A Kind of Online Support Vector Machine for Blind Multi-user Detection

CDMA is a time-varying system. The background noise of channel and the relative position between base stations and users are continuous, the joining and leaving of users are stochastic. These factors result in that the properties of signal received by users are changing continuously. Based on this, we propose a online support vector machine for blind multi-user detection method. The main idea of this method is that the detector updates the training dataset periodically and retrains the SVM classifier with an incremental training algorithm. So we can trace the changing of the system and improve the performance of the signal detection.