There is a whole new science about to be born, in and around WiFi, which we at Faultline have referred to numerous times in the past, but this week we found the first mention of it out there in the industry, at ASSIA the broadband performance specialist.
It came up in an ASSIA blog from the CEO and founder, whom the company refers to as the inventor of broadband, with some cause – John Cioffi. He talked about collisions in WiFi and ways to alleviate them, but he also talked about cooperation between competing WiFi Access Points, to better use unlicensed spectrum. The entire blog humorously focuses on whether or not we need fiber to the smart watch.
We agree with Cioffi that there is a problem. And in no time at all, perhaps 2 to 3 years, this could become a debilitating problem for WiFi.
The collision protocols of WiFi are between the end user clients and the Access Point itself, and if clients are frequently uploading content, then the Access Point itself has to stop and wait (a random period of time) before it continues its workload. The more video that an Access Pont has to deal with, the more it wants to behave as a one way passive pipe, but the rise in cloud apps and social media is unlikely to allow this.
Also if one of the clients in a home is a DVR, then there can be two continuously sending stations – the internet pipe to the home AND the DVR, if these are not the same device. They would interrupt each other constantly, or rather be waiting for a chance to speak constantly.
Perhaps both of these need to be an Access Point in their own right? But that would need another layer of complexity to manage it.
Cioffi says that the answer is to have some form of policy manager in charge of each of the Access Points in and around a home, which manages this contention, and this is not the first time we have come across this. StreamBoost at Qualcomm Atheros does precisely this.
But a policy manager to simply manage one Access Point only gains you more granular levels of quality of Service. Cioffi is talking about managing multiple WiFi resources to achieve one work load.
He says, “For instance, an apartment complex of 20 units all with 100 Mbps fiber or VDSL service has a 2 Gbps data rate, which can be shared if WiFi systems have some level of smart policy underlying their usage. Essentially, such policy would reverse the contention issues to create an efficiency advantage.
What he is suggesting is that instead of perhaps bonding multiple fixed lines to bring more bandwidth to a home, you bond the signals when they get there, in the airwaves using WiFi.
Given his past in inventing Vectoring and cross-talk removal you can see where he is going, and it’s even more complex than he lets on. If you have a complex highly video centric WiFi environment, with two or three TVs getting UHD streams and little Johnny in his bedroom also getting high density video across the same network, it might break. But there may be resources on next door’s network that can cope with the delivery of “some” of those steams, or even some of those packets inside some of those streams.
This is far more common than we imagine. When a remote WiFi client is reached through too many walls and ceilings, it can slow the delivery of packets to that client right down. As a result, this can bring down the entire Access Point. We call this the bad apple effect. One way of serving that bad apple might be to send to a home gateway or Access Point that is not so busy, and have it deliver that client’s signal. This could be determined at a nearby cloud processor which knows about all the difficulties of all the Access Points, and spreads the load accordingly.
If you have a neighbor who is out and who’s home gateway is on, AND you have his agreement for the operator to run a second SSID, then an automated policy might deliver the remote stream through the neighbors Access Point. The WiFi on his home would automatically use beamforming or diversity to get accurate data to the neighboring client in the best condition it could. This could lead to a situation where instead of each hoe having their own Home Gateway, what they do is sign an SLA to be delivered a certain amount of broadband, and allow for a Home Gateway to be installed on their premises. As long as they get their allocated SLA bandwidth they are happy.
No-one would be any the wiser, and an operator who deployed this type of policy might be able to reduce help desk calls, reduce lost packets and improve customer satisfaction and network resilience.
When your home gateway has a fatal problem, instead of being cut off, your entire home broadband operation could be run from a combination of nearby neighbors’ Access Points. Hence the resilience.
At Faultline we have often said that once you get TV delivered over broadband, and move to highly automated homes through IoT technology, then if your broadband fails, you could not work, watch TV, or even turn the lights on. In that scenario broadband failure becomes catastrophic and a separate secondary broadband route to each device needs to be in place for emergencies.
Cioffi goes further and talks about not sending quite so much broadband traffic upstream to the internet, if the DSL line connection is not fast enough to carry it.
In his mostly telco world he is thinking about keeping WiFi and VDSL in alignment with one another, and preventing failure across the two systems.
There are many more things that can happen in a home network, such as squeezing adaptive bit rates deliberately to ensure that multiple streams of video do not break either the capacity of the broadband line which brings them into the home or the capacity of its WiFi to redistribute video around a home.
Enterprises have used expensive internal switches to schedule WiFi to ensure it was not overpowered by the contention that Cioffi refers to. He points to ASSIA’s CloudCheck as one candidate for making sure you are working on the right connection. However CloudCheck is a very small step, simply scanning WiFi to tell you the strength of the various WiFi signals around you at the same time as telling you their VDSL backhaul speeds. It is perhaps the tip of the iceberg in what we see as a long journey to smooth the path of WiFi and broadband within the home.
But he gives us some clues where his next steps may be, mentioning US libraries where typical broadband speeds per user are down at the 1 Mbps level, and says this can be similar in schools. If he puts an automation tool in front of CloudCheck to switch between different WiFi services which use different, less saturated broadband lines, and do this automatically based on their saturation, he is partway to a new broadband tool for WiFi connections that serve large numbers of people – convention centers, airports, hotels. Given that ASSIA manages broadband for customers like AT&T, it is a small step to extend this type of tool to a management tool for AT&T, and he suggests, even regulators, although at the moment it is an iTunes downloaded tool to empower consumers.