MEET THE TECH EXEC
CTO IoT Strategy and Senior Principal Engineer Strategy and Solutions Enabling Division of IoTG, Intel Corporation
To download the full magazine and read the full interviews, click here.
Brian McCarson is the CTO of IoT Strategy and is a Senior Principal Engineer at Intel Corporation. He has a Master of Science in Materials Science and Electrical Engineering from North Carolina State University. Brian specializes in system architectures for IoT, data extraction, data visualization, innovative data analytics, and multivariate problem solving. He leads the Technology and Standards Team within Intel’s Internet of Things Group Strategy and Solutions Enabling Division and is the system architect overseeing the Intel IoT Platform.
According to your 3 videos that you published on Intel.com, where does Security fall into the 3 Phases of IOT? What are your concerns at each stage?
In the first phase of IoT (Connecting the Unconnected) security is relatively easy. You have one object or “Thing” that is stationary and someone has to manually install something to bring the Internet and connectivity to that “Thing”. The physical process of bolting the connectivity on to the back of a pump or putting that into a car establishes the security or trust you need. You can then provision that device, connect to it, and give it an IP address and track it pretty well. It gets exponentially more complex as you start moving to the different phases of IoT. In the later phases you have questions like, “what happens when a device is self-provisioned?”, “what happens when a device wakes up and wants to optimize its environment and use another data source next to it?” and “what happens when a device figures out how to change its configuration and settings to optimize performance?”. That is a much higher degree of complexity than you would see in Phase 1. The best example I can give from that is autonomous driving.
Imagine this scenario: one of the most interesting value propositions for autonomous driving is this idea of platooning. It’s kind of a cool concept. You get on a freeway and as soon as you hit the on-ramp your car basically takes over and you’re just sitting there watching Game of Thrones or something on your tablet and your car pulls right onto the freeway for you and it starts platooning with other cars on the freeway. It can basically operate with maybe a 6-inch distance from the bumpers in front of and behind you. You will have a stack of cars all lined up together just going 80mph down the freeway, reducing drag within a tunnel of air resistance and you’ll get better fuel economy and reduced commute times. But that involves being able to trust the cars in front of you to tell you what they are doing. The only way platooning works is if you can trust that if the car in front of you sees a hazard it will communicate with everyone else in the platoon chain in time for them to respond. But when you commute to work you’re not going to plan your commute with your neighbors so you can platoon with someone you can personally trust. This has to have a seamless connectivity to all the other vehicles and has to happen across multiple makes and models of cars not just certain vehicles. Multiple manufacturers, networks, and passengers need to agree to cooperate and you have to trust that just because you’re communicating with other cars you’re not allowing them the ability to drive you into a barricade if someone with malicious intent joins the platoon.
It’s so much more complicated in Phase 3 when you are considering what information a system will have to share and who it will be allowed to seamlessly communicate with.
What about as far as external threats?
Imagine that someone didn’t like a city or the type of people that lived in that city. If they can hack into even a handful of vehicles that have a Level 5 automation capability where there is no steering wheel, no gas pedal, and no brakes and takes full control, you are completely at their mercy.
That is an interesting challenge that we have to be able to overcome with autonomous driving. How do you allow for seamless communication between vehicles and still be protected from hacking? Many people are thinking about how to do it. We will probably have to build multiple layers of protection.
So, the first layer could be a hard-wired, functionally safe system within the vehicle that has its own private in-vehicle network and allows you to have all the sensors and actuators controlling things independently and nothing can interfere with that. The next layer could be the informational awareness layer where you are sensing the external environment beyond what the vehicle’s sensors can see and you can take those bits of information and then filter as you need to.
That partitioning can be done now. You can have one-way firewalls where no data can come in but data can go out. But part of the problem is that manufacturers are making more vehicles with wireless and wired connections for additional access to media capabilities. One could argue that there needs to be a strong separation between media-based electronics and vehicle control-based electronics in your vehicle architecture.
What superpower do you want most?
Manipulating time, time travel would be the best.
What is computing in the Fog?
Fog is probably one of my favorite subjects. Before I can explain my thoughts on Fog I need to explain my thoughts on the Cloud. The thing that is great about the cloud is that it makes it easy for almost anyone to afford having data center capabilities at their fingertips without having to buy their own ridiculously expensive data center. You can allow people to have little packets of data center compute and storage usage and have a kind of communal data center business model. Facebook is a great example of that. They buy and manage enormous data centers and you only use the tiny little portion that you need for your links, networks, and photos. Amazon Web Services and Microsoft Azure are also great examples of that model and they offer a great economic value proposition that allow individuals or even startups who could never afford the initial costs of building their own data center to jump right into the market.
There are some problems with the Cloud however.
There are some limits in physics and economics that are really hard to overcome. For example, if you own a retail store and you need to stream 4k or 8k resolution video because you want to be able to look at the video and analyze it in real-time so that you can know the demographics in your store and decide how you want to advertise to your customers based on who the people are that are walking through the door. To do this takes A LOT of high-resolution video data. Streaming all of that to a remote data center is very expensive and time consuming and the economics don’t always make sense for a Cloud-based system architecture.
