Basically, there are two approaches to network intrusion detection: signature detection, and anomaly detection. The signature detection approach utilizes well-known signatures for network traffic to identify potentially malicious traffic. The anomaly detection approach utilizes a previous history of network traffic to search for patterns that are abnormal, which would indicate an intrusion. How can an attacker disguise his buffer overflow attack signature such that there is a greater probability of his attack going undetected by the IDS?