WTRS Executive Interview

Interview with Joy Weiss, President & CEO of Dust Networks.

May 11, 2007

George: Can you start by telling us a little about your background and how you came to Dust Networks?

Joy Weiss: I started my career at Nortel – I have an EE degree from MIT, and I did a variety of different jobs at Nortel. I started out in engineering and I ended up in sales, and then eventually running a business unit for Nortel. But the themes during my Nortel days were always clearly technology oriented and networking oriented, but also of trying to tie the two together. Because I did the leap from technology to a very customer facing role that became a really interesting spot to me, of marrying new technology with customer needs.

Since I left Nortel, I’ve been the CEO of two other startup companies here in the Bay Area, one actually based in France (but I was the CEO based here), and one based here in the Bay Area. Even though they were very different companies from Nortel and Dust, they all had a theme of bringing information- bringing more or new information- to people or systems so it could be made use of. At the end of the day that’s what telecommunications is, whether its voice or an email. It was different facets of that equation, combined with the fact that I like startups so much because you get this opportunity to be deep in technology and with the customer at the same time, which is a treat when it works right. Not a big treat when it doesn’t work right, but a big treat when it does.

When I came to Dust, I was coming off of another role and looking for what I wanted to do next, and had started looking at the ZigBee space independently and thinking ‘gosh this could be a really interesting space’. It fit my criteria of interesting technology, but also the need to have a real market pull. I’ve had the misfortune of doing interesting technology when there is no market pull, where you think ‘I can outwait’ or ‘I can market my way into it’, but that’s a tough thing to do. Some companies have done it. It takes a lot of money and a lot of time, and is very tough for a startup to do. I did want to do another startup, because I really like the energy and culture of startup companies, but it had to be nifty enough on the technology side and it also had to be a real market with real traction around the application space.

As I began to dig into ZigBee, it seemed to me that it might be one of those emerging markets with real customers saying “if you build it I will buy it” and also some pretty nifty stuff going on the technology side. Coincidentally, Kris Pister was getting his first round of venture financing in place at the time and was looking for a CEO to run the company, and that’s how I came to be at Dust.

I’ve been here for a little more than three years, and joined coincident with the first round of venture financing, when our products were basically a slideshow - a really good slideshow. But we released our first version of SmartMesh that first year (at the end of 2004) and did our first customer trials with, at the time, a division of Honeywell that was doing energy management, doing a grocery store energy management trial. We did a trial with SAIC, who was doing perimeter security, and it has been quite a great ride since then.

George: What is important for everyone to know about Dust Networks and can you talk about the initial industry focus of the company?

Joy Weiss: Dust Networks is providing embedded technology so that OEMs who are in the business of providing sensing systems can do that more cost effectively for their end customers. Our products comprise our system-on-chip technology combined with our networking software. The hallmark of our product offering is we deliver very high reliability wireless systems that have great flexibility in terms of the network topology that they can accommodate, as well as the range of power supplies that they can accommodate.

The company was really founded around a premise that said: for wireless sensing to really be ubiquitous (which we believe it can be, in terms of this notion of connecting the physical world to the IT world), that you would have to figure out a way to combine wire-like reliability with wire-free economics in the extreme, because you really want to have the flexibility to put a communicating sensor wherever you might want to put it. So the first instantiation of products that we did was based on our protocol, and based on off-the-shelf standards based hardware. We call that protocol the Time Synchronized Mesh Protocol, or TSMP, and we really did prove that you could get extraordinary reliability, 99.9 plus percent reliability, in very harsh environments using this protocol, but we could operate these networks on batteries for years at a time if required.

Now you can also plug-in a node if you happen to have power available, but, again, to achieve the vision that we saw of having anybody be able to say “Gosh now I can go put an application in where I really wanted to put it in, where I couldn’t put it before” we felt you really had to snip both wires. This was the principle: snip the communications wire and snip the power wire and the first generation of products that we did really proved that you could achieve that vision, which previously required you to make tradeoffs. Typically to get reliability with a wireless system you cranked up the power, and this notion of trying to turn down the power while you turned up the reliability was a combo that hadn’t really existed before.