The same is true for fully automated vehicles. If you are having your autonomous car drive your child to school and it sees a hazard, you are going to want that automated car to decide to apply brakes immediately and on its own and not have to dial up to ask if the Cloud if they should apply the brakes. The time latency of making decisions when you have to send information to the Cloud is another problem.
When it is foggy outside that means the clouds are down at ground level and immersed around you. Fog Computing is a metaphor for bring the power of Cloud Computing (which by definition is remote) down to where users and “Things” are in an immersive way. Fog is around us, it’s immersive. If you think about an autonomous vehicle it’s a data center on wheels, even if it doesn’t have a wide area network connection it can keep you safe without it. Fog is interesting to me because it can solve key problems: cost of data transmissions, reliability of access to the cloud, and the latency or time it takes to make a decision. Fog is not for every application. But for some it can completely change the way they operate.
The cost of computing, connectivity and storage is dropping so fast because it is commoditized. Take a single transistor, the same kind that sits in in the brain of your tablet, smartphone or laptop, it can switch on and off up to 10 billion times in a second. To put things in perspective, if you wanted to switch a light switch on and off 10 billion times it would take you over 200 years without any sleep, food or bathroom breaks. Yet the cost of producing those transistors is less than the cost to grow a single grain of rice in rural China. The cheapest unit of food on the planet is more expensive to produce than the most complex unit of compute. That’s the phenomenal reality of the compute and connectivity we have in the world today. Fog computing is now becoming practical whereas 10 years ago it would have been cost-prohibitive. I get so excited when I try and imagine the possibilities. It’s going to be interesting to see how our homes, schools, workplace and the rest of our daily lives will evolve with Fog technologies coming to market.
What did you want to grow up to be when you were a kid?
I wanted to be a paleontologist from a really early age. I was completely obsessed with dinosaurs when I was a kid. I was born in New Mexico. My grandpa was an outdoorsman and I would go out rock hunting and gold mining with him. I got my undergrad in Geology but unfortunately, there just weren’t any jobs at all at the time I graduated. I switched my direction and decided to go into Materials Science and Electrical Engineering and this helped me switch to the semiconductor industry.
What song best describes your work ethic?
My go-to jam before giving a presentation is always Dr. Dre because it gets me pumped up. For my last Intel talk in China I had them play the intro to “Nothing but a G Thing” as I walked out on stage. But the song that best describes my work ethic would probably be something from Radiohead. Maybe “No surprises” by Radiohead. The lyrics are kind of interesting and that’s my favorite band and that song was my ringtone for a long time.
If you were giving guidance to someone in engineering what advice would you give them?
I’ll give a few answers to that question.
Here is my technology answer: I think as a society, we’ve figured out a lot of problems with connectivity, compute, and storage and have many of those issues worked out. We haven’t figured out how to replicate the human mind and the things we take for granted. You can walk down the street and see someone walking towards you and your brain automatically registers that it is someone you recognize but you don’t necessarily know why. Your brain scanned and registering the shape of their facial features, the way they walk, the way they carry themselves and decided that you recognize them as one of your friends. We take for granted how easy that is for us to do, but we are still figuring out how to do that with computers and cameras. But some of the advancements in compute technology around cognitive neural networks, machine learning and deep learning are helping get us closer. In the field of science called Biomimicry, we are starting to replicate some of the ways biological systems like arrays of neurons have structured themselves and see if we can apply those natural methods to the way computers think. The whole field of artificial intelligence to me is one of the most fantastic technology areas in the coming decades.
How do you teach a computer to do much more than if/then statements? Teach it patterns, and sub patterns. Teach it to observe and learn and make its own if/then statements. It’s a very different way of approaching the science of computing. We used to only assign computers to do mundane tasks and workloads. What prohibits us from assigning computers do the miraculous? Apply computing technology to perform complex analysis of someone’s DNA and their blood profile and discover they are at risk for kidney failure. Then recommend changes to their diet and medications/supplements to protect their health. Why can’t you employ a computer to do more of that advanced thinking. If I was in college I would focus on AI machine learning and deep learning and not much else.
This is the marketplace answer: What differentiates scientists from engineers is that scientists ask questions and test hypotheses to learn. Engineers use science and those same methodologies to solve problems. If learning and advancing the knowledge humankind has of our universe motivates you, then academia is likely the best path for you. If you want to be addressing the market issues that people are facing with their daily lives, learn what isn’t working and what can be improved with technology to address those issues, then engineering may be a better choice for you. Addressing market problems with end users in mind is the best way to make money in business, combine that with the right technology and you have a recipe for engineering success. I tell all my employees that I’m not interested in just doing cool experiments or inventing cool technology. I’m interested in solving real end user and real customer problems with technology. I like the combination, that’s where the most interesting magic happens.
[vc_cta_button2 h2=”” title=”LEARN MORE” color=”belizehole” accent_color=”#ed884e” link=”url:http%3A%2F%2Fwww.managedsolution.com%2Fmeet-the-c-level-interview%2F|title:managedsolution.com|”]
MEET THE TECH EXEC INTERVIEWS
Managed Solution is conducting interviews as part of an outreach initiative to share trends and engage technology enthusiasts in the southwest.