Belmont Club


Aviation Week describes the coming of pervasively distributed sensors, all linked together by a wireless network, to be everywhere yet nowhere: listening, watching, waiting, understanding. Best of all they’ll look like a fly, a spider or a blade of grass.

Gaggles of mechanical grasshoppers, flies, bees and spiders–each a relatively dumb creature–can be networked into very smart networks to conduct intelligence, surveillance and reconnaissance.

In the last decade, remote sensors arrays have been changing from somewhat obvious, hard-to-mask, mechanical objects to autonomous, self-propelled, insect-like devices that can climb walls or jump up stairs and then lie dormant until motion, noise or vibrations trigger their activation.

The grandchildren and great-grandchildren of “WolfPack”–a coffee-can size, air-dropped network of ground sensors–include fast-moving spiders, high-jumping grasshoppers, bees with detachable surveillance payloads and sensor-equipped dragonflies.

Development of BAE Systems’ WolfPack worked out the dynamics of connecting a series of low-cost, not-so-smart sensors to create a very smart network. That network could, for example, monitor and analyze nearby communications and map the information flow. It then could trigger electronic jamming or even the injection of a data stream of algorithms that captures low-power traffic, attacks communications protocol stacks and otherwise manipulates a foe’s flow of information. A second-generation WolfPack added a propulsion system to manipulate the modules and recharge the batteries.

“Advanced communications can move intelligence, surveillance and reconnaissance (ISR) around the battlefield in real time,” says Lt. Gen. Dave Deptula, the U.S. Air Force’s Deputy Chief of Staff for ISR. “These ISR sensors now are transformed into the nodes of a truly global, net-centric weapons system.”

All this is amazing. But consider what it implies: bandwidth dominance, the control of space, algorithms and operating systems and most of all the power to crawl into any abode and overhear practically any thought on the planet. If these sensor systems every come to pass, and there’s a strong probability they eventually will, what kind of world will it be and with what forces in contention?

“You want to monitor activity someplace, because you think bomb fabrication is going on,” Penkacik explains. “What would the swarm look like that you deploy to provide surveillance for a week? It would involve a flying robot with an imager that would perch on a building across the street. It would probably be in a sleep mode until an acoustic or seismic sensor on a crawling robot [inside the target building] detects a vehicle arriving. Through the ad hoc network, the sensors will wake up and begin the surveillance mission. You can track the activity and introduce additional robots into the swarm.”

Decentralized data fusion is being examined as a way to build even more intelligence into the swarm. If each robot provided a separate data stream, it would overwhelm the operators’ BlackBerry-size control devices. The effort to convert the mass of surveillance data into a refined stream of useful knowledge through distributed processing is being worked through BAE Systems Australia at the University of Sydney.

“Everybody does a little piece of processing and you may have larger algorithms executed by multiple micro-scale processors,” Penkacik says. “It’s analogous to using home computers to help solve enormously complex mathematical problems rather than buying a Cray [high-speed computer].”