Abstract: The traditional statistical testing based methods have the shortcomings of low efficiency and high false positives.To solve this problem,according to the characteristic of periodicity and short burst in LDoS flows,a detection method against LDoS attacks has been designed and implemented based on feature extraction using wavelet transform.The proposed method extracts five feature indices of LDoS flows through wavelet multi-scale analysis of network traffic.Experiment results show that the method,capable of detecting the LDoS attack,achieves high detection rate with low computation cost,and hence has good practical value.