Innovation Pipeline

Based on the number of construction cranes operating on UC Berkeley’s campus these days, there’s a lot of building going on. In the packed Sibley Auditorium, there was another kind of building going on. But this annual gathering of computer science and electrical engineering professors and students had more to do with building a picture of next generation technologies likely to shape the marketplace.

Like any lecture hall, there’s always bound to be a lot of hand waving going on, but the approach was grounded in a rich mix of academic insight based on research findings, and grounded in market realities. Also, it’s rare to see overt displays of irrational exuberance here. This affair was all about discussing the kernels of innovation upon which a new generations of university spin offs and start ups will find their place in the sun.

First up was professor Michael Franklin, who revved up his talk with a 40-year look back over the data revolution movement. From the 1960s mainframes to minis of the 1970s and the birth of micro computers in the 1980s – which ceded the throne to the dawn of the web, Franklin pinpointed how we’ve finally arrived at the latest episode: the wireless device generation. As the world shifts into the age of wireless devices, it brings with it a new set of challenges of bridging disparate data repositories across a widespread network.

In this distributed wireless infrastructure, there’s the overarching challenge of taming the flood of data. And there are two paths through that mess: one is providing summary information and the other is delivering data immediately. Franklin’s research focus on these issues have led him to help found in 2005 a start up called Amalgamated Insight, which has venture backing from Diamond Ventures and Onset Ventures.

Other efforts underway by Franklin and his grad students are the Dataspace project. That database effort has been zeroing in on the big headache of every expanding demands of “data everywhere” which has led to a wide range of efforts –none of which are focused on coordinated agenda. The most acute information management problem centers on a large number of diverse, interrelated data sources. The trouble is there’s not a convenient way to manage these dataspaces in integrated, or principled fashion. Franklin’s dataspaces project outlines a new agenda for data management, which entails the breadth of work going on in data management today along with other research objectives.

Tightly related to the exploding growth of data source are a new generation of wireless enabled network nodes. Kris Pister, who founded Dust Networks, gave a presentation on the future of wireless networks. The promise of an explosive wireless sensor market didn’t quite happen. Now there’s a more level expectation for the growth of these networks.

Among the good news is several of the challenges that have vexed wireless sensors or motes has been battery life, lack of standards and greater reliability. The key standards are taking shape IEEE 802.15.4; Wireless HART and ISA SP. There are better radios now for the sensors running 2.4 Ghz frequency and the power consumption has been reduced by 30X. On all three fronts, Pister laid out the progress from battery life reaching the five to ten years life expectancy and the requirement of four nines reliability. The two year outlook is that there should be a mote on a chip.

And at that time we should have stronger security for these new networking automatons. Add to that the push to reduce cost in manufacturing these autonomous networking devices. Industrial automation is the beachhead for wireless sensors. Despite the market being challenging, there’s a lot of strong opportunity ahead.

Closely tied to the future of wireless motes is the push to reduce the size of the transmitters. Miniaturization is the biggest bottleneck, said professor Clark Nguyen in his presentation on The Future of RF Micro-electromechanical systems. But the prospect of using MEMs to build these RF channel selection electronics is proving to have a strong future. And these channel filters tend to take up a great deal of the real estate in these miniature devices. And the prospect of more channels needing to be handled from data to video to voice.

Using the ultra-precision of MEMs technology and surface micromachining used in IC foundries, professor Nguyen and his researchers have been able to build 2 micron thick resonators to handle the challenges.

It was also another first for the event. This year, the Hass Business School got involved by teaming up with EECS to give semifinalists in the business plan competition a chance to present posters to 28 venture capitalists from a total of 22 firms which included 3i Technology Partners, Diamondhead Ventures, Hummer Winblad Venture Partners, Mohr Davidow Ventures (MDV), New Enterprise Associates (NEA), Onset Ventures, Sand Hill Angels and US Venture Partners.

When the Bears symposium comes around next year, the agenda is more than likely to give a candid picture as to whether the predictions about what’s next were good bets. Lee Bruno