coding gains of the proposed scheme over the uncoded. QPSK modulation system either in terms of decoded block-error rates or in terms of decoded bit-error.
N92-22018 Bandwidth Satellite
Efficient Coding Communications*
for
Shu Lin University of Hawaii Honolulu, Hawaii 96822 Daniel J. Costello, Jr. University of Notre Dame Notre Dame, Indiana 46556 Warner H. Miller James C. Morakis William B. Poland, Jr. Goddard Space Flight Center Greenbelt, Maryland 20771
RESEARCH
The trol
purpose
coding
reliability coding
*
of this
scheme by
PURPOSE
research
to achieve
using
coded
is to devise
large
coding
modulation
an error gain
with
reduced
complexity.
Supported
by NASA
Grants
PRECEDING
NAG 5-931
PAGE
BLANK 269
NOT
and NAG 5-557.
FILMED
and
conhigh de-
CODED
•
To achieve bility,
•
In
a 3 to 5 dB
the
fact,
decoding
to
complexity
•
MODULATION
complexity
is quite
simple,
no
(BCM)
(TCM)
of the
gain,
moderate
relia-
modest.
the
matter
whether
or block
coded
decoding trellis modu-
is used.
to achieve complexity
and
is quite
coding
lation
mentation
gain
a 3 dB
modulation
coding
coding
achieve
coded
However,
ALONE
coding
gains
increases
exceeding
drastically,
decoder
becomes
unpractical.
270
5 dB, and
very
the
expensive
the
de-
impleand
A BASIC
•
How ity
can by
we achieve
using
complexity
coded
QUESTION
large
coding
modulation
?
271
gains with
and
high
reduced
reliabildecoding
AN
•
Use
coded
nated
•
A
in
cascaded)
short
(trellis
•
modulation
(or
good
ANSWER
or
bandwidth
block)
efficient
is used
powerful
outer
code.
With
properly
chosen
nated
coded
modulation
large
coding
as the
Reed-Solomon
gains
width
efficiency
This
combination coding
of three
worlds,
efficiency
and
concate-
outer
scheme
not
decoding
and
complexity.
272
relaas the
a concate-
only
can
with
good
achieve band-
implemented
and
of achieving coding
and
codes,
modulation
a way
code
is used
can be practically
offers
reliability
code
code
reliability
of coded really
inner
and
high
also
modulation
(RS)
inner
and
but
with
coding.
tively
nated
conjunction
gain,
concatethe
best
bandwidth
PROPOSED
•
A concatenated
SCHEME
( or cascaded
) coded
modulation
scheme.
For NASA
high-speed
large data file transfer
satellite where
communications
for
very high reliability
is
required.
The
outer
symbols
code from
C2 is an
GF
(n2,
The outer
code
•
The inner
code is a bandwidth
PSK
RS
code
with
(2b).
•
M-ary
k2)
is interleaved
code of length
to a depth
efficient r_l
of rn.
block
and dimension
kl = tab. •
Under
the same
gated
concatenated
research
project,
coding
273
we have
with TCM
inner
investicodes.
THE CONCATENATED
• The
with
The
CODED
outer
which
OVERALL
code
C,
is an
is designed 0 _< t,
inner
MODULATION
(n,,
to
correct
< [(n_-k_
)/2J.
code
C,
k_ ) RS t,
is a 2'-PSK
SCHEME
code
or fewer
code
over
GF(T),
symbol
of length
errors
n,
and
dimension
iffil
where
k, = mb.
•
The
outer
code
•
The
encoding
C_ is interleaved
to a depth
of m.
consists
of two stages,
the
outer
and
inner
consists
of two
the
inner
and
outer
encodings.
•
The
decoding
stages,
decodings.
•
When block
•
In the
the
receiver
is erased
event
transmission
and
fails
to decode
the
receiver
of an erasure, or accept
a received raises
we could
the
erroneous
274
block,
the
a flag.
either
request
blockwith
a realarm.
r
k2
_J
b-bit
1st •
byte
•
•
Section
m-th Section
Ist
k=-th
n2 -th
Segment
Segment
Segment
Figure
I
A Segment-Array
275
ERROR OF •
Let
THE
Po,
P.,
decoding,
•
P,,
be
an erasure received
Lower
bound
• Let
OVERALL
and
entire
have
PERFORMANCE
been t5
on
block
for
probabilities
of
an incorrect
a correct
decoding
for
an
respectively.
Pc and
derived
denote
the
and
code
SCHEME
upper
bounds
an AWGN
on
P,,
and
P°,
channel.
a lower
bound
on P,.
is an upper
bound
on the
V
• Then
l-P,
of a decoding • Let •
/_°,
The
failure
denote
a decoding
an upper
performance
modulation
and
bound
of the
scheme
total
probability
error.
on
proposed
P,,. concatenated
is measured
by
the
coded
pair,
/_,,
and
1-Po. We
can
scheme either terms
in
compute
the
over
uncoded
terms
of decoded
the of
coding
gains
QPSK
decoded
bit-error
the
proposed
modulation
block-error rates.
276
of
rates
system or
in
•
For
data
used
as
file the
transfer,
the
measure
block-error
of the
error
rates
should
performance
be
of
the
scheme.
• There
are
two
types
of bit-error
rates,
denoted
P.,
and
P,2.
•
P_,
is computed
1=',, using
the
based
on
the
block
P_, posed
• P,,
is a measure scheme
using
the
when
failure
performance
retransmission
based and
on the
of the
pro-
is allowed.
total
a decoding
probability error
1-Pc
of a code
of
block
approximation,
P,2
•
). P.,
of bit-error
is computed
a decoding
probability
approximation,
P,, = •
error
P,,
is used
as the
the
scheme
when
= (d,/2n,
measure
).
(1 -
P, ).
of bit-error
retransmission
lowed.
277
is not
performance available
of or
al-
TWO CONCATENATED
SPECIFIC
CODED
MODULATION
SCHEME-
• The
outer
code
code
over
GF(2')
used
to correct
• The
inner
16, D[C,
code
I
C2 is the NASA which
standard
has minimum errors.
C,
code
is an 8-PSK
(255,223)
distance
up to 16 symbol
]= 4, R[C,
SCHEMES
with
RS
33. It is
n, = 8, k, =
]= 1 and _/[C, ]= 3 dB (over
uncoded
QPSK).
• The
outer
• The
overall
code
is interleaved
effective
to a depth
rate of the
scheme
of m = 2.
is
R., = (k, /n, ). R[C,] = 0.STS.
• The
inner
be decoded
code with
has
a 4-state
a soft-decision
278
trellis
structure
Viterbi
decoder.
and
can
ERROR
PERFORMANCE
• With
SNR
= 9 dB/symbol
• With
SNR
= 10 dB/symbol
P,,
and
1-Pc
is very
small.
(6.57 dB/infor,
(5.57
dB/infor,