In this experiment, we add real background traffic on top of single DoS attack traffic in experiment I. The setup is shown in Figure 4. The background traffic is collected from a research office at National Electronics and Computer Technology Center (NECTEC). Users in this office are 35 undergraduate students. All hosts are on the same broadcast LAN 100 Mbps. The captured interval time is two minutes. We apply the same thresholds from previous experiment I. Table II shows characteristics of captured traffic and accuracy of each sub-experiment, averaged after five repetitions.
From Table II, we found zero false positive in all but the port scan experiments, similar to the result of experiment I. It turns out that if traffic data contains real background traffic,
the false positive in case of port scan is less than that of pure attack traffic. This is because the background traffic interrupts the reverse flows of port scan, causing slower arrival of response packets.