The US military is working on ways to improve security on wireless networks. It’s suggested adopting some of the tactics used by credit card firms and eBay.

The project is the work of the Defense Advanced Research Projects Agency, which is effectively the military’s gadgets division. It’s previously worked on making a robot that can (sort of) outrun Usain Bolt, as well as trying to figure out the techniques used by flying snakes.

The latest work is looking at the way military service staff use wireless devices to communicate in remote areas where there’s no reliable networks. These usually involve setting up direct links between individual devices and building an ad-hoc network. As Darpa puts it, the devices “not only provide access to the network; they are the network.”

The way this works means that the devices have to work in unison to work out the most efficient way to share frequencies, power and information. The big problem is that at the moment this only works because every device trusts the information it receives from every other device.

That in turn means that if a single device is captured by enemy forces, or its communications intercepted and manipulated, the entire network could be compromised. (Darpa also notes that a device being inadvertently misconfigured can also lead to disruption.)

Darpa is now calling for external proposals to develop ways to get round this problem. It’s key principle is that the network should use tools that assess each individual device (a node in the network) for trustworthiness. If the network detects something suspicious, it needs to reconfigure itself to stay active without risking security.

Wayne Phoel, who’s managing the program, notes that solutions could draw inspiration from existing business and social networks. For example, he points to the way credit card firms look for unusual patterns of behavior such as unexpectedly high purchases or a sudden change of location to trigger a security alert. He also pointed to the way buyers and sellers rate eBay individual transactions and build up a big picture assessment of whether a person is trustworthy.

(Image credit: Darpa)