The other dimensions that we tried to simultaneously solve for were that we wanted the system to be scalable, because we also believed that you would end up with very large networks, but it also had to be really easy to deploy. Because we knew that this was going to go into applications where there were people well trained in deploying sensors, but not well trained in wireless, we wanted to ensure that there was a minimal hurdle to deployment, it was critical that these looked like a wired deployment. Not that you would literally be pulling wires and cables to it - but the special knowledge about wireless would be minimized, so that truly anybody in an organization across the application space we were considering, all the way from somebody who works in an oil refinery to somebody who works in their living room, would have an excellent experience of being able to use this technology.

So reliability, scalability, power flexibility and ease of use all ended up being proven out in our earliest deployments to be achieved using this basic protocol (TSMP) on top of industry standard technology. But the place that we got the most resonance in terms of “gotta have it now” ended up being the industrial market, in particular industrial process.

I like to put applications in two buckets that I call ‘wire replacement’ and ‘wirelessly enabled’. ‘Wire replacement’ applications are not literally where you pull up the wires to put in wireless: historically, you had deployed sensor networks that were wired, and now you will deploy them wirelessly. ‘Wireless enabled’ applications are those that you simply wouldn’t do before, either for cost reasons or for safety reasons or for a variety of other reasons, it was simply impractical to deploy those applications until they became wireless.

As a young company you are always looking for a ‘quick big hit’ that creates some market momentum that you can then ride and expand from, and industrial seemed to us to have the magical ingredients. It was a ‘wire replacement’ application, so they already understood sensor applications, they already put sensors in and that wasn’t new to them. The new part would be the wireless part. That was a big tick mark. There were OEMs who already sold sensing systems, everything from the sensor up, because you have to know what to do with the sensed data when you get it. This has been one of the hurdles for systems such as RFID, where it’s a great idea, but the eco-system has to be so built out before you can put in the first tag, that it necessarily takes a long time. The nice thing in industrial is that the systems that make use of sensed data already exist, and the OEMs who provide the sensors also provide those systems. So there was a nice sort of network effect, if you will, in a single partner/channel and you also had a very concentrated customer (base) of big companies with big problems who bought directly from the OEMs who could solve those problems. So that meant a very short food chain of getting to market, and a value proposition that translated very well and very easily from us to the OEM, and from the OEM to the end user.

George: So the value proposition worked all the way down the selling chain?

Joy Weiss: Right, and it’s a short selling chain. We looked at a variety of other opportunities where the interest in sensing was high, but where the number of players that you had to get around the table was also really high to make that come together. Candidly, the other thing about industrial is that it has among the highest ROIs of any of the applications because the cost of cable and conduit in that domain is so expensive, and because you are in environments where you have to have intrinsic safety etc… so it became a sweet spot for us very quickly. We ended up focusing there, and found that the eco-system extended to standards bodies that were also working in that domain, so it completed the circle for us.

We were working on a standards based radio, we were interested in standardizing our protocol and this proved to be an excellent venue for us to do that in an environment where the values we brought to the table, the reliability, the scalability, the power flexibility, the ease of use, were all where the bar was being drawn for this industry. In other words, they needed what we had or they couldn’t do what they needed to do. So it was a very nice fit for us and that’s where we spent the better part of the last two years, intensely focused on industrial. We do have some customers that are outside of that, but the primary focus for us has been industrial

George: So the customers outside of this main industrial focus, is this where you are seeing more of the wireless enabled applications opposed to the wire replacement applications? Or does the map happen that way?

Joy Weiss: Interestingly, a primary application in industrial is the ‘wire replacement’, but for the same application we’re also seeing ‘wirelessly enabled’ deployments. In other words, we’re seeing companies saying ‘not only will I put in my new sensors based on this wireless technology instead of wires, but in an environment where I already have sensors I will overlay wireless technology, where I never would have put wires before because it is cost prohibitive.’ So the neat thing in industrial is that the baseline is ‘wire replacement’, but in fact it’s a very healthy mix of ‘wire replacement’ and ‘wirelessly enabled’.

