ECOC 1975 - ECOC 2013

171
RANDOM CODING FOR DIGITAL OPTICAL SYSTEMS
C. Game & A. Jessop
Introduction
The choice of a transmission code is a compromise between producing a
signal which is easily regenerable against one which maximizes repeater
spacing. The code should also provide an independent means of measuring
the binary bit error rate. Over the years wire line systems have evolved
that employ line codes which produce a signal having adequate timing
information for wide-band (low Q) timing extraction circuits and which
can be transmitted by a system with low-frequency limitations. Optical
fibre systems, while retaining the same need for simple and cheap timing
extraction circuits, do not have the same low-frequency limitations.
Wire line systems avoid low frequencies to obviate the need to equalize
cable down to low frequencies, to facilitate the protection of equipment
from surges and to facilitate power feeding and supervision over the
same cable as the signal. These restrictions do not apply to optical
fibre systems, the low frequency limitation being one of a.c. coupling
within the repeater. The implications of this difference and those listed
below should be investigated before choosing a line code for optical systems
Factors Affecting Choice of Code for Optical Systems
Choosing a code involves consideration of the following:
(a) means of limiting the laser mean pulse density
(b) spectrum manipulation (e.g. elimination of d.c. content)
(c) repeater timing information
(d) in-traffic error monitoring at terminals
(e) location of faulty repeaters
(f) data transparency
(g) code re-frame time if applicable
(h) efficient use of information capacity
(i) general simplicity of repeater
(j) non-linearity of source and detector
(k) dispersion in the fibre (pulse broadening)
This paper shows that all the requirements above can be met by the use
of scrambled binary as line code except for point (d) which can be met
by a little added redundancy. Whether or not point (e) needs to be met
depends on whether in-traffic location of a failing repeater is a
requirement. For an optical fibre system with negligible pulse broadening
the maximum repeater spacing is obtained with a binary system whether the
source is peak or mean power limited. This is because the increase in
signal-to-noise ratio due to reduced bandwidth by increasing the code
radix is more than offset by transmitting less power per level. If pulse
broadening is significant t.hen a higher radix code could maximise repeater
spacings. However. progressive improvements in optical fibre attenuation
have usually been accompanied by reduction in dispersion; with currently
available multimode fibres transmission is mainly attenuation limited for
speeds up to 140 Mbit/s.
C. Game & A. Jessop are with STL Ltd., Harlow, Essex, UK.