TCP Port reuse seems to be detected incorrectly as TCP.Split.Handshake

[Tristan_Cober]

Have a client who has been having a large number of Scans against their IP, and we are repeatedly receiving alerts about split handshakes (hundreds a day). I have looked at a packet capture on the WAN interface, and as far as I can tell these are false positives being triggered due to port reuse. At no point did the internal server on 184. reach out to 108. which would be required for the split handshake; it was only inbound scan traffic from 108. to 184. which should never trigger this signature.

Message meets Alert condition The following intrusion was observed: "TCP.Split.Handshake". date=2020-10-04 time=23:55:11 devname=<redacted> devid=<redacted> logid="0419016384" type="utm" subtype="ips" eventtype="signature" level="alert" vd="root" eventtime=1601880911201944543 tz="-0700" severity="medium" srcip=108.<redacted> srccountry="United States" dstip=(Internal of 184.<redacted>) srcintf="wan1" srcintfrole="wan" dstintf="lan" dstintfrole="lan" sessionid=43120 action="detected" proto=6 service="HTTP" policyid=5 attack="TCP.Split.Handshake" srcport=3658 dstport=80 direction="outgoing" attackid=26339 profile="default" ref="http://www.fortinet.com/ids/VID26339" incidentserialno=1351879586 msg="a-ipdf: TCP.Split.Handshake, TCP split handshake at state: ESTABLISHED" crscore=10 craction=16384 crlevel="medium"

Anyone experienced similar?