Wednesday, June 11, 2008

Attempt to Classify Early Warning Systems

Over the past 2 months, since my work completed as the researcher/project manager evaluating a last-mile hazard information dissemination research, a contract received through LIRNEasia, I've been taking a stab at classifying early warning systems (EWS). The research so far does not reveal a concrete abstraction for this classification effort. Most of the work done are domain specific, thus financial specialist trying to classify financial EWS, engineers trying to classify engineering EWS, so on and so forth.

The question that pops in my mind is "how does one distinguish between two similar EWS designs for the same purpose; i.e. pick the best?" or "how does one enumerate the capability and capacity of a given EWS?" or "how does one decompose an existing design to depict the possibility of adding on to extend the value to service other risks?" Current thinking is to design systems to the decision makers liking. Also there is no regard for including response systems in the design. As I see a EWS designed without taking the "customer attribute"; i.e. the response system in to consideration, is like throwing darts in to an empty space; i.e. no target.

The works so far leads me to believe that three main parameters that can classify any EWS, whether it be natural (as in the animal kingdom), engineered, social, or economic, are by understanding them through observer-controller (predictor corrector) systems, complexity theory, and Markov processes. These three primary fields give us the tools to define the operational orientations, capabilities of the design, and the expected capacity in real conditions.

I am testing the above mentioned framework on four examples: community-based last-mile hazard warning system (I was personally involved in), debt crisis financial EWS, Dam failure (safety) EWS, and a EWS based on the Traceability of Agriculture markets. The classification scheme inclusive of the enumeration theories are working in my favor. Although the exact simulated values for the system design capabilities and expect capacity are yet to be determined.

I would greatly appreciate anyone working in the same arena or has any interest in discussing the aspects of solving this classification problem to share their opinions with me through dialog.