Forums

Question about Reed-Solomon error control code

Started by Unknown April 4, 2005
Hi,
I am simulating equalizer with Reed-Solomon channel code a frequency
fading channel. The fading channel has a severe spectral null, I use
frequecy domain equalization to contre the null. At present I think the
equalizer works well. When I add the RS(170,180) (8 bit) channel code
for a data frame, I find the BER performance is not improved at all. It
is even worse a little bit than that of without channel code. I know I
have to account for the confidence level. All in all, the results are
very similar. When I control the channel to generate shorter
consecutive errors, i.e. not as severe as the first one, I can see some
improvements.
My question is: for a severe fading channel (not AWGN), a short t
(error corection length t/2) RS code would not improve the performance?
Or, it should absolutely improve the BER? I cannot find any error until
now.

Thanks in advance

>>>>> "freelait2000" == freelait2000 <freelait2000@yahoo.com> writes:
freelait2000> I am simulating equalizer with Reed-Solomon channel code a frequency freelait2000> fading channel. The fading channel has a severe spectral null, I use freelait2000> frequecy domain equalization to contre the null. At present I think the freelait2000> equalizer works well. When I add the RS(170,180) (8 bit) channel code freelait2000> for a data frame, I find the BER performance is not improved at all. It freelait2000> is even worse a little bit than that of without channel code. I know I freelait2000> have to account for the confidence level. All in all, the results are freelait2000> very similar. When I control the channel to generate shorter freelait2000> consecutive errors, i.e. not as severe as the first one, I can see some freelait2000> improvements. freelait2000> My question is: for a severe fading channel (not AWGN), a short t freelait2000> (error corection length t/2) RS code would not improve the performance? freelait2000> Or, it should absolutely improve the BER? I cannot find any error until freelait2000> now. I think you can answer this if you know how long the average fade length of your fading channel is. If your average fade length is t/2 symbols long or more, the RS code won't be able to correct anything, so performance may be the same or worse than if you had no RS code at all. There's a lot of handwaving here, but I think it gets to the point. Ray
what is the BER during the fade before the RS FEC and after the RS FEC?

RS FEC has a threshold below which it cannot improve the BER and will
make it worse.  

Mark