Superimposed training for multiple antenna communications
A training-based transmission technique for multiple-antenna communication systems is disclosed herein where the transmitted symbols are a superposition of the training symbols and the data symbols.
1. A method of wireless communication comprising:
receiving information symbols to be transmitted to a receiver; receiving training symbols known to the receiver; and
arranging the information symbols and the training symbols so that the training symbols are not transmitted in a single training phase but are interspersed with the information symbols in a superimposed transmission, the superimposed transmission occuring over a noncoherent communication channel with multiple antennas being utilized for either the superimposed transmission or for reception at the receiver, the receiver using its knowledge of the training symbols and of the arrangement of training symbols within the superimposed transmission to generate an estimate of characteristics of the noncoherent communication channel having a coherent interval of T, and the superimposed transmission of the training symbols and the information symbols occupying the whole coherent interval.
2. The method of claim 1 wherein there are M T transmit antennas and wherein the superimposed transmission can be represented by a T×M T matrix X where
X=X T +X d
and where X T is a matrix representing the training symbols and X d is a matrix representing the information symbols.
3. The method of claim 1 wherein the information symbols are codewords which are a linear combination of dispersion matrices with symbols to be transmitted.
4. The method of claim 3 wherein the training symbols and the dispersion matrices are generated using simulation-based optimization with gradient estimation.
5. A transmitter for wireless communication comprising:
one or more antennas for transmitting across a communication channel to a receiver;
a symbol arranger which receives information symbols to be transmitted to the receiver and training symbols known to the receiver and superimposes the information symbols and the training symbols into a superimposed transmission, the training symbols arranged so as to be interspersed with the training symbols in the superimposed transmission rather than being arranged in a single training phase; and
an encoder which generates the information symbols in accordance with a coding arrangement, the information symbols generated by the encoder being a linear combination of dispersion matrices with symbols to be transmitted.
6. The transmitter of claim 5 wherein the communication channel has a coherent interval of T and wherein the superimposed transmission of the training symbols and the information symbols occupies the whole coherent interval.
7. The transmitter of claim 6 wherein the transmitter has M T antennas and wherein the superimposed transmission can be represented by a T×M T matrix X where
X=X T +X d
and where X T is a matrix representing the training symbols and X d is a matrix representing the information symbols.
8. A receiver for wireless communication comprising:
one or more antennas for receiving a transmission from across a communication channel, the transmission being a superposition of information symbols and training symbols;
a training symbol extractor which receives the transmission and which extracts the training symbols, the training symbols being arranged not in a single training phase but interspersed with the information symbols in the transmission; and
a channel estimator which receives the training symbols and which estimates characteristics of the communication channel based on the training symbols, and
a decoder which decodes the information symbols in accordance with a coding arrangement and operates in accordance with
arg
min
l
=
1
,
…
,
L
Y
-
ρ
(
X
T
+
X
d
,
l
)
H
~
F
where Y is the transmission received, X T is a representation of the training symbols, X d,l is a representation of a possible information symbol, and {tilde over (H)} is an estimate of characteristics of the communication channel as provided by the channel estimator.
9. The receiver of claim 8 wherein the information symbols decoded by the decoder are a linear combination of dispersion matrices with symbols to be decoded.
10. The receiver of claim 9 wherein the training symbols and the dispersion matrices are generated using simulation-based optimization with gradient estimation.