Development of hardware-software model for signal spectrum computation using Fast Fourier Transform based on FPGA

The object of research is the implementation methods of an adaptive hardware-software model for signal spectrum analysis using Fast Fourier Transform (FFT), implemented on a Field-Programmable Gate Array (FPGA) followed by processing in the software part. This solution combines the advantages of hardware acceleration and software flexibility. The proposed model is aimed at solving the problem of creating an efficient tool for real-time signal processing, taking into account limitations in accuracy, latency, resource usage, and data retention for further processing and analysis. The model is designed with scalability in mind, both in terms of increasing the number of processing channels and extending the FFT length and precision level. Its development included stages of modeling, synthesis, debugging, and testing close to real-world conditions. The structure of the model was thoroughly designed, data representation formats and rounding procedures were optimized, and the FFT algorithm was adapted to the specifics of the chosen platform. Altogether, this ensured high accuracy of spectral analysis and efficient use of FPGA resources, as confirmed by experimental data. Practical testing of the system in real time was conducted, during which such parameters as result accuracy and power consumption were evaluated, considering the efficient use of logic elements and memory blocks. The obtained results logically reflect the advantages of the hardware-software implementation, the usage of optimized data formats and rounding procedures, as well as the successful adaptation of the FFT algorithm. This allowed achieving a balance between high spectral analysis accuracy at the level of 3.97 kHz with an FFT length of 16,384, a twofold reduction in the required memory size, and a 0.25 ms decrease in FFT result transmission time. The practical applications of the developed model cover a wide range of fields, including embedded signal processing systems, modern real-time measurement devices, as well as mobile or energy-efficient systems, where real-time processing under low power consumption is critical. Thanks to its versatility, the model can be integrated into more complex digital signal processing systems, expanding their functionality.