Second of three parts: The need for multidisciplinary approaches; lots of unknowns about what the IoT will become; how the general public perceives the IoT; the need for security; privacy issues.
Semiconductor Engineering sat down to discuss the future of the IoT with Oleg Logvinov, director of market development for STMicroelectronics’ Industrial and Power Conversion Division; Martin Lund, senior vice president of the IP Group at Cadence; Naveed Sherwani, president and CEO of Open-Silicon; and Damon Hernandez, a member of the Web3D Consortium. What follows are excerpts of that conversation.
SE: The amount of data that needs to be stored is astounding, even if it’s only for a couple weeks. How do we deal with that?
Hernandez: Even in the building industry, between that side and the geospatial side, the real world is being digitized right now. The real world 3D is happening as they solve their own data interoperability. All the smart systems that go into buildings and into shoes that we talk to, that building needs to understand who’s in there. Maybe the person likes the building at 70 degrees or 60 degrees. Then you can make better systems. It’s not just one thing. It’s an integrated system and we need to make sense of it. That’s where the power of being able to see data and then interact with it—the simulations beyond that—is a challenge. How do you interact with that data and get better questions out of that data.
SE: What does the semiconductor industry have to do to make the IoT real? We’ve talked about reducing power and cost for 20 years, but what do we need to have sensors everywhere?
Logvinov: Power reduction is important, integration is important and footprint is important. All of those things we know well today. But what we have overlooked in many areas is who is going to be successful going forward in providing solutions. It’s not enough to push the limits of power consumption and cost and all of those singular trends. The most successful ones will be companies capable that can sense multiple facets of the Internet of Things. If you think about an intelligent thing, it’s a smart phone shrunk to the size of a grain of rice. It’s a smart thing with processing power inside and sensors capable of sensing multiple things including gas composition in the environment. It has to have the ability to communicate constantly and even visualize the information you have. So if you think about the number of disciplines that you have to become proficient in, it’s not only semiconductor process technology. It’s also thinking about how you have to create the system, the software layers you have to put on top of it and the information needed to make sense out of it—whether it’s communicated to another machine or device. That multidisciplinary approach is going to take over.
Sherwani: It has to be all of those things, but what we have been good at in semiconductor design over the past 30 years is a huge amount of overdesign. With the Internet of Things, you cannot do that. If you look at microprocessors and network processors, they deal with a lot of silicon gates going to some obscure cases that rarely or never happen. In the case of the Internet of Things we have to think differently. We have to start with a use case and then design for the use case and nothing else, because every gate is sucking power. That’s something we’re not good at. What we are good at is building RTL on RTL on RTL for something that happened in 1991 on top of RTL that deals with something that happened in 1977 and still sits in that RTL. We have to learn is optimal design, not overdesign.
Lund: I disagree a little bit because to some extent it comes down to thresholds. When you pass a certain threshold that says you can run on a battery for 10 years, you don’t care if you can run on a battery for 11 years. When you pass the threshold, it doesn’t matter. Gates are free. If you take a wafer and you deposit 100 billion wafers on it and add one gram of material to it, it’s essentially free. The power isn’t free but the gate is. But there’s another aspect of this. Is it application driven? Will the semiconductor drive the Internet of Things or is it human ingenuity that will come up with some problems that are worthwhile solving? But if you step back and think about where we are right now, the Internet of Things means a Nest controller—an intelligent thermostat. It’s a cool concept, but it’s not rocket science. We’re just scratching the surface. It’s like being back in the 1980s saying the Internet is going to be great, but what is it? People were saying that ISDN would be bigger than IP (Internet Protocol) back then. We can’t predict it all. The semiconductor enables it because it’s almost free and can be done in high volume. But how we connect it all together and allow these things to function is critical.
SE: The only thing we’ve heard about the IoT is that we can program our oven or toaster from a distance. What does the smart home really look like in the future?
Hernandez: Going into the home is a sensitive area because of privacy issues. When you talk about things like Nest, there’s a very high geek value. It’s great, but what is the ROI. What is the sensor cost? Already most houses come in over budget and over time, but adding all these things may not be essential. Do you care if your toaster sends you a message that the toast is done or whether it just pops up like it has for the past 100 years? When it comes to homes it has to be affordable, it has to add relevance to people’s lives about why they’re going to network that. If you think of the game Watchdog, this is what the majority of people think the Internet of Things will look like. In the game, you play a cyber hacker who is able to access the city operating system or Internet of Things to find out information about what people do and don’t do, and then essentially go after them like Batman. That’s what the public is thinking of the Internet of Things right now. We have to combat a lot of security issues. We also need to have home repair manuals so people can repair stuff on their own homes.
SE: When everything is connected, everything is connected to you—which is what creates a security problem. How do we solve that without increasing the cost, adding to the amount of processing required, invading people’s privacy and still get the job done?
Logvinov: When we talk about security, we put too many meanings into it. Security means access controls and being able to reject or accept a device joining the system. When we talk about the Internet of Things, we really talk about the privacy of the data. That is a multi-layer filtering of what you can and cannot access. When we talk about encryption of the data and optical cables, that’s one area. Security in the sense of access or no access, becoming or not becoming part of the network, and then privacy of the data protection, that’s where the complexity comes into play. But even if we create a system that has all of those elements, what if someone gets hold of a device, decodes it and then understands how the system operates. Whatever it uses in terms of passwords and encryption keys has to be hidden from intrusion and not detectable from the outside. Those are multiple dimensions of what we call security. When we start talking about exposing our data, that’s an issue. In Germany there is a huge outpouring against smart meters because those meters provide real-time measurement of your power consumption. It’s too revealing about your lifestyle.
Lund: There’s another version of that, which is the nuclear centrifuge that stopped working several years ago. Even the most secure network today that is not connected is vulnerable to attack. How about living on the East Coast and there is a winter storm and someone turns off the heat for 1 million people? That’s a real security problem. When we talk about the Internet of Things it’s not just about data privacy. It’s also about real security. If someone hacks your car and turns off the brake system, that’s a whole different level. We haven’t seen the first attempts of that, but there are studies that show you can hack a car through the CD player. You can do that with planes, too. But on a happier note, with the Internet of Things, people are willing to take on some risk if the value is there.
Logvinov: That’s a fundamental point. With paper money, people developed counterfeit bills. We developed plastic (credit/debit cards) and people figured out a way to misuse it. Now we’re switching to online transactions and people are finding ways to hack into our accounts. Even when we make it more secure, someone finds a way in. But if the value is there, people will buy into it.
SE: How much of security is on the semiconductor side, and how much is on the software side?
Sherwani: We see only the hardware aspects of encryption, and that’s really expensive. Many of the security protocols are expensive in terms of power, not in terms of real estate. Real estate, as Martin said, is almost free. Every gate is burning power, and many security protocols take a lot of gates and burn a lot of power. So how do we come up with cheaper security protocols?
Lund: Software security is a big deal, though. We’ve been living with this stuff for years without knowing it. Your set-top box at home is a very secure device because people don’t want you to hack the account.
To read part one, click here.