TCP/IP Stack Fingerprinting
What You'll Learn
You'll learn how subtle differences in how operating systems implement TCP/IP networking can reveal the exact OS running on a remote system—without needing a login or software banner.
The Detective Work of Network Stacks
Every operating system (Windows, Linux, macOS, BSD) implements the TCP/IP protocol suite slightly differently. Think of it like handwriting: everyone writes the same letters, but subtle differences in how they form those letters reveal their identity.
When a computer sends network packets, it makes tiny implementation choices:
- TTL (Time To Live): Windows typically uses 128, Linux uses 64, older systems might use 255
- Window Size: The initial TCP window size varies by OS and version
- TCP Options: Different OS versions support different options (timestamps, selective acknowledgments) in different orders
- Flag Combinations: How systems respond to unusual flag combinations (like SYN+FIN together) differs
- ICMP Responses: Error message formatting and even whether to respond at all varies
How It Works
Tools like Nmap's -O option send carefully crafted packets—sometimes with unusual or edge-case flag combinations—and analyze the responses. By comparing the pattern of responses (window sizes, TTL values, option ordering) against a database of known OS behaviors, the tool can identify the operating system with remarkable accuracy.
For example, a response with TTL=64, window size=5840, and specific TCP options in a certain order might clearly identify a Linux 2.6 kernel.
This works even when banner grabbing fails because it exploits lower-level implementation details that administrators rarely think to disguise.
Key Takeaway:
TCP/IP stack fingerprinting identifies operating systems by analyzing subtle implementation quirks in network packets—like TTL values, window sizes, and TCP option patterns—that differ between OS vendors and versions.