Error detection/correction and fault detection/recovery – Pulse or data error handling – Digital data error correction
Reexamination Certificate
1999-07-30
2002-03-19
Chung, Phung M. (Department: 2133)
Error detection/correction and fault detection/recovery
Pulse or data error handling
Digital data error correction
C714S755000, C714S780000
Reexamination Certificate
active
06360345
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a decoding method of turbo codes used in a digital communication system and more particularly to a decoding method of turbo codes and a device for the same, for activating related decoders in parallel, optimally weighting on the components of log-likelihood ratio generated from respective constituent decoders, and combining weighted log-likelihood ratios to obtain diversity gain, thereby reducing a bit error rate and decreasing the number of iterations.
2. Description of Related Art
Generally, as a channel error occurring in a digital communication system, there are noise and fading which is a phenomenon of signal distortion caused by random characteristics of transmission materials. To cope with this error factor and increase reliability, an error correction method is provided.
A representative commercialized method among the error correction methods used for digital communication systems is a method using a convolutional encoder and a Viterbi decoder. As users' demand for multimedia information, such as video and data, increases recently, an error correction method accomplishing a much better bit error rate is required. Among various techniques for enhancing error correction capability, a turbo code has been provided as a novel technique and now is in the center of attention.
According to the turbo code system, data sequence and interleaved data sequence are respectively encoded for transmission, and reliability obtained through respective constituent decoders is communicated during the decoding, thereby enhancing reliability.
With reference to the accompanying drawings, transmitter and receiver of turbo code as described above will now be reviewed.
FIGS. 1
a
and
1
b
are block diagrams for respectively showing conventional turbo encoder and decoder for code rate 1/3.
As shown in
FIG. 1
a,
the encoder comprises: interleaver
10
for scrambling the data sequence (d) according to a prescribed rule to provide outputs; first convolutional encoder
11
for convolutionally encoding the data sequence (d) to provide outputs; second convolutional encoder
12
for convolutionally encoding the interleaved sequence of the data sequence (d) to provide outputs.
In
FIG. 1
a,
a reference character, x
0
, is the same as the data sequence (d), x
1
is the sequence that the data sequence (d) has been encoded and x
2
is the sequence that the data sequence (d) has been interleaved and then encoded.
With reference to
FIG. 1
b,
the structure of the receiving party corresponding to the transmitting party having such structure as described above will now be described. The sequences y
0
, y
1
and y
2
are the decoder input sequences corresponding to the sequences x
0
, x
1
and x
2
respectively. The decoder comprises: first constituent decoder
21
for receiving and decoding based on the sequences y
0
, y
1
and the a priori information from the second constituent decoder
22
to output a log-likelihood ratio and provide extrinsic information to the second constituent decoder
22
; interleaver
20
for receiving and scrambling the sequence y
0
and the extrinsic information from the first constituent decoder
21
according to the rule as in the interleaver
10
to provide outputs; second constituent decoder
22
for receiving and decoding based on the interleaved sequence of y
0
, the sequence y
2
and the a priori information from the first constituent decoder
21
to provide outputs; and deinterleaver
23
for receiving data outputs from the second constituent decoder
22
, recovering the original order of sequences to output a log-likelihood ratio and provide ethnic information to the first constituent decoder
21
as a priori information.
Before undertaking description on the operation of the conventional turbo decoder having such structure, it may be advantageous to set forth definitions of certain terms for the purpose of helping to understand the decoding of turbo code. The terms to be defined are “log-likelihood ratio” and components of the log-likelihood ratio, “extrinsic information”, “a priori information”, and “channel value”.
The log-likelihood ratio, (d
k
) means the log value of the ratio of the a posteriori probabilities of the d
k
. It can be estimated that d
k
is “1” if the log-likelihood ratio is a positive value and d
k
is “0” if the log-likelihood ratio is a negative value. As the magnitude of the log-likelihood ratio gets larger, the reliability of this estimation gets higher. The following formula is the definition of
1
(d
k
), the log-likelihood ratio of the first constituent decoder
21
.
[Formula 1]
wherein, “observations” are the input sequences of the first constituent decoder
21
, namely, the sequences y
0
, Y
1
and the a priori information from tile second constituent decoder
22
. (Similarly,
2
(d
k
) is defined as the log-likelihood ratio of the second constituent decoder
22
where “observations” are the input sequences of the second constituent decoder
22
, namely, the interleaved sequences of y
0
, the sequence y
2
and the a priori information from the first constituent decoder
21
.) Such log-likelihood ratio can be divided into the following three components:
[Formula 2]
wherein, the second term in the right side is the a priori information from the second constituent decoder
22
, the third term is the channel value of the d
k
and the first term is the extrinsic information obtained from the present constituent decoding and provided to the second constituent decoder
22
as a priori information for the next constituent decoding.
Therefore, a priori information means the data transferred from the previous constituent decoding and used for obtaining the log-likelihood ratio of the present constituent decoding, and extrinsic information means the data derived from the log-likelihood ratio obtained by the present constituent decoding and used as a priori information for the next constituent decoding.
Based upon the terms defined above, the conventional decoding method will now be described. Primarily, the first constituent decoder
21
computes a log-likelihood ratio using the sequences y
0
and y
1
, and provides the extrinsic information that is one of components of the computed log-likelihood ratio to the second constituent decoder
22
as a priori information. The second constituent decoder
22
computes a log-likelihood ratio using the interleaved sequence of y
0
, the sequence Y
2
and the a priori information from the first constituent decoder
21
, and provides the extrinsic information that is one of components of the computed log-likelihood ratio to the first constituent decoder
21
as a priori information.
According to the system described above, the sequential path of first and second constituent decodings is defined as 1 iteration. After the time when the number of iterations, p, exceeds 1, the first constituent decoder
21
uses the extrinsic information from the second constituent decoder
22
as a priori information when computing the log-likelihood ratio.
After the iterations are performed a predetermined number of times, the hard decision is made on the latest log-likelihood ratio against a threshold “0” to estimate the d
k
. As the number of iterations increases, the bit error rate gets smaller but the degree of the improvement of the bit error rate gradually decreases. Since increasing the number of iterations results in the increase of decoding time, the number of iterations cannot be unlimitedly increased especially in real-time communication system.
According to the conventional decoding method, when a code rate is 1
, the required number of constituent decoders is n−1 and the number of log-likelihood ratios obtained at a certain decoding time is 1.
SUMMARY OF THE INVENTION
Accordingly, the present invention is directed to a decoding method of turbo codes and a device for the same that substantially obviates one or more of the limitations and disadvantages of the related art.
An objective of the present i
Kim Sang Wu
Park Sung-Joon
Bachman & LaPointe P.C.
Chung Phung M.
Korea Advanced Institute of Science and Technology
LandOfFree
Decoding method of turbo codes using a weighted parallel... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Decoding method of turbo codes using a weighted parallel..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Decoding method of turbo codes using a weighted parallel... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2845034