Today’s network intrusion detection system is plagued with working in a dynamic environment. These systems must be able to cope with detecting new and unknown attacks. Current research has shown that the use of anomaly-based multi layered feed forward neural networks can provide help. However, these systems have issues that need to be resolved. These issues involve the architecture and preparation of using these systems in a network. This paper will describe the problems, what has been done to overcome these issues and address any future work needed.