RE: Bit errors, is this really an issue ?

["Marc Dumon" <[e-mail]>]

Dear Jordi,
I'll try to give you an answer to your question:

For a satellite link the BER is a specified requirement. For a specific BER
one can determine the Eb/No (Energy per bit / Noise spectrum density) from
the formula
" Probability of making an error=1/2 * erfc(Eb/No)^(1/2). (erfc is the
complementary error function)."

Example : if the satellite link requires a BER of 10^-5, the above formula
gives us a Eb/No of +/- 9dB.
Conclusion : The BER can be reduced by increasing the [Eb/No](dB). This can
be achieved by increasing the transmitted power (Eb)
and/or reducing the noise (No).
In a satellite link the Eb/No can be increased by using bigger antenna's,
more power, and/or better equipment (with lower system noise).
It's logic that the cost of the equipment increases aswell.

Another technique to reduce the BER (for a fixed Eb/No) is by using error
control coding.
There are two important types of error control coding :
Error detecting code (ARQ) and forward error correcting (FEC).
Error detecting code can detect that one bit in a code word is wrong but it
cannot know which bit it is. The complete word must be retransmitted (ARQ).
With forward error correction (FEC) a the wrong bit is corrected without
retransmission.

The "coding gain"is the difference in Eb/No with and without coding.
FI : Coding gain for BER = 10^-6 with FEC (7/8) = 3,7dB; for FEC(1/2) =
6,4dB; and for FEC(1/2 + Reed Solomon) = 7.9dB.
Example : for a certain system with a given Eb/No of 3dB the BER increases
from +/- 5*10^-1 (!) without error coding
to +/- 10^-7 if FEC (1/2) with ReedSolomon is used.
To achieve the same quality (BER = 10^-7) without error correction you need
an Eb/No of +/-12dB.


So your conclusion is correct : a satellite link can approach fiber BER,
provided adequate FEC is used and/or enough power is available.

Now the rain :
This is more complex. The rain attenuation increases with increasing
frequency (Ku-band is more sensitive to rain than C-band),
and depends also on the polarization (attenuation for horizontal
polarization is much bigger that for vertical polarization)
and the geographical location, the time of the year and the elevation.
The ITU publishes statistical tables in which the world is divided in
different rain zone's. (Belgium is in rain zone E)
This tables show the fraction of time that a given attenuation will be
exceeded.

Example : for Belgium at 12 GHz and H-polarization :
An average of 99.999% of the year the rain attenuation does NOT exceed
7,38dB (0,088h/year it exceeds 7,38dB)
99,991% of the year the rain attenuation does NOT exceed 3,42dB (0,789h/year
it does)
and 99,000% of the year the rain attenuation does NOT exceed 0,41dB.

If you have a system with an Eb/No of +/-8dB without error correction you
will have a BER of +/- 10^-3 under clear sky condition.
An average of 0,088h/year the rain attenuation exceeds 7,38dB resulting in a
broken link.
An average of 0,789h/year the  rain attenuation exceeds 3,42dB resulting in
an BER of > 10^-2
and 99,000% of the year the BER will be +/- 10^-3.

Now if yo use exactly the same system but with FEC + RS you will have a BER
of <<10^-7 with a Eb/No of 8dB.
An average of 0,088h/year the rain attenuation exceeds 7,38dB still
resulting in a broken link.
But average of 0,789h/year the  rain attenuation exceeds 3,42dB resulting in
an BER of <10^-7.

I hope I gave you an answer on you question, and if you need additional
information,please do not hesitate to contact me.

Marc Dumon
____________________________________________________________________
Marc Dumon
ORBIT satellite telecom nv.
Lange Leemstraat 144A
B-2018 Antwerp
Belgium
Tel : [phone]
Fax : [phone]
e-mail : [e-mail]











> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]]On
> Behalf Of Jordi Nelissen - Alcatel Bell Research
> Sent: Thursday, September 24, 1998 2:56 PM
> To: tcpsat
> Cc: Jordi Nelissen@RC
> Subject: Bit errors, is this really an issue ?
>
>
> Dear tcpsat,
>
> I have a question with respect to the effect of BER on TCP performance
> on satellite links. From draft-ietf-tcpsat-stand-mech-06.txt, I read :
>
> 'Typical bit error rates (BER) for a satellite link today are on the
> order of 1 error per 10 million bits (1 x 10^-7) or less frequent.
> Advanced error control coding (e.g., Reed Solomon) can be added to
> existing satellite services and is currently being used by many
> services.  Satellite error performance approaching fiber will become
> more common as advanced error control coding is used in new systems.'
>
> As I understand this, in optimal operation, a modern
> satellite link can
> approach fiber BER, provided adequate FEC is used. Further on I also
> read :
>
> 'FEC should not be expected to fix all problems associated with noisy
> satellite links.  There are some situations where FEC cannot
> be expected
> to solve the noise problem (such as military jamming, deep space
> missions, noise caused by rain fade, etc.).'
>
> Let us take the most common situation, i.e. rain fade.
Can someone
> explain to me its quantitative impact on BER. Is the satellite link
> completely out of order when it rains or can we still find rescue in
> methods like ARQ or some TCP which is able to distinguish corruption
> from congestion ? Is there information available on the noise
> characteristics on modern satellite links ? (BER, distribution of bit
> errors, ...)
>
> thanks, Jordi
> --
> _____________________________________________________________________
> Jordi Nelissen - [e-mail]
>
> Alcatel Research - Network Architecture - Traffic Technology
> Francis Wellesplein 1, 2018 Antwerp, Belgium
> Phone: external: [phone], alcanet: +6057192
> Fax:   external: [phone], alcanet: +6059932
>
> Visit our webpage :
> http://www.rc.bel.alcatel.be/projects/networkarchitecture/traffic
> _____________________________________________________________________
>