CiM-550 IP Enabled Satellite Modem

Rev. 2

Forward Error Correction

CD/CIM550.IOM

When Comtech EF Data first introduced the Turbo Coding option in the CiM-550, only Rate 3/4 QPSK was offered. This is an excellent choice for applications where maximum bandwidth efficiency is desired, in combination with high coding gain. However, with Firmware Version 1.15 (and later) Comtech EF Data has introduced two further code rates - Rate 21/44 BPSK (very close to Rate 1/2) and Rate 5/16 BPSK (very close to Rate 1/3).

These two rates were developed to address an entirely different case, namely that of transmission from very small antennas, with limited transmitter power. For a dish antenna, the gain is directly proportional to its area, and the lower the gain, the less directional the antenna becomes. Thus, in satellite transmission, even though the dish may be perfectly pointed at the desired satellite, if the beamwidth is wide enough, adjacent satellites in the orbital arc will also be illuminated. This is a potential source of interference, and for this reason the ITU (International Telecommunications Union) place strict limits on the power spectral density (also referred to as flux density) of signals arriving at adjacent satellites.

One obvious method to reduce the level is to spread the transmitted signal over as wide a bandwidth as possible. In the past, this has sometimes been achieved using a Spread Spectrum modulation format, but this has severe penalties as far as demodulator complexity is concerned. However, by using BPSK modulation, and high FEC code rates (up to Rate 1/3, for example) the power spectral density may be reduced. Taking Rate 1/2 QPSK as a baseline, moving to Rate 5/16 BPSK Turbo Product Coding gives an immediate reduction in power spectral density of 5.0 dB.

The increased coding gain of this FEC method allows a further reduction in transmitter power. Using Rate 1/2 Viterbi with concatenated Reed-Solomon as a baseline example, Rate 5/16 provides 1.5 - 2.0 dB improvement in coding gain. Putting these two factors together yields an overall reduction in power spectral density of approximately 7.0 dB. This simultaneously permits a smaller antenna, and reduced transmitter power. The disadvantage is the increased spectral occupancy of the carrier, and it will depend on the particular satellite operator to determine if this poses a severe economic problem.

When operating at these higher code rates (21/44 and 5/16), the demodulator is forced to operate in a region where the Ebt/No (also referred to as Es/No) is negative - in other words, there is more noise than signal. For this reason it is more difficult for the demodulator to accurately measure the Eb/No, and so users may see more variation than normal when viewing the Eb/No on the RX PARAMETERS screen, and when operating near threshold. This is most pronounced in the case of Rate 5/16.

One unexpected advantage of the Turbo Product Coding method is that, unlike concatenated Reed-Solomon methods, there is no pronounced threshold effect. The demodulator and FEC codec can stay locked down to very low values of Eb/No, but at this level the output BER becomes unusable (around 1 x 10-2).

In many cases, FEC methods which provide increased coding gain do so at the expense of increased processing delay. However, with TPC, this increase in delay is very modest. The table below shows, for the CiM-550, the processing delays for the major FEC types, including the three TPC modes:

94