The infrastructure requirements of IoT

You’ve probably read a lot of articles about use cases for IoT – this isn’t another one of those. TowerXchange spoke to IoT guru James Brehm about the infrastructure requirements to deploy, for example, 250,000 IoT sites in the U.S. alone. Is IoT an untapped greenfield opportunity for towercos? And if so, will towercos be willing to become a ‘service provider to service providers’ to unlock that opportunity?

TowerXchange: Please introduce yourself and your vision for the future of IoT, particularly insofar as it will impact communications infrastructure.

Almost five years ago I founded James Brehm & Associates as a boutique market research and consulting firm with a focus on IoT. Prior to this I ran the IoT / M2M practice at Compass Intelligence, having previously served as a consultant at Frost & Sullivan for nine years. I come from an industry background having worked at UUnet, Gateway and Rackspace.

As to our vision of the future; we’re at crossroads now – it’s like Moore’s Law met Melcalfe’s Law met Shannon’s Law and they all fell in love. We’re experiencing hyper growth in the number of end point connected devices, and it will only accelerate over the next four to five years, which is driving a proportionate need for expansion and densification of communication networks.

A lot of what we once called M2M, now IoT, started life as satellite then cellular, and now there’s lots of Wi-Fi out there – short range stuff. Now we’re seeing the proliferation of Low Power Wide Area Networks (LPWAN) from the likes of SigFox, Ingenu, Silver Spring Networks, and quite a few others, many of which are members of the LoRa Alliance. Through these networks we’re able to gather more intelligence, and create more intelligent solutions: from self-healing sensor solutions, to autonomous vehicles. We’ll be able to put fresher food to your table, and medicine and vaccines can cease going to waste.

This is creating a new opportunity for towercos, REITS and building companies to expand and take advantage of the new economy by deploying enabling IoT connectivity solutions of their own.

TowerXchange: Can you describe and simplify the IoT ecosystem in terms of equipment providers, network providers, the role of operators and towercos – who is whose customer?

I’d use the Israeli term a ‘balagan’ which loosely translates to a huge mess! One person’s customer is another person’s partner.

The IoT ecosystem starts with device manufacturers and ends in the cloud. It requires the participation of tower owners, REITs, building owners, wireless operators, non-traditional operators, systems integrators, and software companies. At an infrastructure level it could include the infrastructure that cities own, as a lot of these are not big cell sites. The need for deep indoor connectivity may mean smaller radios on street signs and stop lights to densify the networks.

Traditional tower and carrier economics suggest that the greatest costs have been to acquire spectrum and build infrastructure – from building and putting radios on towers, to trenching copper, fibre, and adding microwave links. We’re now pivoting – going from a time when millions of celphones were transmitting gigabytes of information and generating US$40-100 ARPU per subscriber (in developed markets), to a time when we’re increasingly now talking about a layer of billions of connected devices, transmitting bits of data, generating ARPU measured in cents not dollars. Under this new regime, current tower economics break down – IoT stakeholders cannot pay tens of thousands of dollars to place radios on towers and thousands of dollars a month to keep them operational. But since the tower networks are already operational, to the owners of those towers, IoT represents incremental revenue at very little cost.

Wireless networks have been built on the assumption that QoS would be a critical differentiator, yet many of the IoT use cases are not mission critical; the network just has to be good enough, and you can’t charge the same price points.

TowerXchange: What are the different IoT technologies, particularly insofar as they relate to the different kinds of equipment that need to be mounted on cell sites. In general, what height, weight, power consumption and ground based equipment footprint will IoT equipment have?

I was just shipped a trial device that is a similar size and weight compared to a home Wi-Fi router, could mount to a tower or directly onto a fibre cable strand to a home or business, with a single small antenna less than a meter. You could plug this into a normal household outlet and draw less power than a TV because it’s not always working – it’s only transmitting when signal hits it.

But the IoT isn’t only enabled by these small form-factors; combinations of networks will operate the IoT. Connected cars will run off traditional cellular networks: antennae that are heavy and awkward to install. Urban services will be provided from infill densification sites using small cells. Even the satellite guys, who have historically had massive base stations, are making their solutions smaller and drawing less power – everyone wants greener solutions.

Shannon’s Law is all about propagation of signal: how do we get antenna configurations right both at the device level and utilising the right technology at the tower and base station level. So for example LTE radios need high elevation and high power, whereas LoRaWAN enables the IoT over wide areas, optimally mounted 40-80ft off the ground, which means towards the bottom of a tower. LoRaWAN radios can transmit around 3km transmitting in urban, up to 20-30km in open spaces.

TowerXchange: What will the IoT network look like in terms of connected devices and site footprint, and what could it look like in the future?

The infrastructure of the IoT is less about cellphone to tower, tower to cellphone connections –IoT is about any device that can benefit from being connected to the Internet and/or to other devices. So the connectivity path could be device to device to device to enterprise router or tower.

Whereas the cellular network footprint is largely defined by where populations live, work and travel – concentrating network investments in dense urban areas, suburbs, transport networks and large public gathering places like auditoriums and stadiums; the IoT is a network of things, not of people, and those things are both stationary and mobile.

There will be no single technology underpinning IoT: there will be some Wi-Fi, Bluetooth, cellular and some LPWAN. LPWAN will leverage fixed points of presence like cell towers, rooftops and in-building. The macro tower network will remain critical to cellular, supplemented by a heterogeneous network layer on micro, pico and small cells, often mounted on street furniture or in building.

