Developments on GPON are WDM PON which uses seperate wavelengths as you mention but the kit is not (yet) common and a lot more expensive at present.

The splitters would then split a seperate wavelength to each premise giving vastly higher potential speeds.

P2P systems are easy to do in small / sparse developments. But imagine the dense areas when you would need 1000s of fibres the fibre management becomes interesting! ( Look at Oldham with 61k premises on samknows!) This is why MDUs often use Metallic tails to the individual units and fibre to the Basement (or sub node) this can reduce the difficulty in the 'exchange' by a factor of 2-400!.

See Irvimic's response, 25th November 2015.

Thanks for this discussion guys, it's bringing back memories of my own work with Marconi in the late 90s when we were working on WDM technologies. At the time most solutions for the mux and demux involved an optical-electrical-optical switch which was inefficient and needed power. We were working on a fully optical system which I presume is now a reality?

In reply to a post by andyhurley:

We were working on a fully optical system which I presume is now a reality?

I have a vague recollection of the hardware guys working next to us softies getting a demo of a pure optical switch working in the late eighties. That was Plessey/GPT, so I guess a precursor to Marconi. They were pretty pleased with themselves...

Yeah, the reports I heard were all rather self-important too. I was actually working for Nokia on software for the switches (precursors to the modern DSLAMs I guess - a few optical connections and lots of POTS designed to fit in exchange racks or cabinets). Marconi bought us and we were subjected to the rhetoric of WDM being the future, ironic that Marconi went bust a few months later...

In reply to a post by Irvimic:

If fibre is already in the ground then this would be harder to change from PON to P2P as it requires the costly capital works in order to lay more cables on typically developed land or roadways.

Ironically, that is probably the position we are *already* in, here in the UK.

The biggest saving offered by a PON over PtP is in the amount of fibre at the higher levels of the tree - between exchange and splitter.

The mostly-FTTC rollout has come with a rollout of the first layer of a future fibre access network - the fibre spines running from exchanges to chains of aggregation nodes. These will already, I reckon, be dimensioned more appropriately for PONs than for PtP.

Even though the current FTTC cabinets use PtP fibre, g.fast seems to be heading towards GPON and 10G-PON uplinks. I don't know if that is because it makes it easy to trial, to start with, or because that is the best long-term business case.