Ideas on networking technology begin to mesh

NEEDHAM -- They go by names like i-Beans, motes, and "smart dust." And technology evangelists are convinced they are a next big thing of World Wide Web proportions: tiny wireless sensor devices that can organize themselves into networks to monitor everything from factories and warehouses to office heating systems to wet diapers in a hospital nursery.

"Wireless mesh networking" technology is being propelled by advances in battery life, low-power wireless communication, and super-cheap computer chips that have brought the sensor prices down to the single dollars, and sometime soon to dimes or even pennies. The word "mesh" describes how each sensor collects and passes along messages until they reach a main hub, creating a fabric-like network that keeps working if one or more sensors fail.

A few years ago it was a science project -- or a science-fiction dream -- but mesh networking is slowly starting to crack the mainstream, from deployments at art museums where sensors monitor the security of paintings, to oil refineries, to machine tool factories.

Three firms at the forefront of this emerging industry are based within Route 128: Ember Corp. of Boston, Millennial Net of Cambridge, and Sensicast Systems of Needham. Ember founder and chief technology officer Rob Poor and CEOs Andy May of Millennial Net and Paul Sereiko of Sensicast joined for a panel discussion last week with industry analyst Craig Mathias of Farpoint Group in Ashland and consultant Brendan McSheffrey of GMD Resources in Needham, a pioneer in networked fire-extinguisher pressure gauges being developed by his parents' Mija Industries in Rockland. An edited account of the roundtable, moderated by Globe telecommunications writer Peter J. Howe, follows:

Globe: What are some real-life opportunities for using wireless sensor networks?

Sereiko: We're working on a client project now with the US Department of Energy, monitoring consumption by electrical motors in industrial facilities. Twenty-three percent of all electric consumption in the US -- 23 percent -- is from electric motors that are running in factories. A sizable portion of those motors is not running optimized.

Mathias: They're running when they don't have to, or they're old motors that consume a lot of power.

Sereiko: And the DOE has made a conclusion that we could reduce energy consumption about 3 percent in the US if we could do preventative maintenance and monitoring and control on these motors. So, voila, in comes an application of wireless sensing where it would have been impractical to run a wire to every one of these motors and monitor it continuously, but a little wireless node that's the size of one of the mounting bolts, routing that information back to a centralized decision-system. You can suddenly save a lot of energy.

McSheffrey: We see almost the identical application in pressure systems within manufacturing facilities where almost 20 percent of all cost that goes into creating compressed air for pneumatic systems is lost just because of leaks. At this point in time there is no way to track down those leaks or that loss of energy. This technology will enable us to do so. The gathering the data piece is one of the most intriguing parts of this marketplace in my mind; the ability to gather lots and lots of data we've never seen before, it's like the creation of a brand new type of telescope or microscope.

Sereiko: It's partly applications where you're replacing a wire, but it's also applications where the installation was never practical with a wire. You couldn't wire a field or a forest to determine the likelihood of fire or issues around drought and chemical content of the soil, but wirelesses, it becomes practical to do that.

Poor: The Internet is poking out beyond just the laptops and desktops into physical objects. That's the growth from the top. I also see the RFID [radio frequency identification chips on individual products] grow from the bottom, so to speak, so that just about every interesting physical object in the world that we interact with will in some literal sense be on the Internet.

May: Most of these commercial companies want wireless. They can feel the return on investment. They can smell it. And they have tried it over the past five years, and they have been very frustrated by it.

Mathias: The other big field is security, fire safety, homeland security-related applications. It would be deploying sensors over a wide area to look for gases and toxins that shouldn't be there. Dropping them out of airplanes, leaving them on a beach, tracking shipping containers, looking for particularly hazardous emissions from shipping containers, things like that.

Poor: Including carbon dioxide.

Mathias: Yes, including carbon dioxide, which might mean that someone is in the shipping container.

Globe: Are there applications for consumers in the home?

Sereiko: Oh, yeah. I went away for the Martin Luther King weekend, skiing with my family, and I came home and my house had been virtually destroyed. I had a pipe leak in my attic all the way into the basement. It blew out the furnace. It blew out the hot water heater. It did tons of damage. Bottom line was the humidity level in the house was inordinately high, and the temperature in the house was 42 degrees when we walked in the door. Now, yeah, maybe there's some sort of ADT super alarm service I could have bought.

May: Yeah, there is, but I didn't want to make you feel bad about it.

Sereiko: With this technology it would be a lot easier to deploy humidity sensors, temperature sensors, connect that through a bridge device up to a cellular network, have your cellphone ring auto whenever a sensor exceeded a certain threshold, and instead of my house leaking for three days it probably would have leaked for about 5 or 6 hours.

Mathias: You can go to Home Depot and buy a water sensor. You put it on the floor of the basement, if it finds water, it will go off. What does it do? It rings an alarm. If you're not there, who knows?

Poor: It's not a system.

