Dissertation > Excellent graduate degree dissertation topics show

Study on Channel Estimation of OFDM Systems Based on Combined Estimation in Time and Frequency Domain

Author: LvJinHua
Tutor: ZhaoXiaoZuo
School: Jilin University
Course: Signal and Information Processing
Keywords: OFDM channel estimation Kalman Filtering Wiener Filtering Maximum-Likelihood estimation
CLC: TN919.3
Type: Master's thesis
Year: 2008
Downloads: 152
Quote: 0
Read: Download Dissertation

Abstract

1.IntroductionThe target of broad band wireless communication is to transmit data fast andreliably, while there are two tough problems in front of us, they are multipath fadingchannel and bandwidth efficiency. In Orthogonal Frequency Division Multiplexing(OFDM) system, a high-rate serial data stream is split into many low-rate parallelstreams, thus changed the frequency selective multipath fading channel into flatfading channel in frequency domain which effectively mitigates the effects ofmultipath propagation. At the same time, parallel subcarriers of the system areorthogonal and overlap each other in frequency domain, this can directly andeffectively increase the transmission efficiency. Due to the above desirableadvantages, OFDM provide a reasonable resolvent for our new generation mobilecommunication. In case of getting the benefit form OFDM systems, a seriestechnology need to be carried out. Channel estimation is one among those.Channel estimation is a kind of technology that estimates the channel impulseresponse between transmit antenna and receive antenna. For the sake of resumingthe data from the transmitter, the estimation of wireless channel is necessary at thereceiver. Besides, the frequency domination equalization of OFDM system alsorequests the operation of Channel estimation. So channel estimation is the basis ofsystem’s coherent detection, demodulation and equalization, plays an important rolein OFDM systems.OFDM channel estimation methods can be divided into three kinds. The firstone is based on pilot or training sequence, named Non-Blind Channel Estimation(NBCE). The second class is blind channel estimation method, The third class issemi-blind channel estimation method. NBCE has good capability and is easy torealize, so it’s the emphasis of this thesis. The basic principle of NBCE is make useof some known pilots in transmitter and estimate channel parameter at the receivertake use of these known pilots. According to the special structure of OFDM symbol,receiver can estimate the fading channel by using the correlation between frequencyand time domination. The estimation then can be divided into methods based onfrequency domination pilot and methods based on time domination training sequence. Method based on frequency domain pilot is widely used, time domainmethods and combination of these two methods attracts more and more attentionrecently. This thesis mainly focus on combined channel estimation.2.An improved OFDM Channel Estimation Based on Combined Estimation in Time and Frequency DomainIn this paper we carried out channel estimation method based on combinedestimation in time and frequency domain, operate on both the preamble and thepayload (including pilot). First, a Kalman filter is used for estimation on thepreamble in the time domain. This has been shown to be computationally moreefficient and to result in lower variance of the estimates compared to a frequencydomainKalman filter. Next we conduct channel frequency response estimation onthe symbols of the payload, which consists of two steps. The first step is aMaximum-Likelihood(ML) estimation, solely at the subcarriers of the interspersedpilot data. In a second step the estimates are improved by Wiener filtering. The finalstep is the optimal combination of the Kalman and Wiener Filter estimates. Theresulting estimation formulae are surprisingly simple.However, during the Kalman Flitting, the state transition matrix is consideredto be a constant matrix close to the identity matrix, while not satisfied in practice.Besides, the original values of Wiener filtering is solely at the subcarriers of theinterspersed pilot data, result in low precision. To overcome the inadequacies of theoriginal algorithm, this paper made some improvements as follows.(1) Describe the time evolution of the fading channel by an autoregressive (AR)process. Using the AR model, proposed a new modified Kalman filter (MKF) andapply it to the fading channel estimation. Estimate the parameters of the AR processby minimizing the mean-squared error (MSE). Due to packet-oriented coherentOFDM transmission, Estimation of the parameters just carry out once per block, useone OFDM symbol per estimation.(2) Develop ML estimation algorithms to estimate jointly the multipath fadingchannel and the transmitted data sequence. The algorithm works in an iterativefashion: it computes an initial estimate of the channel based on the pilot symbolsand then operates in a decision directed mode. The algorithm does not require anyprior knowledge about the channel. Exploiting the properties of OFDM systemsgives a very simple structure to realize the joint ML estimate and can efficiently improve the precision.3.ConclusionsThrough computer simulation, we can drop the following conclusions. Channelestimation methods based on frequency domain pilot can finely tracking thevarieties of channel in time domain, but only happens when the time delay iscomparatively small. Channel estimation methods based on time domain preambleperforms well when the channel changes slowly, and much worse when Dopplershift is large. While the combination of the two individually estimates is verypowerful. Meanwhile, our modified method compare favorably with the originalmethod resulting in reduced mean square estimation error and bit error rate.

Related Dissertations

CLC: > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Data communication > Data transmission technology
© 2012 www.DissertationTopic.Net Mobile