narayan1 -- December94
If information theory has been rumored to have copped it time and again, perhaps we should rejoice in its miraculous powers of resurrection. On the other hand, I believe that we are being unnecessarily defensive and should simply ignore the prophets of doom. Consider the typical doomsayer: His (her) impres- sions of information theory consist of ``forward and converse theorems", ``capacity being the maximum mutual information" and achievable after an eternity. Ah, well. Years of success stories including the Viterbi decoder, the Lempel-Ziv algo- rithm, modem design, advances in cryptography, signal processing techniques in remote sensing and radar based on the minimum length description principle, maximum entropy methods in signal processing -- all these are somehow ignored. And these happen to be some of the gains in the application fields alone to which information theory has contributed in a significant way, not to mention its contributions to statistics, ergodic theory, combinatorics, etc. Clearly, those who claim information theory is dead are dead wrong. Well, enough of this morbidity.

It is more constructive to address the question: What can be done to sustain and enhance the vitality of the field?

To this end, consider the fact that contributors to informa- tion theory come from a wide variety of disciplines: Engineers, statisticians, mathematicians and ergodic theorists, physicists, etc. As far as the mathematicians, statisticians and ergodic theorists are concerned, I am very optimistic as they constitute a body of strong researchers for whom information theory is indeed an application. I think we will always get exciting theoretical contributions from them. This is particularly true as new light is shed on interplays between these disciplines and information theory.

This brings us to the engineers who constitute a critical component. Information theory got off to a flying start thanks to a group of engineers who have also been responsible for some of its subsequently stunning success stories. We must ensure that new generations of engineers are motivated into embracing the discipline. And it is here, I believe, that significant reinforcement is needed. Emerging technologies call for a modification of some of our favorite paradigms in information theory. While this development surely does not detract from the significance of existing paradigms, there is clearly a need for new ones which address the fundamental limits of performance of emerging systems which are generally poorly understood. For instance, communication networks, e.g., those offering multi-media services, cannot be adequately addressed based on our usual paradigms involving simple sources and a com- munication link. Indeed, network information theory -- to the extent it has developed -- is inadequate too. Nor does a theory of queue- ing networks suffice. I believe that information theorists can play a very significant role in this field, and we have seen a few recent papers in this context. It is also true that most of the models we have studied thus far are fairly simplistic in the context of a network. There is a real need for tractable and realistic models of information sources as also network situations involving delays, asynchronism, etc. There have been some promising beginnings in this regard, but sustained coherent efforts by more people are need- ed. In summary, the engineering side of information theory needs bolstering.