Speeding Development of Radio Communications

4.29.2005 -- In the old days, all radio communication was implemented in hardware. But once the hardware was built, you couldn’t change the outcome. If you were disappointed with the resulting capabilities, you were stuck with them until the next design-and-manufacture cycle, and that could be a long time in coming.

“But we have other options now,” says Zhongren Cao, a UCSD postdoc from China. “The more you can minimize the impact of the hardware, the more flexibility you gain to enhance the functional capabilities. How do you do that? By moving more of the capabilities into software.”

Team in lab
Magnus Almgren, Thomas Svantesson, and
Zhongren Cao (top to bottom) in lab at UCSD

Cao is referring to a signal-processing radio platform that he and a team of seven other electrical engineers are working with. They are developing a straightforward methodology to easily adapt their theoretical ideas in software to the platform, shortening the time from concept to capability.

“We refer to the platform as a ‘software-defined radio,’” Cao says. “We can eliminate bugs and add features as they occur to us. The best part is that, as the designers, we don’t have to have all the answers up front. This methodology is much more forgiving and the process potentially more creative!”

“Our focus is on signal processing of the radio mechanisms,” he says. “We still need the analog front end with the antennas, but we’re writing the software to enable the platform to function like a radio.”

To get started, the team bought a hardware platform from Lyrtech to try out their ideas and learn how to apply it to their purpose. “This is just a lab platform for our research,” says Cao. “Our longer-term goal is to have an advanced signal-processing platform on which others can conduct research.”

They develop their ideas in Simulink, a graphical programming environment that enables modular code development. Simulink makes it possible to map basic hardware and software functionalities to representative icons. Design then becomes a relatively straightforward task of linking the icons graphically to achieve a certain capability.

Svantesson, Almgren, Cao
Svantesson, Almgren, and Cao

Simulink is useful because it can model many real-world applications with strict timing constraints. The timing requirement for future broadband wireless systems, for example, is in the nano-scale range: Signal processing has to be done within a few hundred nanoseconds. Any algorithm that cannot meet this timing requirement simply won’t work.

Team members code their algorithms, compile the programs, download the binaries to the digital signal processor, run the codes, evaluate the results, then return to Simulink to correct or enhance each software module in an iterative process.

In particular, the group is working on wireless systems with multiple antennas. “One of the hot areas right now in communications,” says Thomas Svantesson, a UCSD postdoc from Sweden, “is MIMO – multiple input, multiple output. The idea is that you can improve the transmission rate with more transmitters at the sending and receiving sites. But, of course, more is not always better, so we need to experiment to see where the tradeoffs lie.”

One of the reasons MIMO is so exciting is its potential for supporting wireless transfer of huge amounts of information, such as is found in video streaming and online gaming applications. The team also is exploring WiMAX, a promising broadband wireless architecture for both fixed and mobile users.

According to Ericsson’s Magnus Almgren, the senior member of the team, “My company is interested in the proof of concept of this methodology. Ultimately, Ericsson wants to be able to provide nearly all communications capabilities in software.”

“We see this as an exciting step forward,” volunteers Cao, “from writing theoretical papers to testing and acquiring real measurements on a prototype system.”

But publishing is still important, Almgren says. Interested in promoting the interplay between theory and practice, he wants to publish measurements taken from the prototype system.

“UCSD could become one of the few places,” Almgren says, “where people write theoretical papers and implement their ideas, then write more papers with the results from the prototype systems.”

The university is the perfect place to do this kind of work, he says, “because here you have knowledge and experience without limitations – in fact, you’re even allowed to fail.” Industry he points out, by contrast, tends to be focused on shorter-term goals and delivering products on time.

“In this case,” says Almgren, “one plus one is greater than two. Individually we’re smart, but it’s the interaction among us that creates the larger impact.”

This impact, he says, also arises partially out of the diversity of the team, which typifies Calit2 research projects: It includes UCSD faculty, postdocs, and industry partners from Ericsson and ViaSat, both paid and volunteer, representing five countries -- Greece, Sweden, China, India, and the U.S.

Almgren has designs on teaching a course related to the platform. “Teams of two could develop their ideas in Simulink,” he says, “download their software to the platform, and run some measurements. They could do most of the development work computationally.”

His long-range plan is to see students “get hooked and ‘grow up on’ this kind of system.” If they did this kind of work, they would better understand what a radio is all about. The result would be better radio channels and communications schemes.

This project began two years ago, arising out of a research review Ericsson had with UCSD faculty and students. Ever quick with an analogy, Magnus came to describe faculty research projects as “raisins” in the sense that the faculty are doing lots of great work, but many of the projects are disconnected from each other.

“We’re providing the cake mix – the radio platform – to which they can contribute their raisins, and we all benefit,” he says.

The development team includes
Magnus Almgren, Ericsson, Sweden
Zhongren Cao (Arnold ), UCSD, China
Nandan Das, ViaSat, India
Per G. Johansson, Ericsson, Sweden
Kostas Liolis, UCSD, Greece
Bhaskar Rao (professor of Electrical and Computer Engineering), UCSD
Thomas Svantesson, UCSD, Sweden
Nirmal Velayudhan, ViaSat, India