Template Tools

We've moved

Everything since June 1 is at  fastnet.news. This is just the archive from before June, 2015

Gigabit Easy with GPON and 10G PON for Decades
Written by Dave Burstein   
Friday, 04 January 2013 01:35

GPON can handle even 4K TVsHarstead and Sharpe of Alcatel base forecast on conservative traffic projections. 10’s of millions of homes, including millions at Verizon, have GPON connections easily capable of delivering a gigabit downstream to each home 99+% of the time. While the total is only 2.4 gig shared, almost never is the actual demand more than 1 gigabit. Any user needing their own gigabit is easily accommodated.

Hong Kong Broadband Network happily sells the gigabit service for about $30 but most are holding back. Australia’s National Broadband Network intends to offer the gig to millions starting in 2014. Verizon won’t sell more than 300 megabits.

That 2.4 gigabits shared, the GPON total, delivers a gigabit reliably surprises most people. The 2.4 gigabits is shared up to 32 ways, an average of about 70 megabits/home. In practice, the total demand is almost never even a single gigabit, leaving more than a gig for any user who demands it. Harstead and Sharpe examine the question of likely change over time as homes require more bandwidth. They believe looking at the peak throughput, not the average, is a much better measure of GPON.

Actual data is the average home draws an average of 100k-200k today, with the Alcatel paper sourcing that figure to Gartner/Cisco and some semi-official Japanese figures. Other estimates from actual networks are similar. Alcatel is presumably informed by the actual demand on the hundreds of carrier networks they support.

Peak usage is higher but still not enough to strain 2.4 gig GPON. Even assuming each of the 32 homes is watching Netflix and has several others in the home online, demand is only 5-10 megabits. Adding some HD video calls and the like doesn’t change that substantially. 10 megabits by 32 users (320 megabits) is less than 1/7th of the capacity of GPON.

It’s slightly more challenging to a carrier selling IPTV over the same fiber. Four TVs running HD in every home at the same time still leaves headroom in 2.4 gigabits, even is some are 3D. It’s only when (if?) most TVs are running at 2160P that IPTV might be stretching GPON capabilities. Those sets are just coming to market and offer little advantage in sizes below 60 inches. Harstead and Sharpe show that even a large volume of 2160P leaves headroom. I and most projections don’t expect that until well into the next decade - and even then GPON, not even 10G PON, probably can handle the load. 

10G PON with four times the capacity is close to coming out of the labs. Even if a significant numbers of TVs were upgraded to 3D 4320P, 10G PON can easily handle the loads. The most probable scenario, I believe after reading this paper, is that GPON --> 10G PON has decades of headroom.

The article’s primary purpose was to compare TDM PON - the GPON standard now and the soon come 10G PON - with WDM PON, considered by some the natural next step. WDM - wave division multiplexing - provides a separate hundred megabit or gigabit wavelength to each home. Ericsson has announced a 40 gigabit WDM system which has more total capacity than 10G PON. But it can only deliver a gigabit to each home. If GPON can match that gigabit, the theoretical advantages of WDM becomes insignificant. Here’s the article abstract

“It has been widely asserted that TDM PON technology will not be able to effectively cope with future increases in subscriber bandwidth demand, and that WDM PON architectures must succeed it. In this article we compare the historical trend of TDM PON bandwidth supply and the growth of residential bandwidth demands to the end of this decade, and conclude that bandwidth supply is growing much faster than the demand, calling into question the need for WDM PON. Further, we distinguish between sustained bandwidth demand and burst bandwidth demand, and further conclude that TDM PON will deliver future ultra-high speed services more efficiently than WDM PON.”

Ed Harstead and Randy Sharpe’s article, Future Fiber-To-The-Home bandwidth demands favor Time Division Multiplexing Passive Optical Networks is in IEEE Communications Magazine • November 2012. Unfortunately, it’s behind a paywall.

WDM PON has been interesting but going nowhere for a decade because it cost more than GPON and carriers didn’t see practical advantages. H&S make a very convincing case that GPON will be cheaper and deliver better peak performance for many years. Unless Internet use changes in unlikely ways, WDM has no place in end-user networks.