George: So there’s a big opportunity for wireless enabled applications and I’m presuming that that cost of deploying them now is much lower than it would have been on a wired network and as a result the ROI on the program, whatever sensor program, is now positive rather than negative?

Joy Weiss: Exactly. BP (who is one of Emerson’s customers) has spoken publicly about their experience in using Emerson’s technology, and they have talked about a 90% reduction in the cost of installation. That’s extraordinary. You don’t often hear about an order of magnitude cost reduction. Usually you hear numbers like 10% less or 15% less – you are usually talking in very small increments.

That’s the vision we have for the ubiquitous use of wireless sensors. It’s because of that step change that we believe that this really is a technology that will pervade a variety of different applications over time. The lowest hanging fruits are the people that already use wired sensing technology and who can get cost savings. You asked, ‘do we see wirelessly enabled outside of industrial’ and the answer is ‘we see it both inside and outside of industrial’.

Here are some very neat examples where they are wirelessly enabled. One of our favorite examples is Streetline Networks (http://www.streetlinenetworks.com). Streetline was founded by Tod Dykstra, one of the founders of Dust, who really had a vision around a complete application, but Dust’s focus is more of a horizontal platform. He had done a lot of work with cities and in city planning, and had a vision around how the technology could be used in city infrastructure.

He founded Streetline Networks, and the initial focus for them is on parking and traffic management. They are using our wireless networking technology combined with their own - they’ve built their own wireless/sensor combo and their own middleware and applications logic, and they are doing trials now in San Francisco in a variety of different parking areas where you can see if somebody is in a parking spot. Eventually, this can be used for traffic management because you know where the congestion is. It can be used to tell people where the parking spots are, which in turn has traffic management impact so you are not sending people to parking spots where they have to drive around and around. And then of course, there is the one that is important to cities, but is the part that consumers like the least, but it actually makes it much more cost effective to do parking enforcement…

George: Revenue generation!

Joy Weiss: It’s a business and you have to go where the money is. It’s extraordinary how much money cities spend on parking enforcement, and if it can be done more cost effectively, then it happens to be one of the few things that they spend on often, and they spend a lot of money on it, because it’s a source of revenue to them. But that’s a totally ‘wirelessly enabled’ application. So basically they are putting this technology into speed bumps. You just put it on the pavement, so you are not tearing up roads, and you are not impacting the infrastructure, but you’re in essence creating this overlay, an application overlay, in an environment that otherwise you could never have put this in and it is delivering extraordinary value.

George: Because the alternative is to dig up the street in some fashion and lay wires…

Joy Weiss: Exactly, which is prohibitively expensive. So technically can it be done? Of course it can be done. Does anybody do it? They do it for traffic circles but that’s the only place that they use wires. They certainly wouldn’t use wires for parking, as it would be exorbitant.

I think that is one of the attributes of wireless sensor networks in general, and the reason that we feel so excited about the growth opportunity, is the idea that you can conceive of an application that makes use of data about the physical world and bring it into an existing venue and deploy it with very little disruption to the goings on in that environment. Whether that’s in an industrial plant, or whether that’s a business building where you are putting it in for energy management, the idea that you are not having to run cable and conduit and tear up walls or tear up streets is really transformational, relative to the kind of applications that we are seeing being enabled out there.

GE Sensing recently announced a product, where they are using this technology as part of a pharmaceutical validation product and going into existing plants and putting monitoring points all over the plant. Pulling cables and conduit is a non-starter. The only alternative is to do it manually, and literally having people walking around doing that. So that’s the part that’s really exciting about this. It’s not required that you have a new facility in order to use the technology… the technology actually brings as much or more benefit into an existing environment, where you are overlaying a wireless communications infrastructure to enable the gathering of more information about your facility, your process, parking, or even ozone (in the case of our partner Novazone). Whatever it is that you want to get information about now becomes readily doable.

