Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: RE: [EFM] EFM Active Architectures


If we take seriously the estimates of the reletive
cost of deployment of a fiberplant vs. the active
electronics (90%/10%, or so), I think that it makes a
lot of sense to give long and careful thought to
insuring that the basic fiberplant topology doesn't
become obsolete. On the other hand, periodic upgrades
to the end opto-electronic packages would seem to me
to be acceptable over the working lifetime of the
fibre. This isn't unlike the periodic upgrades of
NIC's in LAN's.

To burden, say, an ONU with a receiver capable of some
day accepting a 10GBE datastream, would be the sort of
overkill that will make any system prohibitively
expensive. The basic PON arcitecture (SM fiber and
splitter) has, for all intents and purposes, unlimited

This view would suggest that making the ONU an easily
replacable module (like a set-top box) might be a
better choice than hanging it on the side of a house.
Additionally, this solves the problem of power (plug
it in a wall socket) and will save quite a bit of
money by putting the opto-electronics in an
environmentally benign location. Don't underestimate
the difficulty of getting these devices to function
from -40 to +85 degrees.


--- Frank Coluccio <fcoluccio@xxxxxxxxxxxxxxxxx>
> [Second attempt at getting this message out... ]
> Frank Effenberger, 
> Yes, as with Dave's half-PON concept which I also
> like very 
> much, the TDM and slicing options are also of
> interest and 
> certainly deserving of further investigation.
> The thing that occurs to me, at least in the two
> optical 
> tracks of EFM (p2p and pt-mpt), is this: 
> How do either of these proposed designs, or _any_ of
> the 
> others, for that matter, lend themselves to
> extensibility 
> to higher speeds down the road? Stated another way,
> should an
> approach that "locks in" an upper bound on
> throughput be of 
> concern? As in, being able to deliver the next two
> powers 
> of 10 beyond 100Mb/s (since 100 is often cited as
> the target 
> delivery rate at this time), at least, without a
> major forklift
> when they arrive. What happens to be friendly at 100
> Mb/s 
> between the OLT and/or the field "thingie" and the
> ONU at 
> this time may not be as friendly at 10Gb/s or
> higher. 
> Or, is this even something that should concern EFM
> at this time? 
> > 
> > All, 
> > We have considered this topology to some degree
> ourselves.  
> > The variant we considered is to use TDM in the
> downstream, 
> > and WDM in the upstream.  The advantages of doing
> this are 
> > that you avoid the TDMA protocol issues, and you
> save N lasers. 
> > over doing it point-to-point.  You also reduce
> your spectrum 
> > requirement by a factor of 2 (half as many
> wavelengths), and 
> > this can be key if you are using coarse WDM. 
> Since cost is 
> > key, CWDM would be a good direction.  Spectral
> slicing is 
> > also an interesting option.   
> > Versus a TDMA PON, you must add N receivers at the
> CO; however, 
> > an array of receivers is *much* more tractable
> than an array 
> > of WDM lasers.  
> > Lastly, the use of WDM in the upstream retains the
> all passive 
> > outside plant advantage of PON.  No field
> electronics. 
> > 
> > Any interest? 
> > Frank Effenberger. 
> > 

Do You Yahoo!?
Make international calls for as low as $.04/minute with Yahoo! Messenger