Improving accuracy in automatic modulation classification of digital modulated signals using design of experiment method
Loading...
Date
2018-06
Authors
Chan, Wui Hung
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
An automatic modulation classification (AMC) is a system is used to classify the
modulation format of a received signal. It is a system placed in between the receiver and
the demodulator. The AMC is crucial as the classification of received signal must be
reliable to ensure the received information is correct. Therefore, a lot of studies had been
conducted to look for the alternative for the improvement of classification accuracy of
the AMC system. In this project, asynchronous delay tap sampling (ADTS) is proposed
as a technique in modulation classification. From the ADTS, unique and distinct
asynchronous delay tap plot (ADTP) is generated for each of the QPSK, 16-QAM and
64-QAM digital modulated signal. These data are then reconstructed to become the input
of a built-in support vector machine (SVM) classifier in MATLAB. Design of experiment
(DoE) method is applied to improve the accuracy of the AMC system. In DoE,
2
2
factorial design method is applied. The two selected factors are the delay tap and the
sampling period used in ADTS. The results of the classification showed that the accuracy
of the classifier is 95.1%. Through DoE, the accuracy of the classifier using the optimum
values is 97.6%. This shows an improvement in the accuracy of the AMC system by using
the DoE method. In conclusion, the proposed techniques are fully capable of improving
the accuracy of the AMC system.