As to the current maturity of IoT networks in terms of site footprint, I did some math on the U.S. market. We’re estimating a cost of US$56mn to build a single LoRaWAN network to cover the U.S. – a fraction of the cost of the cellular network. At US$200-250 per LoRaWAN base station or gateway, we’re talking about over 250,000 sites needed in the U.S. alone. That will take years to deploy, and the network rollout will initially look like cellular – concentrating first in population centres, but expanding beyond cellular, deeper into rural areas, where there is some of the greatest need: monitoring natural resources, agricultural resources and livestock.

When the big carriers talk about offering 99% coverage, that’s population not geographical coverage, and in the small print you realise that a lot of those people have to travel to get coverage. Even in the U.S. there’s a lot of open space that needs coverage. I grew up in a small town in Iowa with spotty coverage from the nationwide carriers – only the local carrier had decent coverage – and there’s a decent size town or city every 5-7 miles in Iowa. In West Nebraska or Wyoming there can be 30-50miles between neighbours, but in the intervening space important agricultural or mining resources – assets that need to be monitored.

There’s a lot of rural network build out that needs to happen to realise the IoT, but that buildout doesn’t have to be a 200ft mast with trenched fibre optics.

TowerXchange: If there needs to be 250,000 IoT sites in the U.S. alone, how many of those sites are already deployed?

Almost none. In terms of the IoT technologies that need to be mounted on towers and rooftops, this is a greenfield opportunity for towercos. And IoT equipment is not competing with wireless antennas for the highest elevations on the towers – 40ft elevations are optimal for LoRaWAN and SigFox is probably not much different.

On a global basis there are ~70,000 LoRa gateways deployed to date, so a lot of this has to be built out.

TowerXchange: So what does the IoT opportunity look like through the lens of the towerco and that of the mobile operator?

While towercos’ core business of enabling QoS in the world of licensed spectrum need not be disrupted, in parallel they can look at how to deploy IoT solutions on the infrastructure they already have and become a service provider to service providers.

Mobile operators are among the biggest spenders of capex in the world; AT&T and Verizon spent more capex than any company in the U.S. for five years in a row – do they want to spend another US$56mn building an IoT network too? I don’t think so – I think non-traditional operators like Everynet and SigFox will do it. And, since towercos have the sites IoT needs, and critical relationships with mobile operators, they are also well suited to provide IoT networks.

While a handful of mobile operators seem to have an appetite to self-deploy IoT, the reality is that very few are rolling out. Mobile operators have dedicated themselves to the business of acquiring and commercialising licensed spectrum, leveraging cellular technologies that are very much 3GPP dictated. As such, they’re actually very conservative about new technologies, and they don’t really know how to build networks that are ‘good enough’ – for most IoT use cases ‘just good enough’ works – it’s not a competitive landscape defined by QoS like cellular.

Most operators would be more responsive to buying LoRaWAN on a wholesale or revenue share basis. Even Sprint’s IoT factory in the U.S. is really a website for small and medium sized enterprises (SMEs) to buy IoT stuff. It comes down to this: big enterprises build stuff, SMEs buy stuff. Of 6mn companies in the U.S. something like 95% are sole proprietors or zero employee companies. A further 600,000 are small business, 60,000 medium sized, and just 6,000 large enterprises. There is more opportunity picking off the SME buyers of connected devices and leaving telcos to sell leased lines and cloud services to those 6,000 large enterprises. That juxtaposition of customers causes significant confusion in IoT.

TowerXchange: To what extent do the economics of IoT require a reinvention of the traditional towerco landlord-tenant relationship – given the lower revenues in IoT, what does this tell us about how partnerships must be structured to be viable?

There is no standard model yet for IoT-towerco partnerships.

There are some trials, such as American Tower and Everynet building out a wholesale network in Brazil. But the economics of the traditional landlord-tenant model do need to change for IoT. I recently had lunch with the former President of Ingenu who was explaining that he felt one of the biggest challenges to be overcome with IoT was how to deploy fast enough onto towers, and how to get the tower rights cheap enough.

It can be a daunting prospect to re-engineer established and successful business models, and the towerco business model has been one of the most successful innovations of the last 25 years, but just like Ford need to learn to become more like Uber, GE need to go from selling jet engines to charging for horsepower, the idea of a towerco becoming an IoT service provider is a dynamic change, fraught with complexities, with uncertain consumer willingness. But if towercos can come to the right type of agreements with IoT network providers, there could be a big win in enabling IoT networks for both carriers and enterprise private networks.

Towercos must be open minded and experimental – the IoT opportunity may be much greater in the long run in revenue share partnerships than using traditional lease models.

TowerXchange: How do you see the IoT being brought to market? With whose brand will the customer have a relationship?

Right now the tower company is comfortable because they are not a service provider to millions of individual customers. Does the towerco want to become a service provider and manage 1,000 different IoT use cases? Probably not. But as a service provider to service providers, perhaps the towerco sells to the enterprise private networks and sells to the telcos, then the telcos sell the IoT network alongside their SME services.

I think the towercos are interested in anything that generates sustainable revenue at a cost that is not prohibitive.

In big cities like New York, towercos have license agreements with thousands of rooftop owners – those same agreements can be used as templates to put LoRaWAN equipment on grain elevators in rural Iowa, enabling as much as 30-50km of coverage in flat areas.