Development of MATLAB program/mobile applications for frequency response data modelling

Abstract

In modern electrical communications systems and signal processing applications, random electrical noise exists in all forms of medium. A method used in analysis and system modelling work best with noiseless date, but due to the direct impact of electrical noise, it affects the system’s bit error rate or signal to noise ratio. With the growth in complex modern Radio Frequency (RF) systems, analysing the system has become increasingly difficult. The critical question is how to eliminate or reduce the noise capture in the RF systems to mathematically model the system accurately and use the theoretical model to simulate the system techniques characteristic. Prony method and Matrix Pencil are used to understand if improvement could be made and if the analysis is possible in using a mobile device application. The report has demonstrated the use of different modelling methods in system identification data modelling. Out of all the methods, Frequency Domain Prony Method and Matrix Pencil method are the only two methods have produced the most or close modelling with consistent result. Although the accuracies of both methods have a very close relationship with the actual signal, the Matrix Pencil method seems to be slightly less sensitive to noise.