Certain investigations on modulation classification and power allocation for ofdm based cognitive radio networks

Abstract

In wireless communication, the utilization of resources is improved by adaptive resource allocation techniques. Due to additional interference constraints on the primary users in Cognitive Radio (CR) networks, the developed techniques for conventional wireless networks cannot be applied directly. Hence, the work in this thesis mainly concentrates on the resource allocation approaches. Recently, there is an increasing interest in CR based Orthogonal Frequency Division Multiplexing (OFDM) systems. It can achieve high data rate and also its flexibility in dynamic radio resource allocation. This research work deals with the following challenges: digital signal classification, subcarrier allocation, and power allocation in CR based OFDM systems. The digital signal system has two stages; feature extraction and classification. The temporal and spectral features are extracted in the former stage. Temporal features includes: mean, variance, skew and spectral features include spectral centroid, coefficient of variation and the spectral skew of the Intrinsic Mode Functions (IMFs). The IMFs are obtained as a result of the application of empirical mode decomposition on the modulated signal. The digital signal classification system is successful in classif ying three different modulation schemes. Our proposed digital signal classification successfully classifying various modulation schemes such as PSK, DPSK, 64 QAM, MSK and 256 QAM. It provides promising results with an overall accuracy of 97.6% (10 dB), 95.8% (5 dB), 80.2%(1 dB), and 53.6% (0 dB) using nearest neighbour classifier with Euclidean distance measure. newline