Source: Kane Mumford, Policy Tracker
A Facebook employee has filed a patent stating that the firm plans to explore a “next generation data system” to connect devices using millimetre wave (mmWave) radio links as a mesh network. This has been tried before, but could the social media group have found the missing link for rolling it out to rural areas?
The patent’s abstract, filed in the US by Facebook engineer Sanjai Kohli, said the technology “leverages the ‘cloud’ for data management, frequency data computation and analytics. The wireless network is a single frequency network that permits limited non-line-of-sight operation”.
Perhaps Facebook has found something that has eluded previous designers. The use of mmWaves in mesh networks is not entirely new. US startup Starry is the first commercial outfit to offer line-of-sight mmWaves over a mesh network in built up areas in the US, while Radiant Networks, a now defunct UK company, was testing the technology for use in rural parts of the UK more than 10 years ago, including with BT in the Welsh valleys.
“It was a mechanically steerable antenna and there were four of those that could independently rotate through 360 degrees to form a mesh to other nodes,” said Paul Grant, operations director at spectrum management specialist ATDI and former Radiant Networks employee. “It worked, but it was an expensive bit of equipment. It was used at 28 GHz and there were some early prototypes at 5 GHz. There were trials in Wales with BT, but these things always need a following wind to get off the ground. BT were looking at it in a valley area in Wales, bits that weren’t likely to come up for DSL rollout. So it has been tried before.”
Grant would need to see the full details of the patent to understand what Facebook is trying to do, said Grant. “But generally, you don’t need anything too novel to file a patent. It doesn’t have to be cutting edge.”
Radiant used a 16-segment fixed array that switched between antennas to select the appropriate one, which could be similar in design to that described in the patent. The mention of operating on a packet basis is perhaps where the innovation comes in, said Grant. “They could be selecting something intelligently about where the packet is destined for and doing that perhaps on an electrically switched, per-packet basis, maybe a bit similar to 4G technology.”
The challenge is that mmWave is essentially always line-of-sight. “You’re driven by the need for that. So with any mesh network using those kinds of mmWaves, which I would class as 28 GHz or above, it’s really line-of-sight.”
Nevertheless, Kohli’s patent’s mention of a “limited non-line-of-sight operation” might prove to be the missing link that has eluded previous designers in seeking to make a viable mmWave mesh network, Grant said.
Use in the developing world
Most of the useful energy in mmWaves is within a geometric construct called a Fresnel zone, a cigar-shaped ellipse that begins to shrink and become sensitive to any obstruction as frequency increases. That may be one of the reasons Facebook is looking to mesh networks.
Kohli’s patent shows the company is developing the technology for its non-profit internet.org arm, which aims to connect rural and remote communities in the developing world. William Webb, chief executive of Internet of Things standards body Weightless SIG, said lower frequencies have traditionally been seen as more suitable for reaching remote communities. “In the past, you would get to rural communities with low frequencies just because they go further and we’ve seen that with the proposals to use TV white space and so on. Essentially the idea there is you can put up one base station and because its range might be many, or even tens of kilometres, then you can pick up quite a few rural communities in the radius of the base station and that makes it worthwhile.”
However, the range of mmWaves makes for extremely low coverage areas unless there is a regular and relatively dense network of nodes, which raises questions about whether such high-frequency, short-ranged waves can be economically viable. Webb said a mesh network is one potential answer, but that there are still many challenges to be overcome.
“You still have a relatively small number of base stations, but what you rely on is bouncing the signal from one village to another and then another, then back to the base station. You might not need more base stations, but you get the same level of coverage. But it’s very tough to make that work and I’ve never come across anyone who has.”
Mesh networks work in an ad hoc way, with transmitters locating a neighbour and bouncing signals from one to another. They typically operate at low frequencies where omnidirectional antennas are available. At present mmWaves typically only use unidirectional antennas in order to travel viable distances.•