Comtech EF Data CDM-710G - Manual (Page 161)

11–1
Chapter 11. FORWARD ERROR
CORRECTION OPTIONS
11.1 Introduction
DVB-S2 has defined a new generation of performance that boosts throughput by about 30% over
the same transponders using a new type coding that exceeds the capability of concatenated
Viterbi and Reed Solomon coding.
The CDM-710G High-Speed Satellite Modem operates with error correction base upon the DVB-
S2 standard: QPSK, 8PSK, 16APSK and 32APSK with concatenated Low Density Parity Code
(LDPC) and Bose-Chaudhuri-Hocquenghem (BCH).
11.2 LDPC and BCH
LDPC and BCH is a concatenation technique. LDPC is a very powerful coding scheme with
significant, Near-Shannon Bound Performance. In some cases, as the carrier-to-noise ratio increases,
the LDPC error correction starts flaring toward an error floor so BCH error correction follows LDPC
and eliminates the flare for any practical range of error rates.
LDPC also functions differently than Viterbi decoding by using iterative decoding. In this process
the data initially corrected by the LDPC decoder is re-encoded and run through the decoder again
to correct additional errors. Key to this is the soft decision output from the LDPC decoder and a
high-speed processor operating at a rate much higher than the data rate. The LDPC decoder runs
the iterative process as many times as possible before corrected data is finally outputted to make
way for a new block of data entering the decoder. LDPC also uses interleaving to spread the
errors. In contrast, Viterbi error correction operates by passing data through the convolutional
error correction process a single time.
The error correcting capability of LDPC is improved by using large block sizes. This also
increases latency. However, in one-way broadcast applications this is not a drawback. Links with
LDPC normally operate at multi-megabit data rates where latency effects are reduced. The
standard block size for LDPC is 64,800 bits, and for lower data rate applications there is a short
frame block at 16,800 bits that suffers only a small error correcting loss (0.2 to 0.3 dB) compared
to the standard block.