Art of creating a Manet takes masterful engineering touch
Engineers face major architectural, design problems in moving from conception to reality
- By Barry Rosenberg
- Aug 03, 2009
Many people experience dropped cell phone calls, almost on a daily basis, while traveling through areas without a tower to transmit a signal. So imagine the reliability you would get with a communications network that doesn’t need towers because transmissions are automatically routed from sender to receiver via dozens of mobile nodes on the ground.
Such a mobile ad hoc network, or Manet, would be particularly valuable for completing communications transmissions, especially in conjunction with wireless mesh networks, where nodes are fixed, such as the Army’s Combat Service Support Automated Information System Interface communications system. As such, Manet technology would be particularly valuable during rescue operations for catastrophic natural disasters, such as Hurricane Katrina, and in military offenses, such as the Marine Corps assault in the Helmand Province in southern Afghanistan.
A Manet is an autonomous collection of mobile users with wireless devices that communicate via relatively bandwidth-constrained wireless links. Because the nodes are mobile — meaning that they are installed in moving vehicles or worn by soldiers or Marines — the network topology might change rapidly and unpredictably. The network is decentralized, and the nodes must execute all network activity, including topology discovery and routing functionality.
The eventual development of Manet technology is a high priority for military services, particularly for Army soldiers and Marines, both of whom constantly move on foot or in vehicles and in areas without a communications infrastructure. The Defense Advanced Research Projects Agency and Naval Research Laboratory have invested many years of research into developing the architecture for such systems.
“In the beginning, engineers only used Manet to talk about the subsystem routing part of a distributed, possibly mobile, wireless network,” said Joseph Macker, senior scientist and network researcher at the Naval Research Laboratory’s Information Technology Division. “It seems now that Manet is a general term for the whole mobile ad hoc network design problem. This is a broad problem and requires architectural approaches.
“Different missions, different platforms, different applications, different operational environments, heterogeneous wireless technologies and user group requirements all require design consideration," Macker said. "No one size fits all.”
Architecturally, the challenge for engineers and theoreticians is to understand how the Manet portions of the network connect and interact with the rest of the network, including sensor networks, backbone networks and ad hoc edge networks.
“This does not mean some components are not mature — although some are," Macker said. "It means we have a lack of architectural design and thinking. Having the right architectural components to pick and choose from with interoperability at key interface points is important.”
In the early days of Manet research in the 1990s, Macker said the handful of researchers who worked on the technology liked to think of Manet not as an acronym but as representative of the way that the French painter Édouard Manet would approach subjects in a loosely formed manner that wasn’t easy to define, particularly as art was transitioning from realism to impressionism. They liked to think of Manet not strictly as a mobile network but one that could work in conjunction with nonmobile wireless systems, particularly with mesh networks in an urban environment.
“A Manet is an extreme version of a mesh network,” Macker said. “You can make design choices in a nonmobile network that won’t work well in a mobile network. The fact that the nodes are moving creates added complexity on what you have to manage. You can use GPS or other approaches to optimize a mobile network, but it is more unpredictable when things are mobile.”
And that’s where additional theoretical research into Manet is needed.
“In this new context, you are connecting to a relay that is potentially unreliable, may or may not be there from one time to the next, or it might be a different type of relay at any one time,” said Michael Neely, assistant professor of electrical engineering at the University of Southern California and co-investigator on the joint multiuniversity and DARPA project named the Information Theory for Mobile Ad Hoc Networks.
“If there are only two nodes — one that transmits and another that receives — you can do the initial signaling and a simple handshake" between the two devices, Neely said. "That’s not the case if you throw in 20 to 50 nodes and all are interacting with each other, and the channels are changing and breaking, and there is interference because everyone is sending over wireless channels.”
In such a scenario, which Neely said has been simulated in networks with as many as 100 nodes, the network’s efficiency drops by as much as 99 percent because of the overhead protocols necessary to send and receive.
“That’s where the theory comes in,” Neely said. “A lot of advancements in theory suggest we can do a lot better,” adding that the purpose of programs such as the Information Theory for Mobile Ad Hoc Networks and another DARPA program, named Control Based Ad Hoc Networking, is to develop opportunistic learning in a mobile network environment. Those programs would be able to make decisions about how to transmit signals and intelligently adapt to a network of many nodes. “I wouldn’t say that we’re a long way away" from solving the key problems.
One of the transmissions technique that Neely said holds promise is the backpressure routing algorithm, which breaks an old model that there must be an established path from sender to receiver before a transmission starts.
“Not requiring a fixed path before you make a transmission decision can lead to significant gains,” he said. “You might not even know who your recipient is before you send a data packet. In a wireless system, that transmission is overheard within some radius region, and having a diversity of multiple receivers increases the possibility of success for people to do something valuable with the data.”
To complicate the job, there are many different radios and proprietary systems under which the radios operate, and interoperability is a significant hurdle.
“The problem with the military is that it is still buying all these proprietary things being pushed on us,” Macker said. “These radios have their own routing protocols, which hurts us in a lot of ways because it is difficult to evaluate them in many different environments. To do a good Manet, you need cross-layer solutions and radios with an open interface. There has been an evolution regarding open source, but it hasn’t gotten to the network level yet.”
Until now, the interfaces between the radio and network levels have been inefficient and decoupled, but that is expected to change with the ongoing development of programmable radios under the Joint Tactical Radio System program. One of the few radios with any Manet technology is the Highband Networking Radio, developed jointly by Harris and BAE Systems. Late last year, the Army said it would deploy the HNR radio to the 101st Airborne Assault Division’s 2nd Brigade Combat Team in Iraq.
Barry Rosenberg is editor-in-chief of Defense Systems. Follow him on Twitter: @BarryDefense.