Research

Our research work is based on the following standard methods of signal processing and system modeling:

• signal representations, transforms

• analysis, synthesis, and implementation of digital filters

• analysis, synthesis, and implementation of adaptive algorithms

Resonator-based (signal model based) signal processing is a prominent research and development field of the laboratory. Our work is based on earlier results of the department. The main idea of the model based signal processing is that the signal processing is effective if the procedure comprise a model on the signal generator. On this basis resonator-based filters and the adaptive Fourier analyzer have been developed. Our activity in the field of analog to digital and digital to analog converters (including delta-sigma converters) reflects our knowledge in measurement, electronics and signal processing.

Research of acoustic applications of digital signal processing is the specialty of our laboratory. Active noise control was one of our first research topics. Active noise control is applied mainly for the cancellation of acoustic disturbances. This kind of noise control utilizes loudspeakers which generate an anti-noise canceling the original noise by destructive interference. We have developed a resonator-based method for suppression of periodic noise. The new algorithm is more advantageous in many ways, than the usual methods applied for arbitrary waveforms. Based on the results with periodic noise control, new algorithm has been developed for stochastic noise, as well. Recent research topics in this field are the on-line identification of the acoustic transfer functions and the application of sensor networks for active noise control.

Based on the knowledge picked up in acoustics and having some students' interest, our research was extended to the field of digital sound synthesis by consequently utilizing signal processing algorithms. Hence we have successfully synthesized the sound of the organ by an extended additive synthesis. Parts of the method were used for the synthesis of bells. For other instruments, like piano or violin this method cannot be used, since the sound of the instrument is a function of many parameters. We have found that physical modeling is the proper sound synthesis approach for such instruments. The latest quest is the investigation of geometric nonlinearities of strings, responsible for the longitudinal vibrations.

The most recent field of our research activity is connected to sensor networks and distributed signal processing systems. Nowadays wireless networks are deployed not only for communication but for data acquisition or control purposes. The autonomous operation of the units (motes) and the unreliable connection of them calls for the reconsideration of sampling, quantization etc. The distribution of the network means advantages and disadvantages, as well. In our laboratory such questions are also investigated, recently an active noise control system has been built with wireless sensing.