MOBILE armies need mobile communications. Those communications, though, must be secure—and not just from eavesdropping. They also need to be uninterruptible. And that is a problem. Many mobile networks (think Wi-Fi routers or mobile-phone towers) operate via hubs. Destroy the hub and you destroy the network. Even a peer-to-peer system in which messages travel in a series of hops between nodes (in the form of the devices that comprise the system) rather than via a hub, can be degraded by a loss of nodes. Existing versions of such systems, which are usually static, rather than mobile, require each node to be set up individually, in advance, to talk to particular other nodes. Mobility brings a need for constant reconfiguration.
In theory such a system is possible. It is called MANET, an approximate acronym for mobile ad-hoc network. In practice, though, a workable MANET has proved impossible to design—until now.
The main problem is mathematical. To avoid the pre-programming required by existing systems, a MANET must maintain routing tables that keep track of the shortest routes between nodes. Updating these tables takes geometrically increasing amounts of processing power as the number of nodes increases. A second, related problem is that to maintain the routing tables each node has to send regular “hello” messages to all the others within range. As the number of nodes increases, these messages rapidly multiply and crowd out other traffic. The upshot is that a MANET with more than about 30 nodes starts to experience problems, and one with more than about 50 will grind to a halt.
Faced with that, America’s Defence Advanced Research Projects Agency, DARPA, decided in 2013 to launch a challenge to build a MANET with more than 50 nodes, with the carrot of a juicy contract for the winner. The challenge worked. Earlier this year the American army tested a MANET with 320 nodes at its urban-training facility at Fort Bragg, North Carolina. This system can support the equivalent of a brigade-sized expeditionary unit, with voice, text and data traffic at up to 30m bits per second (bps). Each node is a hand-held unit like a chunky smartphone. During the test, users were distributed across dozens of buildings. To make things extra tricky, some of them were in basements.
The new MANET was able to perform so well because it overcame both the routing-table problem and the hello-message problem. Engineers at Persistent Systems, a firm in New York that designed it, simplified the routing task by employing tricks such as remembering the route taken to reach another node and reusing it, rather than working out the best path every time a message is sent. Another trick they perfected is “overhearing”. This occurs when a node happens to pick up a message it was not intended to receive, and can provide a shorter path than the planned one.
To deal with the hello-message problem, the packets that carry those messages are simplified. Louis Sutherland, Persistent Systems’ head of business development, likens the process to saying “hi” rather than “hello, how are you?”
Persistent Systems is not the only company trying to develop better MANETs. Thales, a European defence giant, has one it claims can support 150 nodes at speeds of up to 6m bps. TrellisWare, of San Diego, claims 8m bps over more than 200 nodes. Nor need MANETs be restricted to communications between people. They could enable robots, whether on the ground or in the air, to work in co-operative groups. Persistent Systems already has a contract to supply communications for the American army’s PackBot ground robots.
MANETs will also appeal to some groups of civilians. Rescue workers in places hit by natural disasters, where existing communications have been destroyed, will benefit. So may miners. A MANET can easily stretch from above ground into underground areas. Industrial sites where there is too much interference for conventional communications may be suitable for MANETs as well. Altogether, then, these particular MANETs may soon paint a picture of communications perfection.
Economist