And there are both answers. You have the FQDN also set somewhere on a local DNS server:
You told the UCM to use the FQDN for the external host, so when it sent the message to your computer it resolved the FQDN to the IP on your DNS server and inserted that local IP into the contact and connect headers which it sent to your computer which in-turn told your computer the IP and port to send back-
Once your computer received the 200OK, this is where the call became somewhat official and your computer started sending to the 192.168 address as it was instructed to. The UCM never saw the ACK as it was sent to the wrong IP and thinking that your computer missed the 200OK, it sent another 200OK which has the same local IP because of the FQDN, so it became a wash, rinse and repeat at which point after about 32 seconds of this, the UCM would assume the connection to your computer had failed and it would terminate the call.