George: What can you tell us about the recent award from the World Economic Forum? That was a really interesting, high profile award to receive. Can you tell us just little bit about that?

Joy Weiss: Thank you. We were very excited

The World Economic Forum is a venue where leaders from industry, political leaders, social entrepreneurs come together. They’ve made an effort over the last couple of years in particular to make sure that business and technology entrepreneurs are also part of the discussion there, which really is about ‘how can we make the world a better place’. It sounds very heady, but that is in fact the essence of what ends up getting talked about at the WEF. It’s a pretty intense setting with an extraordinary group of people who are very accomplished, and many of them also have access to extraordinary resources, be they financial or intellectual or civic resources.

And so the WEF started a program called the Technology Pioneers program to solicit companies who had innovative technologies that could potentially help change the world, in the broadest sense, to participate in the WEF. They had hundreds of applications, and chose a small number of companies. We were fortunate to have been chosen in acknowledgement of the fact that they shared our vision for how wireless sensing networking technology can change the world.

That’s a really profound acknowledgement to us because, even though as a young company we are very focused on a particular market and particular product instantiations and the day to day of that, …we also have a vision, which is why you start the company to begin with.

You do get the opportunity to participate in the WEF when you get this award, and so I attended this year, and had the opportunity to talk to a very wide swath of business people, government leaders and social entrepreneurs about the capabilities that we have and immediately, the light bulb went on, and the kinds of applications and opportunities that they were able to dream up in the short time that we had to spend together was really among the most exciting parts of the WEF for me.

When they talked about the challenges that they see in the environment, which as you might guess, was a key theme this year, even though wireless sensing technology isn’t being used in a widespread way for that just yet, there’s no question that this is a piece of how the global climate challenge will be addressed. This technology will be one of the many things that help to address that challenge. So it was really interesting to have the opportunity to hear from people their vision for how the technology could be used, and then also hear them describing problems that I knew the technology could be used for. So we will be attending again this year, and I hope that we have more ideas to bring to the table and more opportunities to hear about important problems that we can help to solve. It was very, very interesting and a real honor to be able to participate.

George: What about ISA-SP100 and Wireless HART? The WTRS view is that Dust Networks started out by using a standard based approach and that contributing IP to the HART and ISA-SP100 standards is just a continuation of that approach. How is Dust helping to drive those standards efforts?

Joy Weiss: That’s correct. So the standard that we adopted initially was the IEEE 802.15.4 standard, and of course the system-on-chip implementation that we are doing right now (our SmartMesh XD product line) is based on our own implementation of our 802.15.4 chipset. But the protocol that we were using was not yet standardized – often it was called proprietary because it wasn’t a standard yet – but it wasn’t our intent to keep it proprietary. We just needed to find a venue where it made sense to contribute that IP. The first such venue was HART. HART released a draft standard two weeks ago, which went through a pretty significant review by technologists and business people from the HART member companies. You can think of the Wireless HART standard as an adaptation, a HART specific adaptation, of the underlying principals of the TSMP protocol, and we participated quite heavily in its development

We are now participating in SP100. SP100 has a considerably longer timeline but SP100 has already publicly stated that they have made some basic decisions about what the technology will look like, and if you look at those basic decisions, (a) it’s a full-mesh network and (b) it’s a frequency hopping network. When you start adding up the components that they have already articulated, you can see that it is quite consistent with TSMP and that’s what we care about. Not that it’s our specific implementation per se, but we believe strongly that TSMP provides great flexibility in implementation and the necessary functional outcome for the end customer, so that the technology is successful in the market place. We are very pleased that both HART and SP100 appear to be adopting technologies that have similar foundations, and look forward to continuing to participate. We participate as an editor in SP100, and of course we’re a member of HART.

