Spectral Correlation Based Blind Automatic Modulation Classification Using Symbol Rate Estimation

Azril HANIZ; Minseok KIM; Md. Abdur RAHMAN; Jun-ichi TAKADA

doi:10.1587/transcom.E96.B.1158

Spectral Correlation Based Blind Automatic Modulation Classification Using Symbol Rate Estimation

Azril HANIZ, Minseok KIM, Md. Abdur RAHMAN, Jun-ichi TAKADA

Full Text Views

0

Share
Cite this

Summary :

Automatic modulation classification (AMC) is an important function of radio surveillance systems in order to identify unknown signals. Many previous works on AMC have utilized signal cyclostationarity, particularly spectral correlation density (SCD), but many of them fail to address several implementation issues, such as the assumption of perfect knowledge of the symbol rate. In this paper, we discuss several practical issues, e.g. cyclic frequency mismatch, which may affect the SCD, and propose compensation techniques to overcome those issues. We also propose a novel feature extraction technique from the SCD, which utilizes the SCD of not only the original received signal, but also the squared received signal. A symbol rate estimation technique which complements the feature extraction is also proposed. Finally, the classification performance of the system is evaluated through Monte Carlo simulations using a wide variety of modulated signals, and simulation results show that the proposed technique can estimate the symbol rate and classify modulation with a probability of above 0.9 down to SNRs of 5 dB.

Publication: IEICE TRANSACTIONS on Communications Vol.E96-B No.5 pp.1158-1167

Publication Date: 2013/05/01

Publicized

Online ISSN: 1745-1345

DOI: 10.1587/transcom.E96.B.1158

Type of Manuscript: PAPER

Category: Wireless Communication Technologies

Cite this

Copy

Azril HANIZ, Minseok KIM, Md. Abdur RAHMAN, Jun-ichi TAKADA, "Spectral Correlation Based Blind Automatic Modulation Classification Using Symbol Rate Estimation" in IEICE TRANSACTIONS on Communications, vol. E96-B, no. 5, pp. 1158-1167, May 2013, doi: 10.1587/transcom.E96.B.1158.
Abstract: Automatic modulation classification (AMC) is an important function of radio surveillance systems in order to identify unknown signals. Many previous works on AMC have utilized signal cyclostationarity, particularly spectral correlation density (SCD), but many of them fail to address several implementation issues, such as the assumption of perfect knowledge of the symbol rate. In this paper, we discuss several practical issues, e.g. cyclic frequency mismatch, which may affect the SCD, and propose compensation techniques to overcome those issues. We also propose a novel feature extraction technique from the SCD, which utilizes the SCD of not only the original received signal, but also the squared received signal. A symbol rate estimation technique which complements the feature extraction is also proposed. Finally, the classification performance of the system is evaluated through Monte Carlo simulations using a wide variety of modulated signals, and simulation results show that the proposed technique can estimate the symbol rate and classify modulation with a probability of above 0.9 down to SNRs of 5 dB.
URL: https://globals.ieice.org/en_transactions/communications/10.1587/transcom.E96.B.1158/_p

Copy

@ARTICLE{e96-b_5_1158,
author={Azril HANIZ, Minseok KIM, Md. Abdur RAHMAN, Jun-ichi TAKADA, },
journal={IEICE TRANSACTIONS on Communications},
title={Spectral Correlation Based Blind Automatic Modulation Classification Using Symbol Rate Estimation},
year={2013},
volume={E96-B},
number={5},
pages={1158-1167},
abstract={Automatic modulation classification (AMC) is an important function of radio surveillance systems in order to identify unknown signals. Many previous works on AMC have utilized signal cyclostationarity, particularly spectral correlation density (SCD), but many of them fail to address several implementation issues, such as the assumption of perfect knowledge of the symbol rate. In this paper, we discuss several practical issues, e.g. cyclic frequency mismatch, which may affect the SCD, and propose compensation techniques to overcome those issues. We also propose a novel feature extraction technique from the SCD, which utilizes the SCD of not only the original received signal, but also the squared received signal. A symbol rate estimation technique which complements the feature extraction is also proposed. Finally, the classification performance of the system is evaluated through Monte Carlo simulations using a wide variety of modulated signals, and simulation results show that the proposed technique can estimate the symbol rate and classify modulation with a probability of above 0.9 down to SNRs of 5 dB.},
keywords={},
doi={10.1587/transcom.E96.B.1158},
ISSN={1745-1345},
month={May},}

Copy

TY - JOUR
TI - Spectral Correlation Based Blind Automatic Modulation Classification Using Symbol Rate Estimation
T2 - IEICE TRANSACTIONS on Communications
SP - 1158
EP - 1167
AU - Azril HANIZ
AU - Minseok KIM
AU - Md. Abdur RAHMAN
AU - Jun-ichi TAKADA
PY - 2013
DO - 10.1587/transcom.E96.B.1158
JO - IEICE TRANSACTIONS on Communications
SN - 1745-1345
VL - E96-B
IS - 5
JA - IEICE TRANSACTIONS on Communications
Y1 - May 2013
AB - Automatic modulation classification (AMC) is an important function of radio surveillance systems in order to identify unknown signals. Many previous works on AMC have utilized signal cyclostationarity, particularly spectral correlation density (SCD), but many of them fail to address several implementation issues, such as the assumption of perfect knowledge of the symbol rate. In this paper, we discuss several practical issues, e.g. cyclic frequency mismatch, which may affect the SCD, and propose compensation techniques to overcome those issues. We also propose a novel feature extraction technique from the SCD, which utilizes the SCD of not only the original received signal, but also the squared received signal. A symbol rate estimation technique which complements the feature extraction is also proposed. Finally, the classification performance of the system is evaluated through Monte Carlo simulations using a wide variety of modulated signals, and simulation results show that the proposed technique can estimate the symbol rate and classify modulation with a probability of above 0.9 down to SNRs of 5 dB.
ER -