In simple terms a PON is an optical network, where upstream and downstream data paths are combined/split using optical spliters.
For a better understanding read: http://en.wikipedia.org/wiki/Passive_optical_network .
Whilst ITU standards already exist for the implementation of PoNs, G.938 and successors, it would seem delivering a system to these standards using ROS and routerboard would require considerable investment in development/testing.
What about a simpler propriety system then?
Consider this.
Headend = CCR1036-12g-4cs ,has 1sfp fibre feeding a 32 way splitter
In each home = 1 off RB2011UAS-2HnD-IN with 1sfp feeding back to the splitter.
Now data flowing on the fiber from the headend to the home has no problems, layer2/3 will ensure it reaches the proper destination.
Problem comes when you consider data flowing from the home back to the head end. You will get collisions and no way of detecting them. lost and corrupted packets.
What is needed is some TDMA, so each home transmits data in turn. G.938 already does this, but the technology I think is difficult from the ROS point of view and the SPF adaptors are expensive.
My question is this then: Could the mikrotik’s propriety nv2 tdma technology (or something like it) be adapted for use on fibre ethernet?
I can imagine such a system having 1GBPS shared download and 10-20Mbps upload per home.
Imagine 1 CCR1036-12g-4cs with 4 X SPF fibre adaptors each with a 32 way splitter = 128 ports.
Port cost is then less than 15$ per port.
Use RB2011UAS-2HnD-IN for the home, then CPE cost is around $100, for a fibre to the home product that includes WiFi.
CAN IT BE DONE ?,
I think at such price breaks we would have what ubiquity would like to call a “disruptive technology”
and maybe a whole new market for Mikrotik products…
MT anounced that they didn’t support passive optical interfaces and didn’t have such plans yet and suggested to use integrated ONT or OLT SPF modules for this case/scenario and even pointed link to Youtube video on one ONT module[wasn’t shipped yet].
Yes PON SPF units are expensive, the proposal here would use standard “off the shelf” spf modules.
I am proposing an alternative proprietary solution, using standard cheap SPF modules, but not compliant with the official G.938 PON standards.
To be more clear:
At head end, Routerboard+ROS with a normal gig Ethernet SPF module transmits standard IEEE 802.3 frames, downstream data together with added sync frames.
An optical splitter (say 32 way) delivers this signal to 32 Homes, each with a routerboard+spf module receiving the downstream data and sync frame.
Each home the uses a “time slot” to time the sending of upstream data, to an optical combiner and back to the spf at the head end, hence avoiding collisions with upstream data from the other 32 homes.
Whilst the “sync frames, and time slots” I propose do not exist in the 802.3 standards, the proposed solution would seem to be similar to the proprietary NV2 mechanism already used by Mikrotik’s Router OS.
This solution can bring mikrotik to more advanced level like zte, huwaei
i agree this is fantastic feature!
We need to hear mikrotik verdict on pon solution on routeros
generally cause xPON downstream is encrypted[which mean some stress on CPU/SoC] most NW HW vendors tend to evade native support thus[to save tiny/outdated SoC in production/products to maximize profit], i suppose.
generally GEM protocol used in one standard and ATM or EoIP[IPoE ?] in another.
generally its about scale of market - downside of HR/manpower/money amount necessarily 2 implement it - outweighing potential profits, especially because most xPON installer/operators/telco tend to lock-down deployment for 3-rd party hardware on-purpose[under various bogus cover stories], further reinforced by proprietary subs-standard firmware tweaking, sadly.
