In the years after World War I, commercial aviation was getting crowded. A excesse of post-war military aircraft being converted for more commercial use brought about a traffic density, causing a congested airscape and causing accidents, making the need for air traffic controllers.
Regulators and authorities responded by instituting air traffic corridors and beacon-based navigation systems—and eventually, around 1930, air traffic control towers began appearing in the UK. In 1935, the first Flight Monitoring Center appeared in Newark. It consisted of a clock, a notepad, some navigation charts, and a not-particularly-reliable radio setup. In a couple of decades there would be radar, too.
We can imagine the current state of drone affairs as analogous to the pre-ATC days of manned aviation. Suddenly, drones are cheap enough for hobbyists and advanced enough to conceivably be used for goods delivery, e.g. Amazon drones, pizza drones, etc. But a significant barrier remains in the form of drone traffic density. How do you keep your pizza drones from smashing into your news drones from smashing into your hobbyist drones?
As described in a paper published this week in IEEE Internet Computing, courtesy of Robert J Hall at AT&T Labs, the answer lies in an “internet of drones.” Drones can best avoid each other if they know about each other, and this is the essence of Hall’s Geocast Air Operations Framework (GAOF) prototype.
“The goal of the work is to demonstrate a path toward an improved system for the operation of drones, with the necessary secure command and control among all legitimate stakeholders, including drone operator, FAA, law enforcement, and private property owners and citizens,” Hall writes. “While today there are drones and drone capabilities that work well with one drone operating in an area using a good communication link, there will be increased challenges when there are tens or hundreds of drones in an area.”
GAOF is an extension of an existing AT&T technology known just as the Geocast System, which is being tested for similar traffic management applications on the ground, e.g. for people and cars. Obviously, adding a third dimension makes things a bit more complex.
The problem, however, is difficult whether we’re talking about driverless cars or drones for the simple reason that an ideal medium doesn’t exist for connecting them. Cellular networks are obvious, but these can change and drop out as a drone moves from place to place. Wireless ad-hoc networks, where drones connect to each other peer-to-peer, are an obvious solution, but carry with them the limitation of not being connected to the greater internet. This prevents the drones from accessing information about drones that are not within line-of-sight (and so can’t be reached with a peer-to-peer wireless network), and general information about local airspace restrictions, should they exist.
The Geocast system works by automatically flipping between the two network tiers depending on availability. Crucially, should it only be able to access a an ad-hoc network, it can gain access to the greater internet via packets relayed via other drones.
“So, for example, if one drone has both tiers available, it can act as a relay, transferring long-range packets into the short-range tier (or vice versa), so that a single-tier drone can receive messages that come via the relay from sources on the other tier,” Hall explains. “This is useful in many scenarios; for example, one could orbit a two-tier-capable drone at higher altitude above an area of operations in a valley where cell coverage was nonexistent, allowing remote awareness and control of drones operating at lower altitude within the valley.”
The drone traffic control problem gets still more interesting. How in the first place can we send packets to a three-dimensional geographical space—that is, every drone within that space—rather than to a list of specific drone IP addresses?
The answer is in geographic addressing (GA), in which circles centered around different latitudes and longitudes are assigned their own address, which is shared among all drones within that circle. Every device that wishes to monitor an area comes with up a query message, which is then sent to a specific geographic address. The drones within that address region send their replies back to the geographic address of the querying drone. All of the said drone’s neighbors get the reply, but if they know they didn’t send the query, it’s easy enough to just ignore it.
(This is a lot like how subnets work on the internet: Your computer sends and receives the same internet traffic as everyone else sharing the same network access point, it just filters it all down to the stuff meant for you.)
As Hall explains, this sort of collision avoidance could be extended beyond other drones and to buildings and towers and other things to be avoided. It’s just a matter of outfitting them with a beacon connected to a geographic address.
“This prototype system has been implemented and tested using simulated drones,” Hall notes. “Aerial field testing with real drones is being planned and will be conducted in accordance with the FAA guidelines.”