The 1215 MHz to 1400 MHz band is important for radio astronomers to observe redshifted extragalactic hydrogen ionic (HI). Observations at these frequencies are complicated by radio frequency interference (RFI) from strong man-made transmissions such as the ARSR-3 Air Surveillance Radar. In this thesis, we characterize some data files recorded at the National Radio Astronomy Observatory (NRAO) at Green Bank, West Virginia, USA, where this RADAR system causes significant data corruption. Using this data, we present a blanking technique to separate RFI from cosmic signal. There are generally two blanking approaches, time window blanking and detected pulse blanking. Compared with time window blanking, the advantage of detected pulse blanking is that the loss of integration time is much less (i.e. less data is discarded). But some pulses fail to be blanked because they are too weak to detect. So in order to blank weak pulses, it is desirable to optimize detection performance. In this work, we will combine these two blanking techniques and present a new Bayesian algorithm which combines Kalman tracking with pulse detection. This new algorithm will help to locate the weaker or missed detections, so as to help improve the performance of pulse blanking.