Wireless Standard Classification in Cognitive Radio Networks Using Self Organising Maps

TY - CHAP

T1 - Wireless Standard Classification in Cognitive Radio Networks Using Self Organising Maps

AU - Baban, Shaswar

AU - Holland, Oliver

AU - Aghvami, Abdol-Hamid

AU - Denkovski, Daniel

AU - Gavrilovska, Liljana

PY - 2013/8

Y1 - 2013/8

N2 - Radio access technology (RAT) recognition can be an important technique to facilitate spectrum sharing, interference avoidance, and cooperation among cognitive radios. As one example among its very many possible uses and benefits, RAT recognition might allow Secondary Users (SUs) to differentiate between the transmissions of Primary Users (PUs) and other SUs, such that SUs might contend for a spectrum band fairly, only not transmitting when they detect the PUs as having started to transmit in the same band. In this work, a practical testbed made up of software defined radio transceivers and computing units has been assembled, and used to transmit and receive extensive samples of representative RATs. A Self-Organizing Map (SOM) with Support Vector Machine (SVM) clustering and classification technique has been developed via a semisupervised learning, to operate on these received samples. Finally, performance metrics have been presented showing almost 100 % classification performance at -20dB Signal-to- Noise Ratio (SNR). This demonstrates the efficiency of this technique.

AB - Radio access technology (RAT) recognition can be an important technique to facilitate spectrum sharing, interference avoidance, and cooperation among cognitive radios. As one example among its very many possible uses and benefits, RAT recognition might allow Secondary Users (SUs) to differentiate between the transmissions of Primary Users (PUs) and other SUs, such that SUs might contend for a spectrum band fairly, only not transmitting when they detect the PUs as having started to transmit in the same band. In this work, a practical testbed made up of software defined radio transceivers and computing units has been assembled, and used to transmit and receive extensive samples of representative RATs. A Self-Organizing Map (SOM) with Support Vector Machine (SVM) clustering and classification technique has been developed via a semisupervised learning, to operate on these received samples. Finally, performance metrics have been presented showing almost 100 % classification performance at -20dB Signal-to- Noise Ratio (SNR). This demonstrates the efficiency of this technique.

M3 - Conference paper

SN - 9783800735297

SP - 115

EP - 119

BT - Proceedings of the Tenth International Symposium on Wireless Communication Systems (ISWCS 2013)

PB - IEEE

CY - Ilmenau, Germany

ER -