Investigation of a VDHL implementation of the data preprocessing for a FMCW MIMO mm-wave camera

This thesis focuses on the resource consumption of transforming multiple data streams from the time domain to the frequency domain, where each data stream represents one sensor channel from a radar camera. The goal is to measure the resource consumption of processing said data streams in parallel and to investigate methods of reducing there source consumption. Real data streams are simulated in hardware and windowed in order to minimize spectral leakage. The windowed signal is then delivered to an Intellectual Property (IP) core to compute its fast Fourier transform which is then passed on for further processing. Finally the number of data streams and IP cores is scaled up so that the resource consumption and scalability of the project can be measured. The result of this thesis is that 23 sensor channels can be processed in parallel with the given IP core from Xilinx, but this would involve a lot of idle hardware. Three improvements are investigated leading to a 39 fold increase in throughput over the naive implementation. These savings free up resources for further digital signal processing, which is critical for real time imaging.