Picture of a dart board and arrow. DART = Dynamic Address RouTing
- for scalable ad hoc and mesh networks

The Grand Vision

On the right, the grand vision behind the DART project is described through a little story. While it is only a work of fiction today, we believe that we are well on our way to make this particular story an everyday reality.


Long Story Short

We believe that local mobile as well as stationary networks can, perhaps even should, be owned and operated by their users, and run off of cheap commodity hardware, including laptops and PDAs. The goal of our project is to make this possible, and feasible, in very large scale.

For more info on our technical approach, please go to the main page.


Project Participants


Jakob Eriksson
Michalis Faloutsos
Srikanth Krishnamurthy


Publications


"Scalable Ad Hoc Routing: The Case for Dynamic Addressing "
INFOCOM 2004 (to appear).

"DART: Dynamic Address Routing"
ICNP 2003 (poster).

"PeerNet: Pushing Peer-to-Peer down the Stack"*
IPTPS 2003






* PeerNet is the old name for the DART protocol, which had to be changed due to people confusing DART with an application layer protocol.
One ordinary morning in an extraordinary future...

Having breakfast at home, you check your mail, and read today's news off your favorite news website. Although you are not aware of it, today you are getting your Internet connectivity from Sprint, the lowest bidder that matches your specifications. Yesterday, the local Starbucks provided you the very same service.

The distance from your home to today's provider is covered by a heterogenous mix of connections. For example, the laptop on your kitchen table connects to your home base station using a wireless connection. The base station, in turn, has a cable that leads to a switch shared by you and three of your neighbors. Dave, your neighbor to the east, has a directional antenna that points at his office, which in turn has a direct wired connection to Sprint. Sprint allows you to share that connection only when you have paid in advance for their service, since Dave's office is not interested in sharing its own bandwidth.

Fred, your neighbor to the west, has an omnidirectional antenna on his roof, which is picked up by a directional antenna in an apartment window two blocks away. This directional antenna is right next to a second identical antenna aimed at the local Starbuck's cafeteria. Through the assistance of these, and other, neighbors, you are able to connect to Starbucks, Sprint, Charter Cable, SBC and many other Internet service providers at your discretion. While other neighborhoods have chosen to buy and share connectivity from a single provider, most people in your community make use of market competition to ensure they get the best possible price.

As you are getting ready to take off for work, your PDA signals for your attention. It's your carpool organizer messaging you to tell you that they will be at your house in 10 minutes, and you better be out there and waiting for them by then, because they're already late. The message packet originated in their car, and was forwarded along the freeway by 7 other cars before it was picked up by a well placed omnidirectional next to the road. It was then sent along a wire to your neighborhood school, and forwarded over a few neighborhood chimneys until it finally reached the antenna on your roof. There was no monetary cost to transmit this message, because it was forwarded on a strictly cooperative basis. You send a message back telling them not to worry.

On your way to work,  that incredibly useful traffic avoidance applet on your PDA queries traffic sensors at a few crucial freeway interchanges to determine which route will be the best choice today. While the transportation authority supplied the sensors with wireless interfaces, and while they do provide them with necessary electricity, they rely on users to take care of disseminating the information. This tends to work well during rush hour (when it matters), but coverage is usually spotty after midnight. As you are driving along the freeway, a steady stream of packets are flowing along the road in both directions, occasionally hopping across your car's tranceiver, to help people avoid traffic jams, exchange insults or flattery, or politely ask permission to cut in front of someone. Sometimes the packets are simply trying find a paid, or free, Internet sink somewhere along the road, to get the latest Hollywood gossip off the web.

Occasionally, especially during the off-hours and in sparsely populated areas, Internet access is not available.  However, sending a romantic suggestion (by voice this time) to that attractive blonde in the blue sports car is still quite possible. Just like her instant and quite impolite rejection ("In your dreams, Honda boy!"), the sound of your embarrassed silence is locally routed through the cars in your vicinity as she speeds into the distance.

The Internet is, and will in all likelihood remain, the premier network for getting globally relevant information, and much of the neighborhood traffic is simply packets being forwarded to and from some Internet sink or other. However, local services are increasingly avoiding the additional cost and complexity of providing service over the Internet. After all, the local pizza joint has no interest in offering its low cost, medium size, two-topping products to people outside its service area. Similarly, while walking through downtown, users are able to bring up local information without paying for Internet connectivity, just like they can send an email to the mayor by finding a route to city hall through their neighborhood. In fact, in an effort to make government accessible, city hall also provides free connectivity to all levels of state and federal government, including police, fire, and emergency medical service.

In a clever marketing move, and at a heavily discounted rate, the local Starbucks provides connectivity to all other Starbucks cafeterias. In addition to bringing in customers that want to do high-performance video conferences with their friends and colleagues across the country (while sipping a sickeningly sweet caramel mochachino), Starbucks is able to provide a community service by connecting communities across the country.



The hardware to make all of this possible is here today. What is needed is a new, more agile and self-configuring software network architecture. That, and a slight attitude adjustment. The Internet is greatest thing ever, there is no doubt about that. However, today, it seems as though the Internet is the only network that could ever matter, and that all forms of networking aims solely at providing Internet connectivity. We argue that while the Internet is an immensely valuable source of information, there are countless applications and services that do not, in fact, need global connectivity. In fact, we believe that many of the possible applications of networking are being hampered today by the strong focus on Internet and global connectivity.

Certainly, if a local pub wants to make its beer selection browsable by patrons' PDA's, the most straightforward approach would be to put up a local base station (with or without internet access) and advertise the availability of the menu through the ether. Today, the standard way to provide such a service is to put a web page on a globally accessible web server, probably in an air conditioned server room far away. Then, the patron must purchase his/her own global internet connectivity from some third-party provider, assuming such connectivity exists in the area, and enter a global WWW address to access the menu. Not only is this unappealing system design, it is also a waste of resources, and an obstacle both for the barkeep as well as his customers.

With the proper software, we believe that community networks can be constructed at extremely low cost, essentially supported only by consumer owned devices. We believe that when wireless connectivity throughout communities becomes free and abundant, in communities ranging in size from a local neighborhood to entire cities or even countries, this will fundamentally change the nature of networking and its applications.