Modulation classification (MC) is of large importance in both military and commercial communication applications. It is a challenging problem, especially in non-cooperative wireless environments, where channel fading and no prior knowledge on the incoming signal are major factors that deteriorate the reception performance. Although the average likelihood ratio test method can provide an optimal solution to the MC problem with unknown parameters, it suffers from high computational complexity and in some cases mathematical intractability. Instead, in this research, an array-based quasi-hybrid likelihood ratio test (qHLRT) algorithm is proposed, which depicts two major advantages. First, it is simple yet accurate enough parameter estimation with reduced complexity. Second the incorporation of antenna arrays offers an effective ability to combat fading. Furthermore, a practical array-based qHLRT classifier scheme is implemented, which applies maximal ratio combining (MRC) to increase the accuracy of both carrier frequency offset (CFO) estimation and likelihood function calculation in channel fading. In fact, double CFO estimations are executed in this classifier. With the first the unknown CFO, phase offsets and amplitudes are estimated as prerequisite for MRC operation. Then, MRC is performed using these estimates, followed by a second CFO estimator. Since the input of the second CFO estimator is the output of the MRC, fading effects on the incoming signals are removed significantly and signal-to-noise ratio (SNR) is augmented. As a result, a more accurate CFO estimate is obtained. Consequently, the overall classification performance is improved, especially in low SNR environment.
Recently, many state-of-the-arts communication technologies, such as orthogonal frequency division multiplexing (OFDM) modulations, have been emerging. The need for distinguishing OFDM signal from single carrier has become obvious. Besides, some vital parameters of OFDM signals should be extracted for further processing. In comparison to the research on MC for single carrier single antenna transmission, much less attention has been paid to the MC for emerging modulation methods. A comprehensive classification system is proposed for recognizing the OFDM signal and extracting its parameters. An automatic OFDM modulation classifier is proposed, which is based on the goodness-of-fittest. Since OFDM signal is Gaussian, Cramer-von Mises technique, working on the empirical distribution function, has been applied to test the presence of the normality. Numerical results show that such approach can successfully identify OFDM signals from single carrier modulations over a wide SNR range. Moreover, the proposed scheme can provide the acceptable performance when frequency-selective fading is present. Correlation test is then applied to estimate OFDM cyclic prefix duration. A two-phase searching scheme, which is based on Fast Fourier Transform (FFT) as well as Gaussianity test, is devised to detect the number of subcarriers. In the first phase, a coarse search is carried out iteratively. The exact number of subcarriers is determined by the fine tune in the second phase. Both analytical work and numerical results are presented to verify the efficiency of the proposed scheme.
|
![[NJIT logo]](http://archives.njit.edu/vhlib/images/njit_logo.gif){kind=logo}
![[Van Houten Library logo]](images/vhl-top-image.jpg)
