MoCA shows off WiFi Backhaul credentials in series of home trials

MoCA’s very existence has been at stake for some time, and for year’s now we have been telling the guys over at MoCA that acting as a backbone for WiFi is the best outcome it can have. Now it has released some interesting research which suggests that it is pursuing that approach.

However, it is a limited piece of research, based on testing WiFi which backhauls itself, with WiFi which has a dedicated one to one connection channel, and WiFi Access Points connected over a bonded MoCA 2.0 connection.

Many operators today are looking at using WiFi purely as their sole means of video connection around the home, and as a result the idea of having multiple Access Points has become very popular with pay TV operators, either in a mesh or an extender configuration.

As that trend has increased, the MoCA opportunity has also increased. In Europe there has been a tendency for operators to use HomePlug for this, but as we approach the UHD era, HomePlug, even the latest HomePlug AVC 2.0, which will never be replaced, because the organization has now been dismantled, is not quite man enough for the job, in our estimation.

The tests were carried out by a company called Dekra in 11 homes around the US in places like Georgia, New Jersey, Virginia, Pennsylvania, Colorado and California. It tried three different architectures – which used 802.11ac as what it calls fronthaul, but which we are happy to call backhaul. The devices were the Netgear Orbi, Plume and Eero, and these were compared with the Actiontec WiFi extenders which integrate MoCA 2.0 bonded backhaul.

The results are in the graph below and are the WiFi performance measured up and down in TCP and UDP traffic tests. Here is one graph below, but they are all pretty much the same.

It’s a dramatic win for MoCA being used as the backhaul, compared to a WiFi-based backhaul only.

The Orbi seem to do quite well, but its performance fell apart as more devices were attached to it. This is perhaps more to do with the final step of WiFi not being Wave 2 devices with Multi-User-MIMO.

The way the Orbi works is that it has two 5 GHz radios working in separate channels – a bit greedy if you happen to have close neighbors as this could kill off any chance they have of getting a spare 5 GHz channel. It uses one 5 GHz channel as backhaul and sets up one box at your home gateway and a second Access Point in the center of your living room. The two use a mesh to speak to one another, and it was launched claiming to be the first mesh to use the QTI tri-radio 802.11ac chipset, although we have seen similar set ups overseas.

So no wonder it outperforms the other WiFi devices, and it must cost more as it uses a lot more silicon and as we said, it would not work at all in a high rise apartment with lots of close neighbors.

Of course the wired MoCA 2.0 performs at spec because it is wired, and is excellent backhaul for WiFi, if you happen to have it installed. It is cheap to install as fixed wires go, but operators are still unhappy making that decision to have MoCA onboard in every device.

This is why those operators who are conducting experiments in multi AP WiFi have begun adding a layer of software control, so that bad apples and sticky clients created by persistent beamforming, can have user control to move each client to the most appropriate AP in sight, not the first one that it connects to in a house. This client steering approach will give most MU-MIMO APs the same kind of performance as those in the graph, where the WiFi chip is good enough and there is capacity on the 5 GHz backhaul. Certainly it would be closer in performance terms than these retail units. Also right now operators are only looking for 100 Mbps in each part of every room, although soon their requirements will shift to beyond 200 Mbps everywhere in a home.

But today, from the store, the ActionTec shows that MoCA continues to have a lifeline for continued dominance of the US scene primarily, and to a lesser extent Europe and China, as the backhaul of choice for WiFi, especially given that there is an existing extender architecture in IEEE1905.1 which is a standard.

MoCA said that the Bonded extenders delivered 800 Mbps downlink in all homes. Orbi achieved just 300 Mbps or better in 50% of homes though some homes were capable of only 170 Mbps while Eero and Plume were unable to reach 200 Mbps at all.

Nine WiFi clients were deployed throughout each test house. Client location remained the same for all five system tests.  For each system, Dekra turned off all clients, setup and enabled new APs, and then turned on all clients. Clients were not manually connected to a particular AP.

Traffic types were TCP and UDP, downlink and uplink. Traffic was sent to all clients individually in 60-second intervals in one test. In another test, traffic was sent to all nine endpoints simultaneously – most closely representing real world usage.