Corrective codes for network nodes

	Zhiyuan Yan

The Internet and other communication networks that contain connecting nodes (terminals, computers or switches) have undergone a revolution in the past decade.
Improvements in speed and efficiency have resulted in greater throughput, or the transmission of more data, and in popular applications that use multicast addressing to transmit data from one node to multiple nodes. One example of a multicast takes place when multiple users simultaneously download the same movie from the same website.
Traditionally, says Zhiyuan Yan, an assistant professor of electrical and computer engineering, data transmission has occurred when each node in a communication network stores a packet of information and forwards it to the next node.
In a newer, more advanced method of data transmission, nodes combine multiple incoming packets into a single packet and pass it on. “Combine-and-forward” data transmission, also called network coding, results in significant performance improvements in multicasts.
But these improvements have a down side, says Yan. Errors due to noise, jamming, or interference can undermine the integrity of the data transmitted in network coding. This threat is particularly severe for wireless applications.
“Networks are very sensitive to errors,” says Yan. “A wireless environment is notoriously unreliable, and prone to jamming and interference.”
Yan and his students design codes to detect or correct errors undermining network coding. The field is highly theoretical and mathematics-based. Yan and his students utilize three branches of mathematics—abstract algebra, finite field theory, and combinatorics—to design classes of error control codes for network coding. They have led the way in adapting rank metric codes to error control in network coding. They also work with subspace codes and constant dimension codes.
“We take codes and design them for our purposes,” says Yan, who holds an M.S. in mathematics as well as an M.S. and a Ph.D. in electrical engineering from the University of Illinois at Urbana-Champaign.
“Each type of code has its own parameters. We are one of the few groups working with rank metric codes and subspace codes.”
Yan has become a pioneer in the field of error control for network coding. He and his students have published more than 50 articles in top journals and conference proceedings. At IEEE’s International Symposium on Information Theory (ISIT) this year in Seoul, South Korea, Yan presented three papers in this field. IEEE, the Institute for Electrical and Electronic Engineers, is the largest organization of its kind in the world. ISIT, held annually, is the flagship conference of the IEEE Information Theory Society and attracts roughly 1,000 researchers from all over the world.
Yan receives funding from the National Science Foundation, from industry and from the Pennsylvania Infrastructure Technology Alliance. He is spending this summer as a visiting researcher at the U.S. Air Force Research Lab in Rome, N.Y., where he applies his error control codes in network coding.
--Kurt Pfitzer