Performance and power management
Chip designer ARM spotted an opportunity to get its nose (ears, wrists, head and legs) ahead of the competition in the wearable tech race, something that the Cambridge-based company has grabbed by the, er, throat.
Focusing on low-powered chips designed to keep wearables going for longer, its processors have nestled their way into a number of gadgets – from Google Glass to fitness devices such as the Fitbit, Nike Fuelband and Jawbone UP – and the popular Pebble smartwatch.
While many wearable devices have comparatively limited functionality compared to their elder smartphone cousins, there are still complex and numerous challenges for the market to overcome.
- 8 bits of wearable tech you need to see now
Noel Hurley, Deputy GM of ARM’s CPU Group, reckons that instead of housing “off-the-shelf” chips, wearables should be powered by ones engineered specifically for particular use cases and run on tailor-made software to eke out maximum performance and battery life.
We spoke to Hurley at the Wearable Technology Show 2014 to find out how the market is progressing, the challenges it needs to overcome and why every wearable tech company should be thinking about growing its platform’s ecosystem.
TechRadar Pro: Wearables are still in their infancy. In terms of aspects like performance and battery life, where are the biggest improvements going to come from in the next few years?
Noel Hurley: The market’s maturing, with a lot of companies currently putting out different wearables that are packing “off the shelf” chips, as it were.
As the market starts to mature and the volumes increase, it makes more sense to engineer chips and software specifically for wearables. That’s where I think we will get the big wins in terms of battery life and also performance and functionality.
Inevitably, off-the-shelf solutions aren’t going to be as efficient as they were designed for something completely different.
Many of the smartwatches we’ve seen so far are guilty of taking Android and putting it onto a smartphone chip (ed: all of them ARM-based) that’s been taken off the shelf, packaged up as a watch and put out onto the marketplace.
TRP: But wearables are picking up traction now – isn’t using off-the-shelf components a necessary evil for companies to beat their competitors to market?
NH: I think it’s a combination of factors. It’s partly about capturing interest, but I also don’t believe anybody has the right answer yet. Many companies are looking to experiment and put different solutions onto marketplace.
They’re playing with different technologies and testing their concepts and ideas, which is the phase I think we’re looking at now. They’re gathering too: the Google guys are getting great feedback from Glass’s concept about what works, what doesn’t work and how now to take that technology forward.
It’s the same with the smartwatch people and the fitness and lifestyle bands. They’re getting products into the marketplace and testing what the consumers like and don’t like.
Supply and demand
TRP: How well positioned is ARM to offer CPUs for different types of wearables?
NH: We’re in most wearable devices in the marketplace today and are in a reasonably good position in that we have a huge range of processor solutions in the marketplace.
People are experimenting – they’re trying anything from very simple sensors on a band all the way up to putting a smartphone on the wrist.
We span that entire gap and many solutions in-between that space too so for us, in terms of processor technology that’s available today, it’s now a case of focusing in on and really engineering to different use cases.
TRP: Does the breadth of ARM’s portfolio put the company in a strong position to meet wearable manufacturers’ needs?
NH: Yes it does – particularly when it comes to sensors in devices, which always need to be on. For example, if I’m looking at a pedometer, such as the Fitbit One, I’m trying to capture steps so it needs to be running all of the time.
The processor that sits alongside that needs to use absolutely minimal, minuscule power consumption because it’s always on.
However, once the user starts to interact and wants to see that data presented to them, then of course you fire up continually changing states using a faster processor all the way up to the super end of power consumption used to deliver a service and experience.
You’re continually going backwards and forwards between those states, and states in-between too.
TRP: How importance is the software that controls these states?
NH: It’s very important in two areas. Designing processors is an expensive business, as is designing a system-on-a-chip (SoC). I can spend tens of millions of dollars designing a new processor or SoC to get a 10 per cent reduction in power consumption.
Similarly, I can achieve that in a matter of weeks with well-crafted software. That’s what I mean about looking at software specifically for use cases.
So that’s one element of it. Another is the intelligence and software that manages use cases and which parts of the system need to be powered in what state. What performance level do I need to just show some steps on a display, for example? Or how about full web-browsing?
At ARM, we’re working on processors to fit the different use cases, and also the software and system elements around how you manage those power and performance profiles, in addition to how you manage the transitions between them. That’s a key area for us.
Android Wear and APIs
Doing the robot
TRP: Google recently unveiled its smartwatch OS, Android Wear. Why didn’t it stick with full Android?
NH: It’s encouraging that software communities and OS companies are looking at how to better engineer software for different use cases. Android Wear is an example of this.
Many of the smartwatches today are running full Android, but only driving a display of a limited resolution. As such, there are elements of Android as it is today that are too big and bloated for what you need to do, meaning there’s lots you can strip away while still delivering the experience that the consumer wants.
That bloat results in inefficiency as you burn power by running software of limited value. Or, if the software stacks are too complex, you burn power that way.
