System and Algorithm Concepts for Automatic Detection and Classification of non traditional signals

A technical approach to exploiting a specific type of non-traditional signal is proposed. This method is highly motivated by the signal properties of a class of signals known as C2 signals. A front to back detection and classification processing chain is discussed which exploits four properties of the signals. These are 1) signal periodicity, 2) the time-frequency nature of each pulse, 3) the duration of each pulse, and 4) the length of time from the beginning of one pulse to the other. The periodicity is detected via the auto-correlation of the signal envelope. The time-frequency information is analyzed by computing a 1 28x 1 28 time-frequency representation of the signal, then performing a 2-13 wavelet transform, and finally utilizing a small subset of the wavelet coefficients as classification features. These features are input to a neural network which is then used as a classifier. Gating functions based upon the final two properties are used to determine if the signals fall with acceptable ranges of signal duration and period. Much of the software for this approach already exists. PSI has extensive experience in transient analysis, and the technical approach is based upon intimate knowledge of the signals to be exploited. Vortical flows are essential features of most aerodynamic flowfields and generally determine flight vehicle maneuverability, aerodynamic load 4 ng and performance. However, our current understanding of the mechanisms of vortex viscous-inviscid interactions is insufficient to allow computational design of aerodynamic and structural design without recourse to experimental verification. At, present, significant efforts are being made to effect design changes which will decrease dynamic loading and improve fighter aircraft agility, maneuverability and performance. But, although significant progress has been Made in computational aerodynamics and associated structural response, reliable design changes cannot be made without recourse to extensive wind tunnel and flight testing. To alleviate this problem, it is proposed to develop innovative vortex measurement and control techniques and to conduct a series of experiments which will provide a detailed vortex flowfield modeling data base which could be used to guide and assess the performance of computational codes.