ECE: Electrical & Computer Engineering
ECE News

Hunting in the Wild
MANET

Print Version

image of article

.pdf Print Version
260 KB

More Information


For more information, visit the Official website of the MANIAC Challenge .

Stimulating interest

A key goal of the MANIAC Challenge is to spur interest in the field among graduate and undergraduate students.

Armed only with laptops, intrepid student explorers hunted this past November in what was possibly the world's most uncooperative ad hoc network &mdash and still managed to bag an elusive, quarry: a six-hop route.

Undergraduate and graduate student teams gathered from seven universities to generate a first-of-its-kind mobile ad hoc network (MANET) and to compete for glory in its uncharted territory. The Mobile Ad Hoc Networking Interoperability And Cooperation Challenge (MANIAC) Challenge was organized by ECE's Luiz DaSilva and Allen MacKenzie and held in conjunction with the IEEE Globecom 2007, in Washington, D.C. It was the first of at least two such competitions, funded with a three-year grant from the National Science Foundation (NSF).

Teamwork

A key goal of the MANIAC Challenge is to spur interest in the field among graduate and undergraduate students.

Teams competed from Auburn University, Bucknell University, The George Washington University (GW), the Technical University of Kosice in the Slovak Republic, the University of North Carolina at Charlotte (UNCC), The University of Puerto Rico at Mayaguez, and Virginia Tech.

The competition's goal was to generate interest in the field among students, while also providing one-of-a-kind opportunities to study actual, uncontrolled, ad hoc networks, where users make their own decisions regarding tradeoffs between self-interest and common network goals.

Although the technology is imminent, MANETs do not currently exist outside of tightly controlled laboratory environment and military deployments, according to MacKenzie. "Questions linger about how well MANETs will work in the wild. Are simulation results reported in the literature too optimistic about performance that can be achieved in these networks?" he said.

The MANIAC Challenge was the first large, multi-hop MANET that was spontaneously formed in a natural habitat, according to DaSilva. "We have a need in this field for data from an ad hoc network that is not controlled by any single research group. We want to see what happens when we don't control every node and the different nodes may have different, and probably conflicting, interests," he said.

Strategic selfishness

The November competition focused on cooperation in routing and packet forwarding. Cooperation is one of the biggest issues surrounding MANETs, DaSilva said. For a stable MANET to exist, all nodes must cooperate, but still retain some selfish behavior in order to achieve their own goals. "Will users of wireless ad hoc networks trade off bandwidth, signal strength, or speed to ensure system effectiveness? If so, how? What incentives will get users to provide services &mdash such as forwarding and routing &mdash to other nodes?"

Each team controlled two laptop nodes on the network. MANETs are based on the premise of users sporting different hardware and software, so the only requirements were that the laptops have 802.11 capabilities and ran the MANIAC API and OLSR as the routing protocol.

Computer
Pointing

Strategy played a large role in the competition and teams were evaluated on the creativity of the strategy employed. As a group, the participants wanted to create a robust MANET and hoped to spot elaborate, multi-hop routes. However, each individual team wanted to acquire the highest number of intended packets.

To encourage both cooperative and selfish behavior, MANIAC teams were awarded 10 points for every packet they received that was intended for them, but every packet they forwarded for another team cost them a point. Source nodes for each team were not part of the cooperative network, but issued packets and sniffed the network to capture its topology.

The strategies were predominantly selfish and ranged from extremely simple to very sophisticated. Most teams tried to drop packets that would give other teams points, but maintain their reputation so that other nodes would still send them packets.

"We were evil. We dropped every packet with a final destination for the next node."

Other strategies included physically positioning their nodes close to sources, limiting the proportion of packets sent to any single node, and being unfriendly to nodes not in their direct path from the sources. "This was probably the unfriendliest ad hoc network in the world," commented DaSilva at the conclusion of the contest.

"We were evil," admitted a member of the Auburn team. "We dropped every packet with a final destination for the next node. Dropping next-node packets was a popular strategy with the other teams as well.

"Live and let live" wins for creativity

Mirror

The team from Kosice won first place in creativity for their "live and let live" strategy; "Their strategy was ingenious," commented MacKenzie. The Kosice strategy included manipulating the routing protocol itself to prevent traffic for other teams from being directed towards their devices at all.

"We were hoping to see the teams come up with fairly simple strategies that when you put it all together would lead to reasonably efficient outcomes for the whole" MacKenzie said. "They didn't do the things we expected, but that is part of the fun of it."

"Everybody decided to be a bad guy," DaSilva said. "This confirms the prisoner's dilemma." The prisoner's dilemma is a classical game theory situation in which two players who may cooperate for gain, or betray the other to go free typically choose to betray each other, even though this result is inefficient for both parties.

But did it work?

Organizers

Allen MacKenzie (left) and the MANIAC Challenge organizing team were looking to encourage student interest while creating and studying a live MANET testbed.

During the competition runs, teams roamed throughout the lower two floors of the Washington Hilton, while the non-network source nodes remained stationary. Many teams roamed, looking to be near certain sources or just for better network placement.

The network had the common stresses of any wireless network with interference and non-users trying to gain access. In spite of the hostility of the nodes, a complex topology appeared, with the greatest hops measured at six, which is extremely rare.

"I'm surprised it worked," DaSilva said. "So many things can go wrong &mdash these networks are pretty fragile. This was less fragile than I expected. This doesn't mean they are robust networks, however."

The UNCC team was almost always the multipoint routing node and took first place in the performance runs. The Bucknell team &mdash the only team of undergraduates &mdash came in second.

"The most fun was to see people actually enjoying participating in an experiment of this type," said DaSilva. "Getting students and faculty excited about something like this is an intangible, but it's very important in moving the research forward."

He cited evidence of growing enthusiasm among the participants. "We had undergraduates here and their advisor reports there is now a higher probability they will continue in the field and perhaps get a higher degree." Also, the Kosice team has written a conference paper on their strategy and participation.

DaSilva and MacKenzie are preparing for MANIAC Challenge II, which will focus on power control and spectrum usage.

Data from both events will be available to researchers at
www.maniacchallenge.org.