Viterbi’s Impact on the Exploration

of the Solar System

Proof of Optimality of
Orthogonal Codes

First Appearance of
Eb
> ln 2
N0
 (b) p’l
have made use of the fact that go(p) is concave.
Fig. 4. Graphical construction of EL(R) from I%(P).
m property b) we have lim,,,, &(p)/p =
C. Thus we obtain Error Bounds for Convolutional Codes
whereand an Asymptotically Optimum
1
ponent E,(R) in the lower bound is a positive Decoding Algorithm
decreasing continuous function of R for all
p’ is the solution ANDREWto J. theVITERBI,
equation SENIORE,,(c)
MEMBER, = E,,IEEE and E, is
hical construction of the exponent-rate curve given by (17b).
ot of the function E,,(p) is shown in Fig. 4. We probability of error in decoding an optimal con- Increasing
Ahstraci-The
volutional code transmitted over a memoryless channel is bounded
with rate. The upper bound is obtained for a specific
probabilistic nonsequential decoding algorithm which is shown to be
her consideration of the properties fromofabove(16) and below as aIV.functionA ofPROBABILISTIC
the constraint length of NONSEQTJENTIAL
the asymptotically optimumDECODING
for rates above Ra and whose performance
code. For all but pathological channels the bounds are asymptotically bears certain similarities to that of sequential decoding algorithms.
an upper bound is obtained. (exponentially) tight for rates above &, the computational
ALGORITHM
cutoff
er lower bound on error probability rate for low
of sequential decoding. As a function of constraint length the
We now describe a
performance of optimal convolutional codes is shown to be superior new probabilistic I. nonsequential
SUMMARY OF RESULTS
btained by replacing the sphere packing to thatbound
decoding algorithm which, as we shall show
of block codes of the same length, the relative improvement INCE Elias”] in the
first proposed theuse of convolutional
the tighter lower bound for low rates recently (tree) codes for the discrete memoryless channel,
Manuscript receivednext May section, is November
20, 1966; revised asymptotically
14, 1966. s optimum
it has beenfor rates Rthat > the performance of
conjectured
by Shannon, Gallager, and Berlekamp.17’Division, For
The research for this work was sponsored by Applied Mathematics
Office of R, = Research,
Aerospace E,(l). U.The S. Air algorithm
Force, Grant decodes
this an L-branch
class of codes is potentially tree superior to that of block
AFOSR-700-65.
d (6) is replaced by The author is withby the performing L repetitions ofof one
Department of Engineering, University of
codes of the same length. The first quantitative verification
basic step. We adopt
California, Los Angeles, Calif. this conjecture was due to Yudkinr2’ who obtained
the convention of denoting each branch of a given path
by its data symbol ai, an element of GE(q). Also, although
GE(q) is isomorphic to the integers modulo (r only when
Q is a prime, for the sake of compact notation, we shall
use the integer r to denote the rth element of the field.
I -lim [P 111 F T p(4pb’)
m In Step 1 the decoder considers all qK paths for the
first K branches (where K is the branch constraint length
&P(Y Y I 4P(Y I ~‘>Yl~ = &(i3. (17b) of the code) and computes all qK likelihood functions
ght line of (17a) is tangent to the curve of (6) n;?I1 p(y, 1ai). The decoder then compares the likelihood
P + 1)/p@;(p). Repeating the minimization with function for the q paths:
CLwe find (0, az, a, . . * 4,
E&l = min [(P + l)Ez - &RI 0, a2, a3, ... ad,
* ..................
0 < R < Elbl
=E,, -MC. (a - 1, az, a3, . . . ad
have for each of the qK-l possible vectors (az, a3 . . . ar;).
It thus performs qIcel comparisons each among q path
2 likelihood functions. Let the path corresponding to the
greatest likelihood function in each comparison be denoted
w rates a tighter lower bound than that of
1 is: the survivor. Only the q”-’ survivors of as many com-
parisons are preserved for further consideration; the
PX > & i -NE&) + o(N)lJ remaining paths are discarded. Among the qK-’ survivors
EE/Ma 127b, Class Project 2
 Jet Propulsion Laboratory
Interplanetary Error-Control Codes

No Coding (Pre 1969)

(32,6) Biorthogonal Block Code (1969 - 1975)
K =7, R = 1/2 Conv. Code + Viterbi Decoding (1977 -- 1986)
Plus Reed-Solomon if Data Compression is Used
K= 15, R = 1/6 CC/VD + RS (1986 -- 2004)
Turbo Codes (2004 -- ?)
LDPC Codes (2006 -- ?)
 No Coding: The Early Mariners