When I’ve got a use case that says “the screen size is this, these are the sensors, and this is the type of environment and performance level I need,” I can strip away a lot of complexity and try to hone it down to the most efficient software and hardware for that purpose.
TRP: Wear also deeply integrates Google’s contextual services. Do you think we’ll see more of this kind of thing?
NH: I don’t have have any specific comment on it as it was only recently announced, but certainly a large part of wearables is about augmenting reality – that’s Google Glass’s whole concept. Similarly, Android Wear is tying together data and cloud services alongside a hardware platform on the body.
TRP: Apps will likely drive adoption of smartwatches, but how important will APIs be in allowing developers to create ecosystems around them?
NH: I think if you’re thinking of launching a wearable device you really need to be thinking about not just the hardware, but also how you’re going to deliver and build an ecosystem around it.
It’s about how you’re going to encourage developers to write apps for your hardware, because I think you’ll have a better chance of success by engaging with the many creative software developers out there, and that comes back to APIs. Then there’s the tools and the framework.
It almost harks to the analogies in which the way games consoles came to market in that you launch an initial application and then over time you want developers to bring out new applications for your platform.
When you look at things like Pebble, they’ve done that very well. I think others that have tried to make it very closed are missing an opportunity to engage with developers and consumers. They should aim to build a level of enthusiasm around the platform.
TRP: If a platform takes off with developers, it can have pretty powerful results…
NH: Absolutely. A few years ago there was a fantastic device that was ingenious in that all its creators did was take a ball with a motor and a Bluetooth radio chip in it and a few coloured LEDs, before opening up an API for it.
It’s amazing how many developers wrote iPhone or Android apps just to play games by controlling the ball. allowing it to do things like move around a maze.
From a hardware perspective it was really simple. Just by writing an API they could change the colour of the ball and write games based upon how you control it. They could’ve just done a ball that came with one app, but by opening up the API they were able to have lots of games revolve around a simple thing.
TRP: In that sense, will putting more sensors in wearables allow developers to do more with them?
NH: Yes – I’m certainly in the camp or more rather than fewer sensors as I think that will foster creation and innovation. Couple that with open APIs and app stores and i think you’ll get the creativity, which will keep the interest of the consumer.
Design and business potential
Design for life
TRP: Can the design and physical shape of wearables affect the components that can be used inside? Most smartwatches have been square up until the Moto 360, the first circular one by a ‘major’ manufacturer to attract attention.
NH: There is an interplay, and that’s not necessarily on the processor side but certainly on the graphics side. Most screens today for the smartphone business are rectangular and of a particular size.
I think that with wearables, we’re going to see some more exotic shapes, more exocitic aspect ratios, curved flexible displays and types of display technology. Of course, that then has an effect on both the graphics processing and the associated software libraries.
Because now if we’re looking at software resources, I’m now not dealing with an oblong shape, I’m dealing with different types of shapes, so my software now has to layout the data on different aspect ratios with different display screen technolgies.
So it does have an effect, but it’s more on the graphics side. The processor doesn’t care quite so much.
TRP: Has increased trajectory around wearables impacted ARM’s time to market for its Cortex-M series CPUs?
NH: In terms of where we are today, we have a very broad processor portfolio all the way from Cortex-M0 to Cortex-M0+ processors to ones fractions of a grain of sand. They span all the way up to the latest 64-bit tablet processors. We have a good range in-between too.
There are other elements looking forward, such as security and looking at how you secure the data, which is a key aspect to this marketplace. Do I see an increased cadence? I think there’s a natural increase cadence, but wearables has a way to go until it gets as frantic as smartphones.
Everyone’s bringing out new phones every six to nine months, but I don’t expect wearables will change rapidly. We’re confident we’ve got the entire processor range to be able to satisfy that change for a generation or two.
TRP: Let’s switch angle – what potential do wearables hold for the healthcare industry? You said that blood pressure sensors are difficult to implement.
NH: Blood pressure has been a bit elusive. Sometimes in the medical field there are regulated areas for taking things like heart rates and so on – it’s currently measured with a big cuff. if you could have a more unobtrusive way of measuring blood pressure, I think consumers as patients would go for it.
There’s also an opportunity to better monitor patients using wearables, which means you can get patients out of hospitals and into homes, which has an advantage for the medical industry because it’s expensive keeping people in hospitals.
Also, patients that spend time in hospital tend to say that the food’s rubbish, they can’t sleep and want to go home.
Check out our coverage of Qualcomm’s Life at MWC 2014.
TRP: How quickly will wearables take off at the business end generally?
NH: You see it many times where people play around with technology in the consumer environment and then industry picks up on it. With Glass-like technology, for example, you can wander around and do stock takes hands free, that sort of thing.
I think it’ll go down that route. When looking at other routes, it goes beyond wearables. Sensors in cars, for example, can affect your insurance depending on how you drive. There are lots of scenarios that will unfold over time.