Kathy Cwikla, Secretary
Current Research
The Communications and Statistical Sciences Research Department was formed in October 2005 by merging the Mathematics of Communications Research Department and the Statistics and Data Mining Research Department.Communication Theory and Signal Processing
Our work in Communication Theory and Signal Processing spans a spectrum of goals, encompassing both fundamentals and practice. But in either case our motivations are to expand knowledge base and, with occasional exceptions, are not project specific. We are fortunate that some of our people have, in the past, been members of development organizations, and have kept ties to these organizations as they evolved into business units. This continuing interaction provides valuable input regarding the direction of our investigations.One major research goal is the search for fundamental limits, usually disregarding complexity and other practical issues, and is broadly described by Information Theory. Present efforts include difficult generalizations of point-to-point Shannon Theory to multi-access problems involving a sharing of a communication resource, with particular emphasis on wireless applications. Basic capacity studies of multiple antenna communication systems are included here.
In addition to this fundamental work we also engage in studies which propose or evaluate techniques, systems, or devices that are much closer to the practical art. Such investigations can be related to signal design, equalization, synchronization, performance evaluation, or communication theory for wireless systems. They can relate to new system proposals or existing infrastructure. The study of tradeoffs involved in choosing between equalization or coding to correct certain channel impairments on communication channels, or on recording channels modeled as communication channels, is part of this work. The search for constrained block codes that provide coding gain is of particular relevance to magnetic recording.
A very long standing project has been spectral estimation, the estimation of the power spectrum (perhaps short term) of time series. The theory includes non-stationary time series and non-uniform sampling; the applications are often to unusual and difficult situations. Spectral estimation techniques have been applied to global warming, black-holes, and so-called g-modes of oscillation of the sun. The latter modulates the solar wind, a flux of energetic charged particles capable of inducing damagingly high voltages on long ocean cables our in expensive communication satellites. Prediction of this phenomenon would enable advance action to prevent damage.