Radio Wave of the Future
Modeling of communications protocols helps the military design next-generation software-defined radios.
In 1997, the U.S. military set up the Joint Tactical Radio System (JTRS) to design the next generation of military communications. The core technology: radios that could simultaneously handle both voice and data transmissions and that could be upgraded to new communications capabilities with the simple addition of software. The multibillion-dollar program is now on the verge of fielding these radios throughout the military, thanks in part to work in Lincoln Laboratory's Communications and Information Technology Division, which is giving developers of the system early insight into how the radios will work in the field prior to their full deployment. The aim is to help the military develop practices and tactics for using these radios in combat and peacekeeping operations, as well as to pave the way for improvements in future generations of JTRS radios.
JTRS waveform protocols automatically adapt to changes in user locations and needs. The network in the right is partitioned into two halves where each segment is fully operational by itself. Once the two network partitions move toward each other, various JTRS protocols, including medium access control (MAC) and routing protocols, dynamically collaborate to form one fully connected network. The live simulation statistics depicted on the upper right corner of the network topology screens illustrate various JTSR performance metrics (such as routing convergence, routing overhead, throughput, and latency) during the simulation.
The idea behind JTRS was to replace the 25 to 30 different radio systems in use by the various branches of the military with fewer software-based radios that would operate over all of the military's radio-frequency spectrum allocations below 2 gigahertz. The radios would provide voice, video, or data communications. Each radio employs a complex set of protocol standards that govern their operation. In the parlance of JTRS, all the protocols that each radio must execute are called a waveform. The waveform software on each radio must run protocols to form networks, without access to fixed infrastructure such as cell towers or fiber-optic cables. Moreover, protocols must form these networks whether the radios are carried by planes in the air, vehicles on the road, or soldiers on foot. "The military needs anywhere, anytime connectivity," says Tom Macdonald, leader of the Laboratory's Wideband Tactical Networking Group.
In order to provide useful communication, each waveform's core protocols perform many complex tasks without involving the radio operators. This ability of the waveform to hide the complexity of creating and maintaining communication links allows operators to focus on their military mission. The core protocols in each waveform spell out the rules of the electronic "handshake" by which one device connects to another. They predict rates of errors and apply correction or mitigation techniques to combat errors introduced by noise in the radio channel or from adversary jammers. They figure out how to form networks with a changing number of other moving radios. They determine the best path to reach each other radio and which data is the most important to send immediately. And because these are military communications, the protocols have to be secure—a requirement that can involve encrypting the transmissions. To optimize performance, the system's designers need to know how well not only each individual protocol works, but how all the different protocols interact. Complicating matters, the protocols' behaviors affect the design of the devices themselves, from the shapes and size of the antennas, to the number of computer chips required, and to the type and life span of batteries.
One difficulty with figuring all this out, says Siamak Dastangoo, a specialist in communications and networking and the Laboratory's principal investigator on the project, is that a variety of government contractors are responsible for different parts of the system. A large system integrator like Boeing might design the overall architecture and oversee contributions from other industrial partners, like ITT Corporation, which might build the devices; a specialty company like BBN Technologies might design some of the waveform's networking protocols. And all this work may be going on simultaneously. "A lot of times the device is being designed as the protocols are being developed," says Dastangoo. This concurrent development can be challenging both for the protocol designers who may not know what the hardware is capable of and for the hardware designers who may not have all the details of the protocols that their device must implement.
This is where Lincoln Laboratory comes in. Under the direction of the JTRS Program Office, Dastangoo and his colleagues have built software models of all the protocols used in one of the JTRS key waveforms, so they—as well as device manufacturers and protocol designers—can run simulations and see how they function. These models allow the development of a system-wide perspective of how all the protocols interact. The models facilitate easy capture of diagnostic information during the design process. They also provide an ability to upgrade each component or protocol separately.
It may seem as if such models wouldn't be needed and that it would be ideal to test the actual protocols on the actual hardware. However, that would take a room filled with racks of routers, computers, and other equipment—a setup that would not only be expensive but also would be difficult to transport, making it hard to share among the different contractors that are developing the different constituent protocols and hardware components. Furthermore, it would take many years before all the protocols and hardware were sufficiently mature to build up this capability, and at that point it would be more difficult to go back and change each individual protocol. Lincoln Laboratory's work helps look at large-scale end-to-end network performance much earlier than waiting until all the hardware and software is complete. The models distill the very complex protocols to their very essence, in a compact form that could be run on a laptop. Dastangoo compares the work to taking 1000 pages of text and cutting it to 50 pages that still tell the whole story.
Using a commercially available network-modeling environment called OPNET, Dastangoo and his team create streamlined behavioral models of the protocols to simulate how a network performs under different conditions. They can see how protocols written by different companies interact with others and identify potential trouble spots. They can add or disconnect communications links to mimic moving radios in complex environments like city streets or dense forests. They can try different types of network traffic—voice, then text, then video, for example. "We can go in and pinpoint where the problems are," Dastangoo says.
To make sure their model is accurate, Dastangoo and colleagues use results from the actual protocols running on a few real radios to assess the fidelity of their computer models and tweak them where necessary. After this confirmation at small scale with real equipment, the simulations offer a cost-effective and simple way to investigate the performance of the much larger networks required in the field. Chris Burns of the Space and Naval Warfare Systems Center Pacific says this capability is an important outcome of the project. "That's the unique part where JTRS will really benefit going forward, to use models and simulation to save the government lots of money," he says. Burns is the modeling and simulation lead for the Airborne and Maritime/Fixed Station arm of JTRS, the group in charge of the radios that will be deployed on Army and Air Force aircraft and Navy ships.
A key benefit of the Lincoln Laboratory–generated models is that the government can freely distribute them to all the individual protocol and hardware designers so that everyone involved in the process can obtain more complete and more accurate information on the work being done by the other organizations. This early knowledge helps avoid problems when the different protocols and hardware devices are integrated. Dastangoo points out that Lincoln Laboratory was well suited to this role in providing early technical bridges between the different contractors because the Laboratory is an independent organization with a systems perspective. "We are ideally positioned to bring together all the information from different partners," he says.
The original project focused on one waveform, but having accomplished the modeling with the first set of protocols for that waveform, the team has been asked to tackle other waveforms. Dastangoo says it will take about two more years to develop and refine models for these additional JTRS waveforms. Lincoln Laboratory will then hand the models off to the government to maintain and update as needed.