Signal Processing Between Theory and Applications

Abdelhak Zoubir
Editor-in-Chief, SPM

Recently, a colleague visited my performance and gave a talk on the performance of third-generation and fourth-generation wireless systems. The seminar included a video that described the setup of the laboratory and numerous real-data results. This is quite unusual when the talk is on research in signal processing for wireless communications and, in particular, when it is given to an audience working mostly on statistical signal processing. Rather than theoretical development, the talk focused on interesting experimental validation. Among other outcomes, it was shown that implementations of the current standards still operate about 10 dB away from the Shannon bound. This conclusion was only possible with extensive (and expensive) experimental work performed by a group of academics, engineers, and research students. It should be stressed, however, that the validation would not have been possible without a deep understanding of theoretical results in the area.

At departmental meetings, conferences, or just at informal gatherings, we frequently discuss among colleagues the trend of signal processing as a well-established (sub)discipline of engineering. A large part of published work in signal processing journals and conferences is on theoretical analysis and algorithm developments. This is of utmost importance as it is the case, for example, of deriving statistical performance bounds, and of devising new techniques for target tracking and detection, parameter estimation, high-resolution imaging, and direction finding, to mention a few. These theoretical developments, however, should serve clear applications, drawn from their specificities and learn from their competing constraints. In essence, the applications help mentor our thinking and guide our intuition, putting our contributions in context. The experiments associated with the applications are also important, as they allow the testing of existing approaches and the revising of the assumed models and fine-tuning their parameters.

Over the past few years, a new theory of compressive sensing has emerged. Unlike in current practice where we compress measurements after sensing, this new theory allows the signal to be sampled and simultaneously compressed at a greatly reduced rate. In recent years, we have witnessed a flood of research papers in signal processing journals and conference proceedings, and IEEE Signal Processing Magazine (SPM) has been the home of an outstanding, well-cited lecture note [1] that appeared in the July 2007 issue and also a special issue published in March 2008, which consisted of many well-cited articles on the topic. The new theoretical advancement requires that we also put into practice this wonderful theory. There have already been a few systems that were built based on this new paradigm, including the one-pixel compressive sensing camera [1], which has also been used for hyperspectral data acquisition [2]. There is surely more to come...

A high impact of research on societal development is fundamental, and signal processing has a lot to offer to achieve this goal. Since its existence, signal processing has revolutionized numerous areas in the society we live today. There is also more to come, and SPM is a key player in this enterprise. Past and present editorial teams of SPM have tirelessly strived to identify and develop key areas for future research, which have had a large impact on the scientific community and the Society at large. In particular, the special issues of SPM have been enlightening and motivating for new research directions.

For the Sep'12 issue of SPM, I invited Fulvio Gini, an energetic and dedicated area editor (Special Issues), to write a brief on his editorial work. (Please read Fulvio's article " Maintaining SPM’s High Standards " by following this link.) He articulates the mechanisms of maintaining high-quality SPM special issues and raises some challenges of the future.

Our highest mandate is first-class publications as feature articles, special issue articles, or in columns and forums. I wish to encourage more colleagues to get involved and participate; I welcome your suggestions for timely and most interesting topics for special issues, feature articles, or columns.

REFERENCES

[1] R. G. Baraniuk, “Compressive sensing,” IEEE Signal Processing Mag., vol. 24, no. 4, pp. 118–124, 2007.