nota

MASSive sensornets are the future Internet

One of the highlights from IEEE MASS was the panel discussion on the state of the future internet. Feng Zhao (Microsoft Research), Tarek Abdelzaher (UIUC), Rick Han (Univeristy of Colorado, Boulder), and John Heidemann (ISI/USC) traded thoughts on how the next Internet is shaping up — not suprisingly, there were many slides of the traditional Internet “cloud” touching the wireless sensor network “cloud”. One common theme was enabling sciences for the masses — the idea that anyone and everyone should and will have access to the coming onslaught of pervasive sensors. Feng showed the new MSR sensorweb that integrates real-world sensors with Microsoft’s earth-scale mapping service. “Data is King”, Feng wrote as he discussed how every-day users will write mash-ups of sensor feeds. Rick Han discussed “sensing for the masses” enabled by “cheap technology and open innovation.” John Heidemann brought a reality check and questioned the panel about the definition of a “user” of this future Internet. But, John followed up with a new term, “slogging”, that he uses to describe the coming age of things blogging (see previous post, doesn’t really roll off the tongue like “blogjects” or “spimes”). John added two twists to the future Internet — the idea of federated sensornets enabling the “citizen scientist” (John cites Mark Hansen as coining this term).

While I think these ideas were perfectly academic in scope, the reality of free access to a global treasure trove of sensors is far-fetched at best. Take GPS for example, free for all, but resolution limited by a few. Take Google Maps for example, “free” for all, but resolution-limited by a few. Feng Zhao put it best in his answer to an audience question about his opinion on business models for sensornets and the reality of every seeing a $1 wireless sensor node — Feng replied the answer lies in amortization of cost, not necessarily an inexpensive sensor. In other words, providing access to data for those willing to pay for it. So for those holding access to the data when “data is king”, it’ll be “good to be the king”.

IEEE MASS 2006 under way

IEEE MASS (Mobile Ad-hoc and Sensor Systems) started with a terrific keynote from Professor Kumar (UIUC). The general thesis of the talk concerned the acceptance and future research of wireless sensor networks as hybrid control systems rather than data-centric sensing networks. Prof. Kumar proposed academia is in the “age of system building” and cited the current state of wireless sensor network technology as the third generation of control. The third being networked embedded control, preceeded by electrical digital and electrical analog generations. (Ok, as a mechanical engineer, I have to say fifth generation — what about mechanical control preceeded by pneumatic!).

Professor Kumar concluded with two rather poignant ideas. He proposed theory as the limiting factor, not technology, in developing and deploying wireless sensor network hybrid control systems. The idea being that all the pieces of technology exist to create complex systems, but the underlying control theory lacks the ability to handle the new paradigm of a multiple input, multiple output system as presented by wireless sensor networks. Following that premise, Professor Kumar argues that theory cannot derive the needed architectures for complex, distributed hybrid systems — rather experience will produce the appropriate architecural characteristics.

Pervasive, Intelligent, Social wireless sensor networks, eh?

The 2006 IEEE IST-WSN workshop was yesterday in Vancouver, B.C. A small group of researchers talked about intelligent systems with wireless sensor networks. Ankur Teredesai from RIT talked about pervasive social networks — the idea being hardware can push social networking. His prime example is the integration of IM onto cellphones. Ankur’s ideas are very much akin to blogjects, and I think the presentation at a technical conference confirms the evolution of the present day social networking technology from the browser.

The highlight of the day-long workshop was the panel discussion — Peter Graham from the University of Manitoba, Ankur Teredesai, and Alex Coleman from the University of Alberta gave thoughts on the current state of wireless sensor network research. Power management — either the lack of or glutony therein gave cause to a good debate that ended not about power and energy effiiciency, but rather about ethical issues and the need for a top-down design emphasis. The leave-off point: research can benefit by (futher) involving the application/user. And we’re left with that old design mantra: use what you build.

Grapes, Wildfires, Mountains: Wireless sensor networks take a hike (outdoors)

It seems experiences with real world wireless sensor network applications for the great outdoors was the theme for this past week. Peter Tsepeleff, CEO Grape Networks, gave a talk at the WSN-SIG about a depoyment in several California vineyards. Peter and company deploy agricultural sensor networks to help vinters improve yield and/or quality of the grapes. A neat niche application — developed in a new unit: acreage. They figure the way to address scalability is designing and deploying their system in units of 100 square acres. Regardless of the exact number of acres, this definition of strategy, constraint, and objective — all rolled into one — for deploying an outdoor wireless sensor network is great to see.

At the latest CITRIS event Nick Sitar, Professor & Director of the Earthquake Engineering Center at UC Berkeley, talked about his experiences in monitoring wildfires and mountains with wireless sensor networks. In the wildfire deployment, wireless order carisoprodol sensor nodes were staked on poles six feet above a burning grass fire — and survived the fire! Turns out grass fires only reach 80 C or so that high above the ground. Professor Sitar’s next goal is to outfit an entire moutain face in Yosemite with wireless sensors!

So what were the common grounds between the two? Range — or more accurately how the actual operating range in the field didn’t come close to the data sheet 100’s of meters outdoor. Maybe a revisit is in order of the great work presented at Sensys 2003 about the gray areas of communication performance in wireless sensor networks as a function of distance. What happens to the range with different semi-permiable matter between two nodes, like smoke, vines, etc?