McSheffrey: How much additional would it cost to put a radio in that, connect it to a micromesh sensor network that works with your phone? The answer is not very much, and the utility goes up by an infinite amount.

Globe: You've all been able to raise some money from venture capitalists, but many are skeptical about the hype-to-business ratio in mesh networking.

May: The challenge we have in the venture community --yes, we have been able to raise money, but I'm talking to a lot who don't want to invest in us or this yet -- is, they see it as, it hasn't happened yet. They're not investing in industrial automation. That marketplace, if you go to a trade show for industrial automation, you don't see a lot of venture-backed companies there. You see the homegrown "my family started this business in Milwaukee." So the venture guys have had a challenge adapting their thought process to the fact that we're going to sell a lot of this stuff into markets that they haven't been traditionally been funding companies in.

Sereiko: It's difficult to perceive that a company like Honeywell or Siemens isn't going to come down and eat the company's lunch. But the reality is, those companies are primarily focused on facing the customer and on managing their customer relationships. If they can license technology from Ember, from Millennial, from Sensicast, and put their company's brand on it, they're fine with that.

Mathias: They don't eat the company's lunch; they eat the company -- often at a very good valuation. So we expect that we'll see a lot of that in this space in the coming years.

Globe: Some venture capitalists say they have a hard time understanding how you make money if sensors cost a dollar or less and run on pretty simple software. Can there be a Microsoft in this industry?

Sereiko: They tend to think there's only one way to make money in any given market, and that's just not the case.

McSheffrey: One of my real passions for this technology is the fact that it creates jobs. It will increase exports. It will create services across many sectors. It is the Internet for the physical world, but it doesn't just create software and service jobs. It creates manufacturing jobs, component jobs, system integrator jobs, installers, value-added resellers, and product development shops.

Poor: Intel right now is pouring tremendous amounts of money into this sensor network arena. They're funding a lab at Berkeley. Not because they think they're going to make money selling little tiny radio chips, because Intel doesn't sell little tiny radio chips, but they see it as trickles of data that aggregate into torrents of data which feed their servers.

May: Driving demand for the big things that they do.

Sereiko: But being somewhat the voice of reason here -- you guys are all going to shoot me icy stares -- you don't want to overhype this like Bluetooth was overhyped. RFID has been overhyped for years. I was in an RFID company that started when it was hyped and died before Wal-Mart ever bought anything.

Mathias: The marketing machines need to be controlled. There's a danger of what we call Howard Dean Effect: You peak too soon.

Globe: How much do you worry about viruses or denial of service attacks or hackers taking down mesh networks?

Poor: It's funny that security isn't the first thing. It's about the fourth. It's cost, it's reliability, it's lifetime maintenance issues, and then maybe you'll get around to security. Ember has incorporated this modern 128-bit encryption standard into the chip, so you can send fully encrypted packets over the air. That's a hedge; it's not a complete solution. But mesh networking solution itself . . . offers a lot in terms of security.

Mathias: The big issues would be around jamming, if someone built a big broadband jammer and tried to shut down all of the electromagnetic communications in a given area. That's very difficult to do.

McSheffrey: There are a lot of applications out there where the wireless will be a feature, not the primary application, so that if for any reason the wireless goes down, the product still works.

Globe: I'd imagine a lot of people with home smoke detectors that periodically go off would worry about false positives.

McSheffrey: The sensor industry, plain and simple, is getting better. These things are moving forward. I've seen a leap in accuracy for a lot of the same reasons we've seen a leap in the accuracy of all sorts of other products.

Mathias: When you design these networks, you design them so that one thing checks another, so that if you get an anomalous reading, you know it's not a problem. Let's face it; sensors do fail.

Sereiko: I have a motion detector in my office. I also have leaky windows in my office. One Sunday, I was getting calls from ADT saying someone's breaking into the office. Well, the wind is blowing so hard that the venetian blind was blowing off the window, triggering the motion detector. Well, let's imagine instead that there was also a C02 sensor. Well, now you could check whether there was a person.

Globe: Or maybe add a strain gauge on the floor.

Sereiko: Exactly. Eliminate the false positive.

Globe: So we really should be bullish on this industry's potential?

McSheffrey: The more we look at the market, the more problems we see that this technology really solves. I kind of feel like in many ways it's 1992 or 1993, and we're all sitting around the table talking about the future of the Internet, but even at this table we can't imagine the Google that might happen in this space.

May: The new technology creates a food chain. Everybody who's in IT will wind up participating in this.

Poor: To me, this is all music to my ears, because it's Moore's Law meeting Metcalfe's Law. Moore's Law says the silicon that runs this stuff keeps getting cheaper. Metcalfe's Law says the utility of hooking things into a system goes up as there are more things hooked into the system. We've got two exponentials working in our favor.

Mathias: Good sound bite, Rob!

Peter J. Howe can be reached at howe@globe.com.