Your browser is not supported. Please update it.

5 June 2015

Ericsson’s IoT math doesn’t add up, backtracking doesn’t help

We often take issue with the headline figures found in forecasts and reports, and this week is no exception. Ericsson has released its latest Mobility Report, and has revised its total connected devices number the wrong way – scaling it down from its 2009 prediction of 50 billion devices by 2020, to today’s 26 billion by 2020 forecast.

Saying that the 2009 Networked Society figure was more a vision than a concrete forecast, we think that the 50 billion figure is already on the conservative side of a prediction, and that to revise a forecast down to half this number shows a prominent misunderstanding of the market – in combination with the bias inherent in Ericsson’s vested cellular interest.

A statement in response to this revision sounds rather out of place for an industry giant, saying “this is the first time we published a connected devices forecast based on thorough analysis. The M2M area has been developing over several years, and while we now communicate our current view of the development in this forecast, the area is still new and much can change.”

You’d have thought Ericsson would have kept its ear closer to the ground, but as its logic shows, this isn’t the case – unless it’s obfuscating the definition of “connected device” for some unknown purpose. If it wanted to sell more networking equipment, you’d expect it to inflate the figure – a tactic that it looked like Cisco tried about a year ago, before the rest of the industry engaged in some rather aggressive one-upmanship.

Cisco’s 50 billion devices by 2020 forecast was the one that created the most buzz in the IoT business, and since then a number of companies have surpassed that figure. ZTE predicts 100 billion as early as 2020, and fellow Chinese vendor Huawei has said 100 billion by 2025. The highest we’ve seen was 300 billion.

So Ericsson moving the figures down seems counterintuitive to our cynical outlook, but nonetheless, it has halved its forecast. Key to the current figures is the growth in mobile. Ericsson says there 7.1 billion worldwide mobile subscriptions in 2014, and that by 2020 this will reach 9.2 billion, at a CAGR of 5%. Global smartphone penetration is pegged to reach 70% in 2020, with mobile broadband networks covering 90% of the world (although whether that is population coverage or territorial coverage is unclear).

Of the total, smartphone subs comprised 2.6 billion in 2014, and are expected to reach 6.1 billion by 2020. By that year, Ericsson says 80% of mobile data will be consumed by smartphones, with 60% of all total mobile data traffic set to be video by 2020.

Africa, the Middle East and APAC are expected to drive this growth, with Ericsson predicting that 80% of new additions in 2020 will originate from these regions – with their spending power finally catching up with their share of the global population.

Total mobile data traffic is set to peak at 140 Exabytes per month in 2020, up from 50 Exabytes per month in 2014. Total mobile broadband subscriptions are growing at 30% year on year, and grew 150 million in Q1 2015, and will hit 7.7 billion by 2020. Ericsson thinks that 5G plans will be available in 2020, and that uptake will be much faster than 4G.

The smart home and the IoT:

“The connected home is driving connectivity in consumer electronics – mostly over WiFi and Ethernet. In total, 26 billion connected devices are expected by 2020, of which about 15 billion will be phones, tablets, and PCs. The total excludes passive sensors and RFID tags.”

Well, here we see the reasoning behind the downwards revision, but it’s an arbitrary decision that doesn’t reflect the nature of the IoT. Based on that definition alone, Ericsson is choosing to ignore the largest sector of the IoT in terms of both device shipment volumes and potential impact – pervasive sensor technology.

The accepted definition of a passive sensor is a device that detects and responds to an environmental input, such as light, noise, movement, vibration, or temperature. Active sensors, on the other hand, emit either light or radio waves and analyze their return to come up with a reading. There is some overlap in applications between active and passive, but both will deployed in their billions in the IoT – and that’s something that Ericsson is apparently choosing to ignore.

One of the key focuses of most IoT initiatives is adding sensors like these to environments or deployments, in order to gather operational data that can then be used to increase the efficiency of the task at hand – whether its smart city traffic management, industrial manufacturing, or water quality control.

But before we tackle these IoT devices, we can see that there’s an issue with the 15 billion phones, tablets and PCs that Ericsson forecasts will exist by 2020. It’s really not unreasonable to think that the average home in a developed economy would contain significantly more devices than the Ericsson prediction, and some rough calculations will show that the number of the other 11 billion connected devices in that forecast should be much higher – considering that we now live in an age of connected white goods.

Some very back-of-the-napkin numbers will quickly show the problem with the math. Using the roughly 200 million homes in Europe, and the 750 million people who live there, just ten connected device per home would very quickly give us 2 billion devices. If it was 10 devices per person, then suddenly we’re up at 7.5 billion, or if we take 100 devices per home, then we’re on 20 billion extra devices for Europe alone, never mind the Americas, APAC and eventually Africa – with the bulk of these devices not being counted in the Ericsson figures.

And 100 devices per home really isn’t unreasonable. If operators finally start providing a smart home as a service offering, which would help reduce churn and increase ARPU, then the home would contain the following: 1 coordinator, likely in the form of a set top or gateway; 3-12 security cameras; 1-2 thermostats; 6-12 auxiliary temperature sensors for the HVAC system; 6-12 zoned heating/cooling unit controls; 6-36 connected lightbulbs, with 1 wall-mounted switch per room; 6-12 speakers and intercoms for whole-home audio; a garage door opener; 1-2 video doorbells; perhaps even automated blinds and curtains; motion sensors in every room and walkway to passively track occupant movements – the list can go on.

Each of those devices has the potential of generating new revenue for pay TV or broadband providers, whether they control the whole ecosystem and make a margin on every product they sell, or whether they offer an open approach that allows users to add their own devices using the leased CPE coordinating unit and a small starter kit.

It’s worth clarifying here that the proliferation of these devices will actually have a fairly negligible impact on the average data consumption of a connected home – as long as the device doesn’t use video. Sensors and actuators deal in messages measured in bits and bytes, not megabytes or gigabytes. The traffic they generate might also never reach the public internet, and will likely remain inside the home network, coordinated by the central controller hub.

That hub will obviously generate some outgoing traffic, in the form of communications with the home owner and the operator, but the biggest potential rise in total data consumption will be found in security cameras routinely uploading their recordings to secure cloud storage, and from home owners accessing live or recorded video feeds to check on the status of their homes.

And that’s just smart home devices. You can begin to imagine if we started including every HVAC system in the forecast, or every piece of city-owned street furniture, connected vehicles, or entire industrial processes – the figures would swell.