5. In generating a pattern, no matter what characteristics the pattern is designed to detect, categorize and compare, there are virtually no limits to its application so long as there are characteristics which can be used to define a relationship, or pattern. Technical applications are easiest to define and three will be offered here to demonstrate the application and adaptability of MESIA. The first being recognition of radar emissions based on their electronic characteristics, the second being mechanical trouble identification, failure reporting and failure projection in a powerplant and the third being monitoring of critically-ill or injured patients in an acute-care setting. 5.1 Current computer schemes to identify radar emitters are based on one of two concepts: matching a set of measured metrics (Radio Frequency [RF]), Pulse Repitition Frequency [PRF]), Pulse Width [PW]), Scan Type [ST]) and Scan Rate [SR]) to stored metric profiles; or matching the full analog signal (using Fast-Fourier Transform [FFT] algorithms) to stored analog profiles. The latter process is similar to the Electronic Warfare specialist identification by sound generated in receiver headphones by an incoming emitter signal. Both approaches have deficiencies: The metric system does not provide sufficient detail to reliably resolve ambiguities; and the analog system involves time-consuming database search and computer-intensive processing. MESIA combines these tehcniques by using the metrics (RF, PRF, PW, ST, SR) to establish groups (patterning of the database) of similar signals to limit search, and using the full signal profiles within the single group identified to resolve on an as-needed basis any ambiguity which might occur. The patterning scheme is based upon use of a five-field, two-digit, hexadecimal address to accomodate the five signal characteristics (RF, PRF, PW, ST, SR):  Since each field would have the possibility of 255 ranges, the total number of possible patterns (combined characteristics of all five fields) would equal 2555, or 1,078,203,900,000 - in plain language, over one trillion addresses because each field would be a subset of the previous field(s), which means that within each one of the 255 RF ranges there are 255 PRF ranges, and within each of 65,025 RF/PRF combination patterns there is a set of 255 PW ranges, and so on for as many characteristics as you would care to define. This first requires that ranges for each identifiable characteristic must be thoughtfully defined to accomodate the real-world dispersion of radar signal characteristics. Once the fields are properly defined and the signal is passed through the "filter", new incoming signals are passed through the same filter used to define the stored patterns, patterns are generated by the new incoming information and those patterns are then compared to the existing stored patterns. Ambiguities resulting from patterns too close to separate or missing information for a particular field can be resolved either by allowing the system to "guess" by disregarding missing information, or using some other separate unique characteristic to resolve the "guess." In the case of radar, ambiguities in identification might be resolved using the audio signature of the stored audio sample and comparing it to incoming signal audio characteristics, if available. Since the pattern-matching will narrow the number of possible matches, ambiguity resolution will be very rapid and can even be considered as a standard option. |