This paper describes a system that has made use of robust off-the-shelf sensor technologies by placing them together in an
array and using intelligent algorithms in a new and powerful manner to extract data that is of interest in
devising an early warning system for water security. The system has been shown to be effective versus a
wide variety of threat agents in a laboratory setting. The use of a unique system for estimating the baseline in
real world systems allows for the identification of small deviations from normal readings in water analysis
parameters. This in turn leads to a system capable of triggering on these deviations.
Once the system has been triggered, the algorithms have been shown to be capable of utilizing the unique
profile represented by a threat agent's deviations to identify that threat agent. Laboratory procedures on over 80 agents to date have shown no significant overlap of profiles. As the database grows, there may be some
overlap in the future, but it is likely that day-to-day plant events will not intrude on the vector space occupied
by agents of concern.
The system also has the capability to learn day-to-day deviations that are unique to a given system. Events
that occur commonly will be rapidly learned, and the rate of false positives to the trigger mode will rapidly
decrease. The systems ability to identify threat agents is not affected by learning and if a system is
compromised by a threat agent in the library the system should alarm and identify that threat from the first
day of deployment. Over all, the system is an invaluable security tool for recognizing system incursions, but
it has the ability to become much more than that. To date, most of the library work has been done on threat agents, but as time allows, the libraries
will be expanded to incorporate common distribution system problems that may arise. Also, as an individual
plant library recognizes deviations and the operators are able to identify them, the system will become a
useful tool in evaluating the day-to-day system health and operational parameters. Includes 28 references, table, figures.