Signal adaptive hardware implementation of a system for highly nonstationary two-dimensional FM signal estimation

Signal adaptive, multiple-clock-cycle hardware implementation (MCI) of an optimal (Wiener) filter for highly nonstationary two-dimensional (2D) FM signals estimation is developed here. It uses results of the space/spatial-frequency (S/SF) analysis in real-time processing of nonstationary 2D signals and is based on the correspondence of the filter's region of support to the local frequency (LF) of the filtered 2D signal and on the S/SF analysis-based LF estimation. The MCI approach helps the proposed design to minimize clock cycle time and to optimize critical design performances related to the hardware complexity, making it a suitable system for real-time and on-a-chip implementation. However, the major advantage of the proposed design is the ability to take variable (signal adaptive) number of clock cycles in different S/SF points within the execution. This property helps the design to optimize the execution time (the main drawback of the classical MCI approaches in comparison to the single-clock-cycle ones), but also to provide the highest quality LF estimation, high S/SF resolution, and a very efficient filtering of nonstationary 2D FM signals. In this way, it is qualified as an optimal solution for wide range of practical implementations. The implementation is verified by a field-programmable gate array (FPGA) circuit design.