and semi-blind channel estimation method. For semi-blind channel estimation method, it is using the information from blind channel estimation algorithm and known sampling symbols to finish channel estimation. It solves the problems of spectrum waste by channel estimation based on pilot symbols or training sequence and high complication by blind channel estimation methods. So the semi-blind channel estimation algorithm is regarded to be a promising way for channel estimation.This article proposed a semi-blind channel estimation algorithm used in the MB-OFDM ultra-wide band system, it is called Interactive Multi-Model based on Kalman filter algorithm, the simple form is IMM-Kalman algorithm. Considering the possibility of estimating the frequency selectivity time-variable channel, this method is proposed. Usually it is considered that the ultra-wide band system channel environment is quasi-static. However, when the transmitter and the receiver are under the moving condition, the Doppler shift would be generated, and would cause the frequency selectivity decline, therefore we propose a new channel estimation method which will have good performance under both the static channel and the moving state. This method used the interactive multi-model algorithm, which have the best performance-to-price ratio usually used in the target tracking domain for mixed system estimation.IMM-based Kalman filtering algorithm is proposed to estimate and track the time-varying frequency selective fading channel. A simple IMM-based filter for channel tracking consists of two models namely the Static Model Filter (SMF) and the Dynamic Model Filter (DMF). The SMF provides a better estimation of the channel parameters when the receiver is static and the DMF gives a better estimation of the channel when the receiver is in moving state. The estimated channel parameters are combined based on model probabilities. The model probabilities are updated each time based on residuals.In order to simplify the computation, the first-order AR process is used to model the channel. To reduce the dimension of the Kalman filter, we futher propose a structure using a low-dimensional Kalman-filter estimator for each subchannel. Channel frequency response gain, is called the AR process coefficient regarding the dynamic model, may be obtained by the Jakes model channel coefficient correlation function, because the channel is the quasi-static or slow time-varying channel, so we could suppose the parameter is not time-varying parameter in a frame, which means it is constant. This kind of supposition for the ultra-wide band channel characteristic is reasonable, and may simplify the computation complexity greatly. Regarding the static model, state transition matrix may be assumed as an Identity Matrix.Besides, this article realized semi-blind estimation to the ultra-wide band channel, because the receiver only needs the known training sequence or the pilot frequency information in the training stage for the initial channel to estimate, but after that the system switches to decision-feedback mode, do not need any of the known information from the transmitter, therefore it made a compromise between the data transmission and the convergence rate and gives better performance.From the simulation with the carrier frequency on 3.432GHz, it is shown that the proposed IMM Kalman channel estimation method could make the channel model more closer to the real channel, due to the real-time update of the model probability, and can track the time and frequency variety,better estimate the channel parameter either in static or moving state, which enable this algorithm to have certain auto-adaptive ability. Simulation graph shows that IMM algorithm based on Kalman filter can give a better estimation and track frequency selective time-variable channel. When compared to the method with just one Kalman filter IMM Kalman method has demonstrated it has the higher performance.