First, by measuring the hopcounts
during normal times, HCF creates a mapping of IP addresses to hopcounts.
Then, if an attacker sends a spoofing packet to the host, it is likely the
hop-count of the packet will not match the expected hop-count for packets from
the spoofed source address. Because legitimate hop-counts may change due to
routing changes, strictly filtering all packets that do not match would lead to
false positives. In order to minimize false positives, HCF only begins filtering
traffic if some threshold amount of packets do not match their expected hopcounts.
This threshold protects against mistakenly filtering legitimate packets,
but it also makes HCF ineffective against low amounts of spoofing packets that
do not reach the threshold. Furthermore, because the range of expected hopcounts
on the Internet is narrow, around 10% of the spoofing packets can be
expected to have the correct hop-count [Jin et al. 2003].
An earlier paper [Templeton and Levitt 2003] also covered many of these
active and passive host-based methods, and they have remained largely unchanged.
For more detailed descriptions we refer readers to Templeton and
Levitt [2003].