Detail publikace

Feature-Based Classification of Digital Modulations Using Various Learning Algorithms

Originální název

Feature-Based Classification of Digital Modulations Using Various Learning Algorithms

Anglický název

Feature-Based Classification of Digital Modulations Using Various Learning Algorithms

Jazyk

en

Originální abstrakt

The paper deals with classification of digital modulations by means of ten characteristic features of modulated signal and four learning algorithms, namely Artificial Neural Networks, Support Vector Machines, k-Nearest neighbors, and Random Forests. 2ASK, 2FSK, 4FSK, MSK, BPSK, QPSK, 8PSK, and 16QAM modulations were chosen for classification. Testing of the methods was carried out by simulation with signals disturbed by multipath fading and additive white Gaussian noise. It was found out that the Random Forests algorithm provides best results with over 99 % accuracy.

Anglický abstrakt

The paper deals with classification of digital modulations by means of ten characteristic features of modulated signal and four learning algorithms, namely Artificial Neural Networks, Support Vector Machines, k-Nearest neighbors, and Random Forests. 2ASK, 2FSK, 4FSK, MSK, BPSK, QPSK, 8PSK, and 16QAM modulations were chosen for classification. Testing of the methods was carried out by simulation with signals disturbed by multipath fading and additive white Gaussian noise. It was found out that the Random Forests algorithm provides best results with over 99 % accuracy.

BibTex


@inproceedings{BUT75345,
  author="Anna {Kubánková} and Radim {Burget} and David {Kubánek} and Artem {Ganiyev}",
  title="Feature-Based Classification of Digital Modulations Using Various Learning Algorithms",
  annote="The paper deals with classification of digital modulations by means of ten characteristic features of modulated signal and four learning algorithms, namely Artificial Neural Networks, Support Vector Machines, k-Nearest neighbors, and Random Forests. 2ASK, 2FSK, 4FSK, MSK, BPSK, QPSK, 8PSK, and 16QAM modulations were chosen for classification. Testing of the methods was carried out by simulation with signals disturbed by multipath fading and
additive white Gaussian noise. It was found out that the Random Forests algorithm provides best results with over 99 % accuracy.",
  address="Brno University of Technology",
  booktitle="The 13th International Conference on Research in Telecommunication Technologies RTT - 2011",
  chapter="75345",
  howpublished="electronic, physical medium",
  institution="Brno University of Technology",
  year="2011",
  month="september",
  pages="1--4",
  publisher="Brno University of Technology",
  type="conference paper"
}