Back to the Future By AL OPPENHEIM

Isaac Asimov once remarked that Bthe secret to being a successful prophet is to never try to predict something that will happen in your own lifetime.[ I hope that at least some of the predictions that I offer in this article are not either too naBve or too fanciful, but in any case, I take some comfort in Asimov’s comment. Looking back at the predictions for 2012 published by the Institute of Radio Engineers (IRE) in 1962, it is interesting to see how accurate many of them were, at least in broad strokes. Major advances in information storage and retrieval, processing of music and sound, the advances in automatic speech recognition were the topics of a number of the articles that seemed to anticipate the future, in many cases, very conservatively. In this article, I offer some speculation on where we might be in 2062, primarily with regard to two topics: the field of signal processing, and university education, building somewhat on the predictions made in 1962. Many of my speculations, I am sure, are obvious and will mature well before 2062. Others would seem to be reasonable speculations at this point but may be totally sidelined by unanticipated paradigm-shifting new technologies. With regard to signal processing, it is a certainty that Bthere will always be signals, they will always need processing, and new technologies and mathematics will always emerge for implementing the processing.[ In 1962, virtually all real-time signal processing systems were analog and any digital signal processing (DSP but it was not called that then) was for purposes of simulation to adjust the parameters for the analog implementation of a system or for offline processing of massive data sets in such application areas as oil exploration, military intelligence, and surveillance. This type of DSP required rooms full of large, expensive, energy-hungry mainframe computers. The Bbig bang[ in signal processing associated with the publication in 1964 of the Cooley–Tukey paper launched the use of the FFT as a major signal processing tool which fundamentally required implementation based on digital platforms. Furthermore, some of the new mathematics for signal processing that was emerging prompted the development of theoretically novel signal processing algorithms that were not realistically implementable in analog hardware. In that era, a fairly small group who was involved in the theoretical development of algorithms was somewhat naively speculating that Bsomeday[ integrated Digital Object Identifier: 10.1109/JPROC.2012.2204109