{"title":"Least Square-SVM Detector for Wireless BPSK in Multi-Environmental Noise","authors":"J. P. Dubois, Omar M. Abdul-Latif","country":null,"institution":"","volume":20,"journal":"International Journal of Electronics and Communication Engineering","pagesStart":1692,"pagesEnd":1698,"ISSN":"1307-6892","URL":"https:\/\/publications.waset.org\/pdf\/4231","abstract":"Support Vector Machine (SVM) is a statistical\r\nlearning tool developed to a more complex concept of\r\nstructural risk minimization (SRM). In this paper, SVM is\r\napplied to signal detection in communication systems in the\r\npresence of channel noise in various environments in the form\r\nof Rayleigh fading, additive white Gaussian background noise\r\n(AWGN), and interference noise generalized as additive color\r\nGaussian noise (ACGN). The structure and performance of\r\nSVM in terms of the bit error rate (BER) metric is derived and\r\nsimulated for these advanced stochastic noise models and the\r\ncomputational complexity of the implementation, in terms of\r\naverage computational time per bit, is also presented. The\r\nperformance of SVM is then compared to conventional binary\r\nsignaling optimal model-based detector driven by binary\r\nphase shift keying (BPSK) modulation. We show that the\r\nSVM performance is superior to that of conventional matched\r\nfilter-, innovation filter-, and Wiener filter-driven detectors,\r\neven in the presence of random Doppler carrier deviation,\r\nespecially for low SNR (signal-to-noise ratio) ranges. For\r\nlarge SNR, the performance of the SVM was similar to that of\r\nthe classical detectors. However, the convergence between\r\nSVM and maximum likelihood detection occurred at a higher\r\nSNR as the noise environment became more hostile.","references":null,"publisher":"World Academy of Science, Engineering and Technology","index":"Open Science Index 20, 2008"}