On Wed, Dec 07, 2005 at 12:22:24PM +0100, Petr Nejedly wrote: > Andy Burns wrote: > > >see http://adslpipe.co.uk/linuxhtpc/images/gtzap1.png > > Just a little note: BER is a Bit Error Ratio and should be the number of > corrected bad bits. > Some drivers report vitterbi error rate (number of soft error recognized > in the vitterbi decoder), > while some report reed-solomon error rate, that is the number of wrong > bits on the input > of the reed-solomon decoder (while there were no uncorrected blocks). > > It's a pity that the drivers mostly report an actual number of error bits > over unspecified (and maybe variable) time span. > It would be much better if the API required reporting real BER (R=ratio), > that is, the ratio between wrong bits and all bits received (over a > reasonable period, > like 0.5s). > In that case the user would immediatelly recognize whether the signal is bad > or already good enough. It is said that DVB's RS corrector can produce > a "quasi-error free" output (i.e. very low probability of uncorrected block) > in case the BER is below 1*10^-4 (a most 1 bad bit in 10.000 bits). > You coud even have a marker on the BER meter, denoting the 1E-4 BER mark. > > Somebody correct me if I'm wrong. Your description is wrong. BER is the bit error rate (usually viterbi) as read from some hardware register. The driver doesn't attempt to give you more than you need: A number which serves as an indication of signal quality. Lower number, better signal. Johannes
