By Ed Sperling
System-Level Design sat down to discuss the Internet of Things with Jack Guedj, president and CEO of Tensilica; John Heinlein, vice president of marketing for the physical IP division of ARM; Kamran Izadi, director of sourcing and supplier management at Cisco; and Oleg Logvinov, director of market development for STMicroelectronics’ Industrial and Power Conversion Division. What follows are excerpts of that conversation, which was held in front of a live audience at the recent GSA Silicon Summit.
SLD: Who’s responsible when something goes wrong on the Internet of Things?
Logvinov: It will hinge on business models, and it depends on how business models evolve. One size doesn’t fit all. Smart-home applications are being pushed very heavily by service providers. But in Europe, where consumers want to push this market forward, they don’t want to give much data to service providers. So it depends on the market, on specific products, and on the service. The models will evolve.
Izadi: Ultimately, the buck should stop with the service provider who provides a service to the end customer. But they need to work with their partners to make sure it’s seamless, because you can have three or four levels down. That’s when things get complex. A lot will go back to having predictability in the system to get those alerts and solve them in a timely fashion.
SLD: This is the downside of 50 billion things on the Internet, right? With one thing, you know who to point to. With 50 billion, you don’t.
Guedj: There are going to be some throwaway devices. If something doesn’t work and it’s really low-cost, people will replace it. The expensive stuff has to be covered by some service. The power companies can benefit from smart grid, which we talked about earlier, and that’s one application that is being deployed because there’s so much at stake there. They have to have a way to solve problems.
SLD: The darker side of this is security. What happens if your drive by wire automobile is hacked?
Guedj: The industry will benefit from that. Someone could argue they missed a meeting involving a $10 billion deal because they got the wrong directions.
Heinlein: It’s going to be a progressive rollout. While self-driving cars are interesting, they’re still some time away. The applications we talked about are not mission-critical. They’re life-improving. But progressively we’ll get better and better at this and security will evolve to do more mission-critical things.
Logvinov: We see a little bit of that today. If you think about the ‘bring your own device’ model for smart phones, who’s responsible for service working? It’s a collaboration between the user and the service provider. To enable that you need a secure container on your phone to isolate your enterprise data from your personal libraries and applications. Some of that is already happening today. As we move forward, we’ll see more and more of that. Essentially, when we start talking about the Internet of Things, it is essentially ‘bring your own device.’ We will be responsible for making purchasing decision about whether to acquire some gadget, and we also will want to expose this gadget to many services. It will be a collaboration between the user and one or more service providers.
Guedj: There will be hierarchical levels where the network may be on infrastructure on top, and the target devices with service providers enabling and controlling local devices.
Izadi: Different levels of aggregation makes a lot of sense. When it comes to safety and mission-critical, having redundancy and fail-over in case something happens are critical. Those are in place today with some of these applications. Failures happen. How do you deal with it?
Heinlein: There’s also the issue of WAN versus wired networks. From a home services aspect, homes are becoming more and more wired—not as much in the U.S. as in other countries, even though that’s the trend—and that’s the gateway to manageability. The local gateway could become the hub of controlling, configuring and debugging the local services. But it’s also the aggregator that goes into the networks.
SLD: How much of the Internet of Things is local versus cloud-based, or something in between?
Guedj: You could not afford to disrupt the network for a very tiny request, so you probably want some aggregation to use the network efficiently, no matter how large the bandwidth will be. But in terms of drawing a line in the sand, there isn’t a hard line. If you’re roaming, sometimes voice recognition works and sometimes it doesn’t. You need some local intelligence. There are clearly things that fall in the cloud, such as aggregation of data, databases and such. There are things that are in the target device, too, such as sensors. And then there are things that are fluid. If you’re doing voice recognition, you may be able to do it in the cloud if you have a very good connection. If you don’t, then you have local parsing. It takes more power and drains your battery, but you have constant quality of service regardless of where you are.
Izadi: Privacy concerns will drive the dial back and forth. We talked about profiling energy use and using this information to determine what can be optimized. One person from the audience stood up and said it’s impossible in his mind to allow a utility to know his energy consumption pattern, because that’s too revealing about his personal life. He would prefer this information be available to him, but confined to his home. This is a cultural issue, and privacy concerns stem from cultural issues.
SLD: When you think about the Internet of Things, and many different things, how much of this IP will be re-usable?
Guedj: The hard-wired technology is very efficient, but not very reusable. The more programmable things you have, the more you can re-use. The challenge in the low-end devices is smart programmability, so you can have programmability in a very efficient manner. The industry needs to strive to have those kinds of building blocks you can optimize solutions on. Otherwise it’s too diverse. But you can’t have one recipe for everything.
Heinlein: If you look at these applications, they’re going to be very different, with different price points, power points, and battery lifetime. Some can harvest, others can’t. It’s going to be a very vibrant time. There will be some things that can be built today using off-the-shelf MCUs and so on, but over time you’ll see more diversity, innovation, and squeezing cost out. The innovation is going to be exciting.
Izadi: From an OEM perspective, this is going to be huge. We’re taking the same equipment, ruggedizing it, and trying to connect it whether it’s the Internet of Things or your home. That means the semiconductors have to support industrial modes, work in a wider range of operating modes. How do you design for that up front?
Heinlein: We see more diversity, not less diversity at this end of the spectrum. In the middle, in the network infrastructure, there’s going to be a need to consume data at the end points of the cloud. That’s where we’ll see low-power servers. But in the embedded market, there’s more diversity, not less.
Logvinov: Hardware ruggedization and adaptability will be important, but we also have to look at the software side. If you take home gateways, for example, how can you make a device in such a way that it can last several years and be adaptable to new services? One of the approaches the industry is looking at today is an OGI (Open Government Initiative) platform where you can implement services as you go and fill them as you need to and create new ones as you need to. The software side of things also will affect the usability of platforms.
SLD: What’s the biggest potential limiter of the Internet of Things? What can slow it down?
Guedj: The diversity and the complexity, but what happen first is the service providers seeing a way to save money or make money and then being willing to sponsor those efforts. In those areas you need a clear business model from the service providers. That’s what’s been missing for consumer electronics. They’ve been searching for an AT&T or Verizon to sponsor the cost of the TV. That’s what’s going to drive the industry first.
Logvinov: It won’t slow down. The train has left the station. It’s moving. The only limiting factor is the speed of innovation. It’s growing and it will find its market
incarnation in smart homes, smart cities, smart campuses. It’s everywhere.