And as you might guess in any large market like this, standards provide a galvanizing effect to that marketplace. Whatever uncertainty may exist tends to go away once a standard is in place. People are not worried about ‘if I buy it from you will it work with the other guy’. So the first key piece was IEEE 802.15.4, and the second key piece in industrial will be to have WirelessHART, and then SP100. The bulk of the initial opportunity we see in industrial is for field devices. The bulk of the field devices are HART devices, so having a WirelessHART standard be first is great, because it unlocks what we think is probably the low hanging fruit in the industrial market .

Within HART there are about 25 million devices already deployed. Maybe only 10% of those devices are actually communicating their digital data back to any kind of host system. So what’s really exciting about Wireless HART is that you can now just retrofit your existing devices and gain instant access to this digital information – without interrupting or interfering with the existing 4-20mA loop. I think you will see no later than early next year, a number of companies with wireless adapters that literally can screw onto a ten year old HART device, and begin participating in a wireless network, just as easily as that. (snaps fingers)

So the notion that you can retrofit comes back to this idea that I said before- the idea that you can overlay a network so readily with this wireless technology because of the fact that with a small battery operated device, that will last for years in the field, you can literally go in and screw it onto something that is already out there and enable rich data to be transferred between that device and a host system, is getting a lot of people excited.

We were at the Interkama show in Hannover two weeks ago, which is a big industrial show. The HART Foundation had a booth there, and there were about ten vendors showing inter-working wireless products on the booth. Several had WirelessHART adapters, which they were demoing there. There were also several control systems, so we are really seeing some very exciting momentum in this market, with most of the major vendors already demonstrating products and their intent to deliver WirelessHART compliant products the beginning of next year.

George: We hear recurring comments that DARPA funded some research in wireless sensing networks at both MIT and UCB and one of the results was the IEEE 802.15.4 radio standard. After that, the design approaches to using these radios seems to be slightly divergent representing the different approaches of these two design centers. Do you have any thoughts on that idea or how the underlying platforms will evolve in the future?

Joy Weiss: DARPA’s mission is to fund innovation and to make sure that the US stays ahead of anybody else on many technology fronts. And so they sprinkled around some money to do hardware research, they sprinkled around some money to do software research and they sprinkled around some money to do systems and algorithms research. You may know that it is not just MIT and Berkeley anymore. There is a very large community in academia that does all kinds of fabulous research in wireless sensing networks in hardware, software, algorithms, and what can be done to process sensor data on the node. It’s a very, very rich field of innovation. But they’re (DARPA) not VCs. They don’t fund products per se, they fund innovation.

I think that because there is such a broad market for the technology that you will see - the great thing about wireless is that the airwaves can be shared - and I think you will see, over time, a variety of different implementations. I think that as more companies get engaged with more customers, we learn more. I know certainly as a company - I was here for our first deployments - we learned a lot from the customers, and tried to bake into our next revision of the product what we learned from that.

I’m guessing that all of our competitors and colleagues have done the same, and I’d like to think that what we are doing in the market now, and what we are doing in the standards bodies now, reflects the best knowledge that exists wherever it comes from.

If you look at the implementation that we did a couple of years back, even though the principles were the same, I would say that the implementation has come a long way in adopting the best of what else is out there. My guess is that we will see the same thing coming from the other camps. They’ll start taking a look at us and saying “That’s pretty nifty, maybe we should try and do that here?” I think that there is more opportunity for convergence than not.

Even though TSMP is a time synchronized based implementation, we actually borrow some principles from CSMA and use them in the implementation of TSMP. Likewise, we are seeing some folks who are using CSMA figuring out how to do synchronization now and even potentially how to do frequency hopping, even though it may be coming from a different side.

Even in the products we deliver today, we provide our customers great flexibility in being able to turn knobs to optimize around a variety of different architectures that they might be implementing in the same framework. And I’m guessing that we will start to see the same thing from others, and perhaps over time, we will see an umbrella under which all of these technologies very peacefully co-exist.

More information about Dust Networks here...

This interview ran in our May 11, 2007 newsletter issue.