Well, what first comes to mind is firewall configuration.
- Configure your firewall to DROP or REJECT all incoming requests FROM the offending IP(s) or IP-range(s).
- LIMIT the number of NEW connections (to the targeted port(s)).
- Configure your firewall to DROP or REJECT incoming UDP requests TO the targeted UDP port(s) (not a good idea if you have vital servers listening on any given port/interface.
Hard to give any practical advice with the little amount of information given.
You are likely to be given more advice if you provide information regarding your issue (circumstances, firewall software, extent of the attack, etc).
Good luck.