Irwin and Joan Jacobs
Professor of Engineering, Electrical and Computer Engineering
392 Rhodes Hall
B.S. 1962 (Yale); M.S. 1964, Ph.D. 1966 (Harvard)
Berger joined the Cornell faculty in 1968. During graduate study he held a Raytheon Advanced Study Fellowship, and subsequently was a senior scientist at Raytheon. At Cornell he is associated with the Center for Applied Mathematics and the graduate Field of Statistics. He was a Guggenheim fellow in 1976, a Japan Society for Promotion of Science fellow in 1980, and a fellow of the Ministry of Education of the People's Republic of China in 1981. In 1982 Berger received the Frederick E. Terman Award (which recognizes an outstanding young educator in electrical engineering) from the American Society for Engineering Education. He has served as a consultant to Raytheon, IBM, Schlumberger, Teknekron Communication Systems and AT&T Bell Labs. He is a fellow of the Institute of Electrical and Electronics Engineers, a member of the Governing Board and a past president of the IEEE Information Theory Group, and a past editor-in-chief of the IEEE Transactions on Information Theory. He is a fellow of the Institute of Electrical and Electronics Engineers and a member of the American Association for the Advancement of Science, the American Society for Engineering Education, Sigma Xi, and Tau Beta Pi. Berger directs the DISCOVER Lab, a research facility for video/audio compression development.
I conduct and supervise research in diverse disciplines including information theory, random fields, communication networks, video compression, signature verification, and coherent signal processing. I am particularly interested in situations in which information generated at several different locations must be transmitted over a network of communication links whose capacity is limited.
We have studied throughput versus delay tradeoffs and robust design techniques in packet communication systems. Other problems concern applying multiterminal rate-distortion theory and multiterminal decision theory to situations in which many remote, correlated sources are connected to a common processor via separate communication links. This work is of significance for multisite signal-processing applications such as interferometry, seismology, and emitter location. We have contributed to the problem of optimum diversity coding, known also as the multiple descriptions problem, in several cases obtaining exact results for the ultimate capabilities of such systems.
The subject of random fields, which is under intense investigation in mathematics and physics, is being studied from the viewpoint of information theory. Analysis, synthesis, simulation, and encoding of random fields are all under investigation. A rigorous basis has been provided by extending the fundamental block and sliding-block coding theorems for sources and channels to situations characterized by multidimensional parameters. Work on rate-distortion theory for non-Gaussian random fields has included the establishment of a critical distortion phenomenon for the Ising model in two dimensions. We are exploring ties with statistical physics, including the information-theoretic implications of phase-transition phenomena. In particular, we have extended the Shannon McMillan theorem to stationary and ergodic fields on trees and have determined the cardinality of phase transition of symmetric fields on closed trees.
Last Updated: January 23, 2001