Mariner 2, 1962
Venus Flyby
Mariner 4, 1965
Mars Flyby
First close-up photographs of another planet.
Mariner 5, 1967
Venus Flyby
 (32,6) Biorthogonal Code +
“Green Machine” Decoding
+ + + + + + + +
+ − + − + − + −
+ + − − + + − −

Mariners 6, 7 (1969) +
+
−
+
−
+
+
+
+
−
−
−
−
−
+
−
+ − + − − + − +
Mars Flyby +
+
+
−
−
−
−
+
−
−
−
+
+
+
+
−
− − − − − − − −
Mariner 9 (1971) −
−
+
−
−
+
+
+
−
−
+
−
−
+
+
+
+ + + +
Mars Orbit
− − − −
− − − − + + + +
− + − + + − + −
− − + + + + − −
− + + − + − − +

The (8,4) biorthogonal code

 (32,6) Biorthogonal Code/
“Green Machine” Decoding
+ + + + + + + +
+ − + − + − + −
+ + − − + + − −

Mariner 10, 1973-1974 +

+
−
+
−
+
+
+
+
−
−
−
−
−
+
−
+ − + − − + − +
Mercury and Venus +
+
+
−
−
−
−
+
−
−
−
+
+
+
+
−
− − − − − − − −
Viking Mars Landers, 1976 −
−
+
−
−
+
+
+
−
−
+
−
−
+
+
+
+ + + +
Mars’ Surface
− − − −
− − − − + + + +
− + − + + − + −
− − + + + + − −
− + + − + − − +

The (8,4) biorthogonal code

 K= 7, R = 1/2 Convolutional Code
with Viterbi Decoding

Voyagers 1&2 (1977-- )

“Grand Tour”
Magellan Venus Radar Mapper (1989-1993)
Mars Global Surveyor (1997- )
 K = 15 Convolutional Codes with
Big Decoding

Galileo (1989 -- 2003)

A Sea of Troubles
Mars Pathfinder (1996- 1997)
Sojourner
 K = 15 Convolutional Codes with
Big Decoding

Cassini (1997 ---- )

Huygens Titan Probe, 2005
Mars Exploration Rover (2003--2004)
Spirit and Opportunity
 A Brave New World :Turbo Codes

Messenger to Mercury (APL Mission: 2004--2011)

Mars Reconnaissace Orbiter (Aug 2005 Launch)
Both use (8920, 1/6) CCSDS turbo code
 The Tanner Graph
Back to the Future: LDPC Codes

1 2 3 4 5 6
 
A 1 1 0 1 0 1
B 1 0 1 1 1 0 .
C 0 1 1 0 1 1
Mars Telecomm Orbiter 2010
1 2 3 4 5 6
And Beyond ? = = = = = =

A B C

22
 SUMMARY
-1
Uncoded
(32,6) Biorthogonal
-2 Mariner 10 (7, 1/2) Conv.Code
Mariner 4
Magellan
Voyager (7, 1/2) + (255,223) RS
Viking
log10 Pb

(15,1/6) + (255,223) RS
-3
(8920, 1/6) Turbo

-4
Galileo
-5 Messenger Pathfinder Voyager
SHANNON
LIMIT
-1.59 dB MER MGS
MRO Cassini
-6
-1.0 1.0 3.0 5.0 7.0 9.0 11.0
Eb /N0 , dB
 Claude Shannon:

“The fundamental problem of communication is that of reproducing

at one point either exactly or approximately a message selected at
another point.”

“Frequently the messages have meaning”

A Tour of the Solar System

On the Occasion of Andrew Viterbi’s 70th Birthday.

Ludwig van Beethoven, Moonlight Sonata

Daniel Barenboim, pianist
 Mercury
Mariner 10
1974
 Venus
Magellan
1990
The Far Side of the Moon
Apollo 16
1972
 Mars
Mars Global Surveyor
1997
The Surface of Mars
Mars Pathfinder
1998
The Asteroid Gaspra
Galileo
1991
 Jupiter
Voyager 1
1979
Jupiter’s moon Io
Galileo
1996
Io above Jupiter
Cassini
2004
Jupiter’s moon Europa
Galileo
2000
Jupiter’s moon Callisto
Galileo
2001
Saturn
Cassini
2004
Saturn’s moon Titan
Cassini
2004
Saturn’s moon Phoebe
Cassini
2005
 Uranus
Voyager 2
1986
Neptune
Voyager 2
1989
Pluto and its moon Charon
Hubble Space Telescope
1994

??
EEarthrise
Apollo 8
1968

We shall not cease from exploration

And the end of all our exploring
Will be to arrive where we started
And know the place for the first time.
-T. S. Eliot
Happy Birthday Andy!