| Popis: |
The system identification technique is a powerful real-time signal processing tool when dealing with the characterization of dynamic systems. This approach uses adaptive filtering techniques and the least mean square (LMS) algorithm to estimate the impulse response of a system in a finite impulse response (FIR) filter format. The main purpose of this paper is to compare the results and processing speed of two ARM-based microcontrollers as well as to give a theoretical background and an implementation procedure of the technique. ARM (Advanced RISC Machine) is an architecture of 32-bit processors usually applied in embedded systems due to its processing power, relatively low consumption, cost and small size. The impulse responses obtained by the method for several passive electronic filters were transformed to the frequency domain and compared with simulations of the circuits' frequency responses. This practice was repeated for both boards and for different parameters of the algorithm. The results have shown that the Arduino Due was unable to deal with higher sampling frequencies due to its processor limitations. On the other hand, the Teensy 3.6, a more powerful controller, yields great results even for higher frequencies.The system identification technique is a powerful real-time signal processing tool when dealing with the characterization of dynamic systems. This approach uses adaptive filtering techniques and the least mean square (LMS) algorithm to estimate the impulse response of a system in a finite impulse response (FIR) filter format. The main purpose of this paper is to compare the results and processing speed of two ARM-based microcontrollers as well as to give a theoretical background and an implementation procedure of the technique. ARM (Advanced RISC Machine) is an architecture of 32-bit processors usually applied in embedded systems due to its processing power, relatively low consumption, cost and small size. The impulse responses obtained by the method for several passive electronic filters were transformed to the frequency domain and compared with simulations of the circuits' frequency responses. This practice was repeated for both boards and for different parameters of the algorithm. The results have show... |
