Training-Based Channel Estimation and Power Allocation for MIMO-OFDM Systems
Date Issued
2006
Date
2006
Author(s)
Chu, Tso-Jen
DOI
en-US
Abstract
This thesis addresses the problem of training–based channel estimation for Multiple-Input Multiple-Output Orthogonal Frequency Division Multiplexing (MIMO-OFDM) systems.
We present a channel estimation scheme based on training guard intervals. This way, the receiver can estimate and track the channel variations semi-blindly based on first order statistics of the received signal. There is no extra loss of information rate as opposed to training sequence based methods , no condition imposed on the channel, and no scaling ambiguity as opposed to second (or higher) order statistical or finite alphabet-based methods.
To achieve the minimum mean square error, the training sequences transmitted from multiple antennas must have impulse-like autocorrelation functions and zero cross-correlation functions. To this end, we adopt the sequences with Zero Correlation Zone (ZCZ) which possess ideal correlation windows.
For multiple-antenna systems, performance of this training scheme will be degraded drastically with the increase of the number of antennas and the delay spreads of multipath channel. We propose to append the training sequences overlapping with data symbols at the transmitter. By this method, the OFDM system can combat longer delay of multipath channel.
Moreover, the issue of power allocation between training and information sequences will also be addressed. By evaluating the performance degradation due to imperfect channel estimation, optimal training power overhead that minimizes bit error rate is found.
We present a channel estimation scheme based on training guard intervals. This way, the receiver can estimate and track the channel variations semi-blindly based on first order statistics of the received signal. There is no extra loss of information rate as opposed to training sequence based methods , no condition imposed on the channel, and no scaling ambiguity as opposed to second (or higher) order statistical or finite alphabet-based methods.
To achieve the minimum mean square error, the training sequences transmitted from multiple antennas must have impulse-like autocorrelation functions and zero cross-correlation functions. To this end, we adopt the sequences with Zero Correlation Zone (ZCZ) which possess ideal correlation windows.
For multiple-antenna systems, performance of this training scheme will be degraded drastically with the increase of the number of antennas and the delay spreads of multipath channel. We propose to append the training sequences overlapping with data symbols at the transmitter. By this method, the OFDM system can combat longer delay of multipath channel.
Moreover, the issue of power allocation between training and information sequences will also be addressed. By evaluating the performance degradation due to imperfect channel estimation, optimal training power overhead that minimizes bit error rate is found.
Subjects
多輸入多輸出
正交頻分多工
通道估測
訓練序列
功率分配
MIMO
OFDM
channel estimation
training sequence
power allocation
Type
thesis
File(s)![Thumbnail Image]()
Loading...
Name
ntu-95-R93942093-1.pdf
Size
23.31 KB
Format
Adobe PDF
Checksum
(MD5):f7915c1f3ac68672ba53462fe9800341