Re: Quick 64b/66b question
David,
We can easily calculate the probability for bad packet acceptance as:
The no. of min sized packets in 125us ~ 2^11.
Probability of falsely accepting atleast one of them in a bad link ~ 1 -
(1-2^-32)^(2^11)
~ 2^-21
So probability of accepting a bad packet at startup = 2^-21 * probability
of bad link at
startup.
All this probably doesnt matter since the link initialization at the
MAC ends
might take atleast 20ms (if not more)
http://grouper.ieee.org/groups/802/3/ae/public/sep00/muller_1_0900.pdf
In any case, I have no problem - other than having to simulate longer -
with having the hi_ber start off high and wait for 125us before asserting
sync_done = true.
Regards,
Birdy
David Gross wrote:
> Hi Birdy,
>
> I have a quick question I was hoping you could answer.
>
> For synchronization purposes, two conditions must be met - frame_lock &
> hi_ber = false. The first ensures that a candidate postion for the 2-bit
> preamble is a valid one (64 valid Sync-Words in a row - i.e.: 01 or 10).
> The second ensures that there is a minimum quality of BER on the link
> (10^-4). Both are required to go into sync. So, my question is:
>
> Why is hi_ber set to false on reset or power_on? wouldn't this allow for
> a possibility where synchronization is established quickly, but a high
> BER might be present (and only after 125us or ~20000 64/66 frames will
> sync_done be false again)? It would seem that reset condition could be
> counter-intuitive to the intent I had originally envisioned. Please let
> me know if I'm way off on this one. Thanks.
>
> -Dave