5 Pitfalls That May Kill The IoT

A couple of weeks ago I participated in a panel titled “The Road to a Trillion Devices” organized by Brian Fuller at Arm TechCon. His closing question was whether we will get to the projected trillion devices in twenty years. My answer was that we may even get there faster. His opening question was what the pitfalls would be to make it difficult to get to trillion devices in the next twenty years – I cited five, namely “security,” “ease of design,” “architecture,” “5G networks” and “mindset.” Here is why I am bullish about those trillion devices and Cadence’s enabling technologies.

Here’s the background of the discussion. At Arm Techcon 2016, SoftBank CEO Masayoshi Son predicted a future with “1 trillion connected devices in the next 20 years,” so that would be by 2036. I had mused here, and here and in a video here whether a trillion devices would actually be possible. To me, it may even be more – given how many devices I already own today and given that I have always been fascinated by exponential growth and Kurzweil’s singularity. Masayoshi Son had thrown out another trillion earlier this year at Mobile World Congress by saying that just Arm itself would deliver a trillion chips for the IoT in the next 20 years.

So what could go wrong? And why am I, for one, bullish on this area despite the hurdles? Here are the five issues and why I am confident that we will figure it out.

Number 1: Security
Given the number of breaches we are experiencing, security is certainly the biggest hurdle. If we as an industry cannot make these devices secure, users will always be hesitant to adopt them as they may pose threats that impacts our credit, our identity, or could even be life threatening if somebody else can hack into a health device. We will figure this out because of CEOs like Simon Segars, who gave a call to action during his TechCon keynote to put hackers out of business. Arm’s security manifesto, together with architecture enhancements in hardware and software, will lead the way here. And industries like ours are embracing security 100%. Cadence, for one, is contributing to this area directly with our tools in the formal world, and we are partnering as an industry too. For instance, I just contributed to a white paper led by Tortuga Logic that centers around security and emulation.

Check. The industry as a whole is committed. We will figure this out.

Number 2: Ease of Design
Chip design is a complex matter. Lots of IP re-use, changing technology nodes, complex verification – you name it. In order to get to a trillion devices, ease of design is crucial. Again, we are well on our way here. For example, the “Design Start Initiative” in which Cadence is participating with hosted design services (think “Cloud” for our EDA tools), spearheaded by Arm, lowers the entry to design. It is a game of ecosystems after all – and other architectures like MIPS, x86 and RISC-V are also lowering the hurdles to design entry in many ways, including open source architectures. And the tools start working together across company boundaries to ease design entry too, like Arm’s Socrates tools with Cadence Verification tools as shown at DAC and Arm TechCon.

Check. Not done yet but we are charging into the right direction.

Number 3: Architecture
The system architecture to carry all the data in an era of a trillion devices is changing rapidly. It’s a challenge and we need to figure it out. Just sending all the data into the cloud and processing it there is not possible for simple reasons like the energy it would take to carry the data. So processing will happen along the way – at the sensor, at the aggregating nodes, at the edge of networking boundaries, etc. To me, being an engineer at heart, I am convinced that we will figure it out. Really smart people like Chris Rowen are working on this – I still remember his brilliant keynote at CDNLive on this topic called “Follow the Data” a while back. Process data as you go.

Check. Smart engineers will find the right architecture.

Number 4: 5G Networks
All of these devices need to be connected. 5G holds the promise of directly connecting devices to the network and cutting out the aggregating component (like your cell phone) in the middle. Worldwide, we are looking at 100s of billions in investment to make these networks happen. The 2020 Tokyo Olympics is the compelling event for broader roll-out. We will get there. Given that I am still struggling with dropping phone calls in some areas of Silicon Valley (on 237 right off Great America no less), timing may become an issue here. But I have no doubt that we will get there.

Check. Timing may be different in different areas if the world, but we will get there.

Finally, Number 5: Mindset
Mindset changes will have to happen on both the developer’s side as well as the user’s side. The latter will be determined by application domains and time itself, and is certainly connected to the security aspects. As soon as convenience or necessity outweighs the risks and concerns, adoption will happen. For the generation ahead of me, my dad is always my litmus test. Wearing a Cochlear implant for hearing has become a necessity and he probably doesn’t even consider it an IoT device. Pacemakers and other health IoT devices may become necessity for some. In other cases, like banking, the convenience of not having to go to the bank outweighs some of the risk considerations. Again, for the generation ahead of me, my dad does all his banking and emailing on his iPad – in that sense, an IoT device that is networked. Looking at the generation following me is even more fascinating. It seems that for my millennial son to my 13-year-old daughter, every device they have is connected and no longer stationary. Both data and the device to access them are not bound to any location – laptop, iPad, phone – it all moves around and even the TV becomes just an occasional stationary item they look at as all inputs come though mobile, connected devices.

Mindset is also an issue on the development side, mostly when it comes to the monetization aspects. Where does the money actually come from? The value moves to the system, so the semiconductor cost for an IoT sensor or other device may actually not be exposed to the end user, and its cost may be cross compensated from ads, health advice, etc.

Check. Mindsets are changing. Timing may be impacted by different application domains.

So will we get to the trillion? Absolutely! Will it take 20 years? Perhaps. I am bullish. It may be faster than all of us think.