Applied Systems Division
l Applied Mechanics Mechanical Design Information Systems Reliability and Human Engineering Electronic Systems Applied Systems Logistics Services
© 1987-2006 Mercer Engineering Research Center 135 Osigian Blvd. • Warner Robins, GA 31088 info@merc-mercer.org • www.merc-mercer.org • 478.953.6800 Figure 4 - Detected Pulses after Processing Figure 3 - Before Processing Figure 3 shows the spectrum at a very low SNR prior to any processing and Figure 4 shows the results after all processing is complete. Figure 1 - Graph Legend Figure 2 - Wavelet and Burst Processing The spectrum is displayed as shown in Figure 1 showing the pulses in a high SNR case. Figure 2 illustrates the steps of processing and the resultant outputs through different steps. After wavelets are utilized to quantify the energy, a determination must be made about the existence of a pulse. MERC used an unconventional method involving High Order Statistics to determine the Gaussian part of a scale to separate the pulsed energy from background noise. Wavelet algorithms decompose a signal into its simpler components (like an FFT) with a localization property that allows them to efficiently portray signals with discontinuities (unlike an FFT). Their multiresolution analysis separates a signal into fine and coarse resolution components. Low Power Detection Mercer Engineering Research Center has explored novel signal processing approaches to solve EW problems with artificial intelligence based algorithms within a digital receiver. Detecting low power RF pulses is typically achieved by improvement of hardware, an expensive alternative. Sensitivity improvements are achieved using wavelets and high-order statistical detectors, improving signal detection by more than 3 dB over the conventional technique, a PolyPhase Filter. The goals of this research were to detect and identify very low power pulsed signals and determine the Pulse Repetition Interval (PRI) and Pulse Width (PW) for identification.