Digital Enhancement and Multirate Processing Methods for Nonlinear Mixed Signal Systems

PhD Student 
David Schwingshackl


 In this thesis digital enhancement methods and efficient implementation techniques which are applicable to a wide class of nonlinear systems are presented. A new polyphase representation is derived especially for nonlinear multirate filters. These polyphase realizations feature one fundamental property: all operations can be performed at the low sampling rate. Consequently, the computational complexity can be drastically reduced, saving silicon area and power dissipation. The developed theory is illustrated by means of examples and several applications including nonlinear echo cancelation as well as pre- and postdistortion of nonlinear systems. A further application is the linearization of single tone transmitters where the presented methods operate in both the time domain and the frequency domain. In addition, the topic of discretization of continuous-time nonlinear systems is addressed, which can be seen as a pre-requisite for digital enhancement methods in mixed signal systems. It turns out that even exact sampled-data representations of continuous-time systems can be derived. Furthermore, the discretization technique is generalized for almost arbitrary hold functions which are used in the digital to analog conversion process. This concept of general hold functions relaxes the typical constraint that the input signal to the continuous-time system must be held constant over the sampling period.  


This thesis is supervised by Gernot Kubin.