Kick Server with Poor Network and Ban for X Time?

We have a geo-distributed cluster. Occasionally there is a local network issue with one server, that creates a situation where it constantly joins and is kicked from the cluster. Resulting in an effective cluster wide lock until the server is manually removed from the cluster by an admin. We have large Timeout values (30s,1M,1.5M) in order to deal with short term WAN issues, but I think it exaggerates this problem. Is there any settings that could help prevent a server from joining (minimum time visible) or ban a server for x time before accepting it again?
Any suggestions would be helpful, or maybe this is a consider for future development, some sort of live black list that is updated the cluster kicks a server, that it will not accept it for x time etc.

Sorry, I just realized that this should be in XtraDB Cluster, not the generic MySQL forum.

Unreliable network is not a good choice for synchronous replication.    But if you must I would consider having some external monitoring to decide when cluster is to be removed from the cluster and when it can rejoin. 

Thanks for the suggestion, yes, real time is not ideal. Unfortunately we went with it for a design and are locked in at this point. The system in question is in a major city in China. Not sure if it is truly unreliable networking, or some form of packet inspection etc. by big brother.