One-on-One: Steven Woo

Semiconductor Engineering sat down with Steven Woo, vice president and distinguished inventor at Rambus, to talk about the IoT and where the real problems are showing up.

SE: What are the big challenges as we move toward an -connected world?

Woo: The challenges we see, moving forward aren’t particularly related, but they do touch on each other and there will be some interrelated impact across them. I would definitely have to list interoperability among devices as a key issue. It is starting to come to the forefront now, and the industry is looking at how to resolve some of these interoperability issues among all of the devices that are looking to be connected.

This year’s CES was very different from the last year, or the year before. It really seemed like this year, there was a lot more work done to at least think through some of the interoperability issues. As well, I saw several entities, Nest, for one, that were seriously talking about their competitive interoperability mechanisms. Their ‘thread’ infrastructure, which is what they call their mechanism of getting ‘things’ to talk to each other. And there were a number of products that were specifically designed to work with the Nest infrastructure.

Another indicator was Qualcomm and other players involved in the AllSeen Alliance. Qualcomm is one of those companies big enough to drive a significant type of ecosystem. It was nice to see that there were such leading companies that can give traction to the interoperability movement. No one really knows what directions will evolve, but certainly, one of the precursors to getting traction in any direction is to have an industry icon willing to try to try things out to get the momentum going.

SE: What will it take to really make the IoT work?

Woo: A big issue is the economics of how to make the IoT work. In most cases it’s really not the technology that is the hard part. It is fairly easy to make devices that can monitor and have low-power Bluetooth radios that can connect to any number of networks. The real question is how you make money at it and develop the correct business models. While the technology is there and interesting, it is not the most difficult part.

SE: Within this business model metric, what are some of the potential solutions?

Woo: There really weren’t a lot of specific solutions that were discussed. One of the themes that came up was what to do with the massive amounts of data that is being acquired by all of these connected devices. It isn’t so much the device, or even the raw data that is important. It is the insight that we get from it. For example, a panelist from IBM said that in the area of connected cars one of the things they were looking to do in the near future was to use the sensors in-vehicle to analyze driving behavior and feed it back to the driver to help improve their driving behavior. One of the points made was that if the analyzed data was fed back to the driver, and it indicated good driving habits, that the driver could share that data with their insurance company in hopes of getting better rates, for example. The interesting point about this is that is while many people discuss the fear of the big brother model, where such data is ascertained without your knowledge or consent, what this model does is feed the data back to you first and let you use it in a way that can benefit you if you so choose. This is a prime example of how the insight that the data provides can be beneficial to you, and has some value that could be monetized.

Then there was power. There is this burning desire to find a way to not have to recharge your IoT every day, or every few days, as is the case with many devices today. The movement is to push the recharge times out for months, or even years, with some of the more remote or passive sensors. The movement is to have ‘watch battery’ form factors, but there is a lot of concern how that is actually going to happen with these IoT devices spewing out so much data. There was just a lot of general concern about how one minimizes power and what is the correct type of operating model for these IoT devices to conserve power. For example, if you have a bunch of data, is it better to send it at a very low data rate, with the idea that is will consume low power, or would it be better to burst it quickly at a high data rate, occasionally, where the device is in the off state most of the time? It seems that the jury is still out on that. I have seen good arguments for both and have seen cases where both work equally well in a specific type of infrastructure. So it will be interesting to see what the industry will come up with—eventually. Most likely someone will build an ecosystem around the connected home or car and make such decisions. And if these decisions are proven out, then they will be re-used in other areas. The first successful implementations will begin to dictate the way things will go.

SE: So what are some of the issues surrounding this massive amount of data?

Woo: Another theme that comes much closer to what our industry is doing is along the lines of what to do with this tremendous amount of data. How do you move it back and forth? How do you store it? How do you analyze it? The issue at the core of all of this is what the computing infrastructure needs to look like to support this ‘Big Data.’

SE: Let’s talk security. What are the problems you’re seeing?

Woo: Security and privacy seem to be an expanding topic. There is a great deal of concern over things like connected cars and homes being hacked, and predictions that in 2015 the hacking of these entities would become a reality as they become more connected. There is concern that, with all the different sensors and connectivity points, each one is a potential vulnerability that has to be able to be individually locked and unlocked to outside access.

There is a general consensus of agreement that security is difficult and it needs to be architected in from the ground up. The infrastructure needs to be able to define and limit access to such I/O ports. Today, many products still haven’t done that. The industry is retrofitting security on top of what now exists. There is also concern that there still isn’t a very well thought out security architecture that has been developed as a part of the IoT infrastructure.

SE: So what do you think will happen going forward.

Woo: Our industry has started to intersect some of the stuff we do with the envisioned infrastructure. We have done a lot of work in memory, interfaces, and security. We are also working on making energy sources more efficient, while at the same time chip power requirements will drop. That is kind of where the chip community is heading.

On the access point security, one of the things we have been looking at is that type of functionality in our security work, and that is one of the driving forces behind the development of our Cryptomanger platform. It does exactly that for port security.

Finally, on the security end, it will be coming to a head this year. The number of high-profile break-ins that occurred in 2014 seemed to be much higher than in previous years. And one can definitely see the impact of security, or lack of, come to the forefront—2015 will see increased momentum in dealing with security.