2 sonuçlar
Arama Sonuçları
Listeleniyor 1 - 2 / 2
Yayın Automatic modulation classification for mimo systems using fourth-order cumulants(IEEE, 2012) Mühlhaus, Michael S.; Öner, Mustafa Mengüç; Dobre, Octavia Adina; Jkel, Holger U.; Jondral, Friedrich K.Automatic classification of the modulation type of an unknown communication signal is a challenging task, with applications in both commercial and military contexts, such as spectrum surveillance, cognitive radio, and electronic warfare systems. Most of the automatic modulation classification (AMC) algorithms found in the literature assume that the signal of interest has been transmitted using a single antenna. In this paper, a novel AMC algorithm for multiple input multiple output (MIMO) signals is proposed, which employs fourth-order cumulants as features for classification. First, perfect channel state information (CSI) is assumed. Subsequently, a case of more practical relevance is considered, where the channel matrix is unknown and has to be estimated blindly by employing independent component analysis (ICA). The performance of the proposed classification algorithm is investigated through simulations and compared with an average likelihood ratio test (ALRT) which can be considered as optimum in the Bayesian sense, but has a very high computational complexity.Yayın A low complexity modulation classification algorithm for MIMO systems(IEEE-INST Electrical Electronics Engineers Inc, 2013-10) Mühlhaus, Michael S.; Öner, Mustafa Mengüç; Dobre, Octavia Adina; Jondral, Friedrich K.A novel algorithm is proposed for automatic modulation classification in multiple-input multiple-output spatial multiplexing systems, which employs fourth-order cumulants of the estimated transmit signal streams as discriminating features and a likelihood ratio test (LRT) for decision making. The asymptotic likelihood function of the estimated feature vector is analytically derived and used with the LRT. Hence, the algorithm can be considered as asymptotically optimal for the employed feature vector when the channel matrix and noise variance are known. Both the case with perfect channel knowledge and the practically more relevant case with blind channel estimation are considered. The results show that the proposed algorithm provides a good classification performance while exhibiting a significantly lower computational complexity when compared with conventional algorithms.
