Close
Close

Published

MoCA repeats field test that rocketed it to initial success

MoCA is very keen to put some distance between its own system performance, that of rival wired home networking groups such as Homeplug and G.hn, but especially when it is considered against WiFi or when working with WiFi.

Intuitively every analyst knows that the protected environment of a coaxial cable produces the same results time and time again, virtually anywhere you place the cable, whereas WiFi obeys laws known only to mathematicians with a flair for the dark side, and its performance is rarely predictable.

It is this which has motivated a field test from MoCA of MoCA 2.0, claiming that at 90% of outlets in any given home, it performs at 400 Mbps in the real world. Even here there remains some confusion. WiFi devices are making claims in 8 way MIMO configurations (Quantenna) this week using MU-MIMO, of something in the order of 10 Gbps, and it’s tough to market against those kinds of theoretical PHY based numbers.

Quantenna has been one of the more vociferous players suggesting that no modern home needs any more networking than WiFi, and given that it will be a few months for even one home to have a working MU-MIMO device in it, and in most cases more like 12 months, it makes comparisons really tricky.

Typically if you go back a year, or maybe two, WiFi RFPs, that is documents which ask for proposals from pay TV operators, were calling for 25 Mbps in every portion of a WiFi home. That’s all the nooks and crannies, from basement to attic and it varies hugely from the drywalls of the US to the cement MDUs of Europe. These days in the 802.11ac world we would anticipate RFPs in WiFi to be pushing the 100 Mbps ceiling, somewhere between 75 Mbps everywhere, up to 100 Mbps. MU-MIMO will not push that throughput any higher we imagine, except to say that it should support more concurrent devices, perhaps 4 or 6 or 8, but that will be a year from now, but even then this is not comparing like with like and there are other pros and cons to be considered.

WiFi cannot do this all of the time, and will slide into a worse performance and back again, given any kind of interference, be it physical – walls, people walking around, or radio – other devices giving off random radio waves, as well as congestion – the 2.5 GHz spectrum which early WiFi used is full of other devices, and 5.0 GHz is filling fast. Also coaxial cable is only any good where you have laid it. You will not plug a tablet into a coax socket, you will only use WiFi for that, and you won’t put coax in your bathroom. The future home network will likely become an amalgam of fixed lines and wireless, and potentially that can be on any level – multiple fixed architectures and multiple wireless technologies. If you add the Internet of Things into the argument, it becomes inherently complicated, and it all has to be designed so that it is all plug and play – there is a long road ahead.

It seems that for some time to come there will be a fixed line backbone in most households in order to stabilize the performance of WiFi, mostly as an alternative backhaul for Wifi from room to room, when WiFi performance plays up. The very existence of an interoperability standard in 1905.1 suggests that this is where we are all heading. The question for MoCA to answer is whether it can survive long term, as that technology, against Homeplug and G.hn. Against HDMI and Ethernet, it has already won, based on their costs due to the need of specialist cable.

MoCA also said that what it calls its MAC rate was either better than 400 Mbps net throughput for 90% of connections, or and better than 350 Mbps for 95% of paths.

If we go back 7 or 8 years, this is precisely the number that powerline was claiming before that technology (not Homeplug, but the UPA powerline technology which was subsumed into G.hn) around 400 Mbps. But this was a PHY rate back then and real world Mac rates were far lower. In fact as early as January 2013 the G.hn support group, Homegrid admitted that its powerline version could offer speeds as high as 400 Mbps in real throughput, from a 700 Mbps PHY starting point, but that tests in China on G.hn chips showed that it tends to offer 40 Mbps to 50 Mbps across 99% of power outlets and 60 Mbps to 80 Mbps at 90% of the power outlets – so by comparison with MoCA, admittedly on Coax, there is a factor of perhaps 5 in it.

The other two candidates then for becoming a home backbone are Homeplug AV2, highly present in claims at CES that it could now produce real world performance up at the Gigabit level, and the coaxial version of G.hn, which like MoCA requires existing in-built coax and in regions where this is common, MoCA has already won the installed base, so perhaps this too is not an option.

At CES this week D-Link said it would come to market with a Broadcom based Homeplug AV2 Technology using MIMO, and to win the US home WiFi backhaul market, all it needs to do is outperform MoCA’s real world performance. We think this is very unlikely.

 

The D-Link PowerLine AV2 2000 Gigabit starter kit boasts that it is the world’s first consumer home networking system to receive AV2 MIMO certification, by using all three wires (line, neutral and ground) and says it is up to tasks such as gaming and transporting 4K video.

Broadcom’s HomePlug AV2 technology has a new physical layer, better noise resiliency and better coverage performance, using spectrum from 2MHz to 86 MHz (close to the 100 MHz of spectrum that MoCA uses, although it is rarely interference free spectrum).

At the chip launch Broadcom was saying that its new chips can provide speeds of up to 1.5 Gbps, about three times faster than older Homeplug AV chips and faster than the speeds G.hn claims for powerline. It is thought of as 500 to 600 Mbps at the Mac level. By adding the fact that this is IEEE 1905.1 compliant, this is supposed to trump MoCA in offering backhaul for unpredictable Wifi.

If we go back all the way to 2012 when MoCA 2.0 chips first came out, the companies involved have always claimed a net throughput of 400 Mbps over 16 nodes and 500 Mbps in two-node Turbo mode, so its Mac performance has translated into a real world performance as you would expect in a protected cable environment. But the powerline world is just as unpredictable as the WiFi world and 90% of outlets are unlikely to get close to that figure, especially under a high level of interference, but we shall have to wait and see if this group responds – we think they won’t bother, as the result would be too low.
So this latest testing claim is the final battle to the death of these two technologies in the US market. Outside of the US, Coaxial cable is not such a pull, so the winners there may be different. We wonder what Homeplug AV2 will come back with, but for our money it is likely to perform at something below the MoCA level in the real world, despite claims of 600 Mbps throughput. Achieving that rate is one thing, sustaining it everywhere, is another and doing it with powerline interference, another yet again.

The MoCA results come from a 108 home test throughout the US, with homeowners self-installing using both Broadcom and Entropic chips and a PC application test on the 100 MHz that MoCA uses, where that 100 MHz was centered anywhere from 500 MHz to 1.6 GHz.

Completion of the field tests is targeted for the March 2015 timeframe and this echoes a similar field test when MoCA came to market in version 1.0 back in 2005 where it verified at better than 100 Mbps net throughputs in 97% of all outlets. 100% after tuning.

MoCA now says it will conduct more tests internationally. It told us that Turbo mode has not been tested (where just two nodes talk to one another for a 20% throughput increase) but there’s no reason why it would not work. The 5% of locations that did not work, performed under 350 Mbps, and the graphs make it look like the worst ones were around 100 Mbps.

MoCA also has a bonded throughput rate, which has a PHY that is double the 700 Mbps at 1.4 Mbps, where the system uses two separate 100 MHz channels. This also was not tested, but it would achieve something close to 800 Mbps we suspect.

Close