This page, as an example, refers to the outlink multi-Mbit/s carrier, from the hub, which is shared amongst all VSAT users to download internet web pages etc. The carrier is similar to a DVB-S or DVB-S2 carrier which carries several MPEG TV and audio programmes.

The carrier on the satellite is made up of a sequence of joined together pulses to make a continuous signal. Each pulse is a symbol. According to the modulation method each symbol represents 1, 2 or 3 etc bits of transmission rate data.

In phase shift keying (PSK) modulation each pulse is a burst of carrier signal with its sinewave zero crossing point timing adjusted forwards or backwards in time to constitute a phase shift. Phase shifts of 180 deg apply in BPSK, 90 deg in QPSK etc. A phase shift of 90 deg represents a time shift of 1/4 of a full cycle of the sinewave. The closer the spacing of the phase shifts, the more difficult it is to distinguish between them at the receive end, so for for each higher order PSK schemes more carrier to noise ratio is required. In 16-QAM modulation the amplitude and the phase are changes from symbol to symbol, making a matrix pattern with the dots even closer together, and thuis requiring even higher C/N ratio.

As a general rule if you have bandwidth to spare, then use a lower order modulation or a higher rate FEC (like 1/2 or 2/3) to spread the signal out. If you have power to spare then use a higher order modulation and/or lower rate FEC (like 3/4 or 7/8). Ideally you want to use all of both the available bandwidth and power simultaneously to obtain the highest user information rate.

If you use larger receive dishes you will always be able to increase the system capacity. If you are doing a point to point link it is well worth using larger dishes - spend more on the antennas and used advanced modulation technique modems, like Comtech Vipersat CDM-570L, to save on the space segment costs. If you have thousands of receive dishes then the aggregate cost of these is significant and you will want to allow smaller receive dish sizes even though this reduces system capacity and increases space segment costs.

Forward error correction is applied to the customer's information data at the transmit end.

so transmission data rate = customer information rate x 1/ (FEC rate).

FEC rate is typically in the range 1/2 to 7/8 so the transmission data rate is always significantly more than the customer information rate.

This page provides a key formula:

SR = Symbol Rate

DR = Data Rate = the information rate. This is the same as the customer
information rate if there is no framing, supervisory, conditional access or
encryption overhead added to the data stream in the modem. DVB modems add
significant overheads.

CRv = Viterbi forward error correction (FEC) Code Rate. Eg. 1/2,
2/3, 3/4, 5/6, 7/8

CRrs = Reed Solomon forward error correction (FEC) Code Rate. Eg. 188/204

If some other type of FEC coding method is chosen, such as Turbo coding, just use whatever FEC rate is selected (e.g. 5/16, 21/44, 3/4, 7/8, 0.95 )

m = modulation factor (transmission rate bits per symbol). BPSK=1, QPSK=2, 8PSK=3, 8QAM =3, 16QAM=4 etc

The bandwidth of the carrier at the -3.8 dB points is approx the same as the
symbol rate.

The bandwidth of the carrier at the -12 dB points is approx 1.28 times the
symbol rate.

The expression "occupied bandwidth" is used to refer to a bandwidth 1.19
times the symbol rate, approx -10 dB points..

* The allocated bandwidth, i.e. spacing between carriers needs to be approx
1.35 to 1.4 times the symbol rate. * If you put the carriers too
close together you will start to see more adjacent carrier interference. If you
put them too far apart you will waste expensive bandwidth, so choose some
compromise that includes some, but not too much interference. A suggested
adjacent carrier interference allowance in link budgets is 28 dB on each side.
You choose. If you can avoid high spectral density carriers adjacent to
low spectral density carriers it will help.

For example: Symbol rate=27.5 Msym/s. Bandwidth = -1 dB 20.9 MHz, -2 dB 24.2 MHz, -3 dB 26.25 MHz, -3.8 dB 27.5 MHz, -4 dB 27.7 MHz, -6 dB 30.3 MHz, -12 dB 35 MHz. For a spectrum analyser view of the spectrum see this page

Some examples:

Mod-ulation and FEC rate and FEC coding method | Minimum threshold Eb/No (BER = 10E-8) Add an operating margin to this for clear sky set up, depending on C or Ku band and rain area. |
Info-mation rate bit/s |
Symbol rate. per info-mation bit rate (e.g. 1 Mbit/s info x 0.667 = 667 ksps) |
Occ-upied band-width Hz at -10 dB
points.1.19 times the symbol rate |
Allo-cated band-width Hz (sugg-ested
carrier to carrier spacing)1.35 times the symbol rate |

QPSK 1/2 rate FEC Viterbi | 7.2 dB | 1 | 1 | 1.19 | 1.35 |

QPSK 21/44 FEC Turbo | 3.1 dB | 1 | 1.048 | 1.246 | 1.414 |

QPSK 3/4 rate FEC Turbo | 4.3 dB | 1 | 0.667 | 0.793 | 0.9 |

QPSK 7/8 FEC Turbo | 4.4 dB | 1 | 0.571 | 0.68 | 0.77 |

8-PSK 3/4 rate FEC Turbo | 6.7 dB | 1 | 0.444 | 0.53 | 0.6 |

16-QAM 3/4 rate FEC Turbo | 8.1 dB | 1 | 0.333 | 0.397 | 0.536 |

16-QAM 7/8 rate FEC Turbo | 8.2 dB | 1 | 0.286 | 0.340 | 0.386 |

Last amended 4 Oct 2007

6 April 2004: link to
dvb-s carrier spectrum analyser plot added.

6 Jan 2009:
Agilent
Application note 1298 - Digital Modulation in Communications systems added
(warning large pdf file 650 k bytes). A really good explanation of digital
modulation.

4 October 2006. Turbo code, adjacent carrier interference, allocated
bandwidth and occupied bandwidth mentioned.

4 Jan 2007, 4 Oct 2007: Table of example modulation methods, FEC coding
and FEC rates added, amended

29 July 2015: amended