PDA

View Full Version : Question Why does GSM system require C to I lager than 9


linhhy
2011-07-30, 01:27 PM
Hi every one,

Who can help me to clear the question "why does GSM system require more than 9 dB for C/I"?.
It required C / I greater than 9 to what purpose, e.g FER, BER, MOS or what?

Thanks for supporting!
ritgail
2011-07-30, 01:47 PM
A successful communication requires a defined c/i value with a acceptable service rate.(Error rate..BER,FER...). GMSK required 9db..
tremalnike
2011-07-30, 05:00 PM
Error correction is designed to work at max when C/I is 9dB or lager . Below 9dB some errors cannot be corrected so BER increases
linhhy
2011-07-30, 05:11 PM
Error correction is designed to work at max when C/I is 9dB or lager . Below 9dB some errors cannot be corrected so BER increases

Hi friends,

In 3GPP 05.05, GSM GMSK required 9db. But I would like to know for what?, e.g BER < 12%, or FER < 1%, or MOS > 3, ...?
linhhy
2011-08-04, 01:23 PM
Everyone can help me?
s52d
2011-08-04, 02:41 PM
Hi friends,

In 3GPP 05.05, GSM GMSK required 9db. But I would like to know for what?, e.g BER < 12%, or FER < 1%, or MOS > 3, ...?

without looking into books/specs:
for voice, old FR codec, minimum BER was specified (using shannon, this is C/I).
Some fading margin is added, of course.
Form BER, assuming proper error distribution, you can get FER.
From FER, you can get MOS.
So, MOS, FER, BER are correlated.

Now... AMR can work with lower C/I, frequency hopping can lower fading margin,
interference is not gaussian - we have to look at burst erasure rate/burst collison rate.
Thus, 9 db or 12 db is just guideline.

With EDGE on top: we try to keep C/I as best as possible for given network topology.

For realy story of GSM specs: look at some 15 years old book. When people
still remembered how chices were made.

BR
s52d
sran
2011-08-05, 02:42 AM
after demodulating, the receiver has to decide what the symbol is. If noise is too much, there will be error symbols which are beyond any error correction algo. This 9dB may be the empirical value?
s52d
2011-08-05, 03:38 AM
after demodulating, the receiver has to decide what the symbol is. If noise is too much, there will be error symbols which are beyond any error correction algo. This 9dB may be the empirical value?

Oh, no. There is quite defined math behind, so it can be easily calculated or modelled.
Information theory - Wikipedia, the free encyclopedia (http://en.wikipedia.org/wiki/Information_theory)
linhhy
2011-08-06, 12:53 PM
for voice, old FR codec, minimum BER was specified (using shannon, this is C/I).
BR
s52d

Dear S52d,

Could you give me any document what can help me to clear that you mention above.

Thanks!
s52d
2011-08-06, 04:06 PM
Dear S52d,

Could you give me any document what can help me to clear that you mention above.

Thanks!

Capacity (http://www.dsplog.com/category/capacity/)

a lot of discussions on Shannon, BPSK, GMSK etc....
rishi
2011-09-18, 01:18 AM
I dont think there might be c/i for 3g as we use wcdma technolgy for the system
RNP/O
2011-09-18, 01:21 AM
I dont think there might be c/i for 3g as we use wcdma technolgy for the system

hi you are wrong for 3G UMTS don't have C/I had only EcIo

thanks
s52d
2011-09-18, 02:03 AM
hi you are wrong for 3G UMTS don't have C/I had only EcIo

thanks

Carrier to Interference ratio is generic term.
EcNo is how it is mesured in UTRAN, it is easy to get real C/I out of it.

Lucky you, if you never had any intrference problem in W-cdma or ofdma systems.

BR
s52d