So-called 'fog computing' offers a way to connect a multitude of devices.

If you've only just come to terms with cloud computing, here comes the next thing: fog computing.

"Fog computing is the necessary next stage on from cloud computing," says Vikram Kumar, the chief executive of KotahiNet. "It is where the edge of networks become intelligent and autonomous."

Fog computing has the features of cloud computing, such as data, computing, storage and applications. But instead of concentrating resources in a data centre, fog moves them closer to where they are used. At times, it can mean putting computing resources in locations well beyond the reach of traditional networks. It turns out this is an ideal way of dealing with the Internet of Things (IoT).

In recent years, some information technology analysts have pushed IoT as the next big movement.


It promises to connect objects that, until now, have been either beyond the reach of everyday information technology, or uneconomic to connect.

For consumers, the IoT might mean fridges that can talk to microwaves. Gadgets that can order the groceries for you.

But consumer applications are nothing compared with IoT's potential to change almost every aspect of industry.

Distributed sensors can track anything capable of being measured, from the moisture falling on a paddock to the speed a truck drives down a road. That information is gold.
KotahiNet is part-way through building a New Zealand-wide wireless network to support the IoT. This month it extended the network's reach to central Auckland, Dunedin and Napier. The Wellington network opened for business in February. By June KotahiNet plans to offer services in 12 cities and regions.

Kumar says his company will build its own wireless networks only in some areas.
In others it will partner with existing regional wireless internet service providers. He says: "These guys are competent and local. They are able to do things that might not be practical if done by a remote organisation. It meant we had to be comfortable building a business that is not centrally controlled."

The key to KotahiNet's business is a technology known as LoRaWAN (Low-Power Wide-Area Network).

It's a type of wireless technology designed to connect small devices over long distances without chewing through a lot of energy in the process. Kumar says it means KotahiNet can use sensors that might use solar power or have a battery life of five to 10 years.

We can build an entire national network ... for less than, say, 2degrees would spend on covering a single city.

Wellington-based Jon Brewer is a wireless engineer and KotahiNet shareholder. He says LoRaWAN is ideal for IoT applications and overcomes many of the problems entrepreneurs have faced in the past when building IoT networks. "It works extremely well and is compatible with NZ regulations", he says.


More to the point, Brewer says LoRaWAN is the first technology of its kind to arrive along with many manufacturers supplying off-the-shelf hardware. He says this has helped to keep costs down.

LoRaWAN is also efficient. "It works better than 3G cellular for sending data in buildings. You usually get about 2.5km coverage in city areas. In Auckland we can cover a 5km radius from Sky Tower. Out of town the range goes for many kilometres. KotahiNet could cover the entire Canterbury Basin with just eight base stations. That would mean connecting tens of thousands of sensors. Nothing gets in the way of 800MHz."

One advantage of LoRaWAN is that it uses unlicensed wireless spectrum, so there is no expensive bidding for bandwidth. The 800MHz spectrum used by KotahiNet aligns with the frequencies used in Europe, which means customers here can buy mass-produced hardware at affordable prices.

Kumar says the technology means the cost of building a network is a tiny fraction of what New Zealand's telcos spent on their cellular networks. "We can build an entire national network covering even the most remote areas in the country for less than, say, 2degrees would spend on covering a single city," he says.

That low cost structure changes the network-building strategy. Kumar says a cellular provider would need to see a high level of demand before investing in extending a 3G or 4G phone network. He says KotahiNet is able to build ahead of any demand: "The costs are so low it needs a non-telco approach."

Compared with Vodafone, Spark and 2degrees, KotahiNet is a minnow. Eventually the telcos will want to reap rewards from the IoT. For now, Kumar says KotahiNet is extending the existing wireless network into new areas. The two types of network have similar infrastructure with wireless access points and backhaul to the internet.

Internet antenna installed by NZ company Kotahinet.
Internet antenna installed by NZ company Kotahinet.

Kumar says cellular networks are best for high bandwidth data, but they only get to so much of the nation. He says LoRaWAN is an ideal way of extending the network, just as Wi-Fi is a great way to extend the reach of broadband networks.

At the moment, devices connecting to cellular networks need a lot more power than LoRaWAN hardware. So much power that they either have to be connected to mains electricity or they need to draw on solar power. That's set to change. Cellular hardware companies and telcos are now working on a standard to deliver IoT-style networking from the cellular network. It is two to three years away and may arrive along with 5G cellular.

When it arrives, there will be a range of wireless services to choose from, each with its own characteristics suiting certain uses. Cellular wins where there are large amounts of data. Kumar says there will always be a need for a low-cost alternative and LoRaWAN scores best in that department.

He says these companies and the IT giants have been here before with cloud computing. "In the early days of cloud, IT vendors and telcos led the charge. They reinvented themselves to be cloud-friendly. They gained real benefits when government and business started using it", he says. Today cloud computing is all about platforms.

Kumar says they've gone down a similar route with the IoT. He says: "We're at the stage when everyone is pushing their own standards and vision". They have what he describes as "point solutions" — one-off products to do a single job. "After a while the market says that's too hard to deal with. They struggle to make various point solutions work with each other. At this point platforms emerge. Platforms are a bridge between point solutions and ability to build large, complex systems."

Kumar says the market is still at the immature stage and point solutions dominate. The move to fog computing will change that. He says the big players are only starting to scratch the surface of the IoT opportunity.


To stay competitive, KotahiNet is focused on two specific areas. First, it aims to be New Zealand specific. Kumar wants to build a network that is tailored to local needs and conditions. Second, he aims to take what he calls "New Zealand's domain expertise" and sell it to the rest of the world. That domain expertise is mainly about using technology in primary production.

He says: "Our first effort in that direction has been to work with olive growers and fruit growers. We can take data from the soil and warn the grower when frost is coming or when spring is on the way." The potential lies in packaging this expertise and selling it overseas.