Signal processing techniques have been applied recently for analyzing and detecting network anomalies due to their potential to find novel or unknown intrusions. In this paper, we present a novel network anomaly detection approach based on wavelet analysis, approximate autoregressive and outlier detection techniques. In order to characterize network traffic behaviors, we proposed fifteen features and applied them as the input signals in our wavelet-based approach. We then evaluate our approach with the 1999 DARPA intrusion detection dataset and conduct a comprehensive comparison for four different typical wavelet basis functions on detecting network intrusions. Our work aims to unveil a question when applying wavelet techniques for detecting network attacks, that is "do wavelet basis functions have an important impact on the intrusion detection performance?". Moreover, to the best of our knowledge, the work is the first to analyze the 1999 DARPA's network traffic using flow data instead of its original raw packet data.