Echolocating big brown bats use wideband FM signals in the 20-100 kHz band for biosonar. The bat's auditory system develops a cascade of representations for the broadcast and echoes that together solve several design problems in sonar in a way that is not mutually exclusive. The first problem has to do with splitting of energy in echoes returned from multiple-glint targets whose range extent of up to 5-7 cm leads to delay spread of 300-400 microseconds, which is much larger than the width of the output of a matched-filter (crosscorrelation receiver) for the bat's signals (XCorr width is ~12 microseconds for an 80-kHz bandwidth). When matched-filter outputs are used for detection, the energy in the multiple-glint returns does not sum to enhance echo SNR because the matched-filter outputs are too narrow. The bat's initial auditory representation, a kind of multichannel A-to-D code, is a spectrogram derived from the outputs of parallel bandpass filters, each of which produces an envelope for its output after half-wave rectification and lowpass smoothing. The integration-time of the spectrogram is 300-400 microseconds, so that the reflections from multiple glints do add together inside this window to enhance echo SNR. This first-stage time-frequency representation is followed by nerve spikes triggered at different threshold levels to represent the time-of-occurrence of the rising or positive-going slope of the envelope. Across bandpass channels, the instantaneous frequency of the FM sweeps for broadcasts and echoes is registered by the frequencies of the neural channels, while the times-of occurrence of those frequencies are registered by the latencies of the spikes. Pulse-to-echo delay is determined by coincidence-detectors operating off the outputs of neural delay-lines that store the FM sweeps of the broadcast and the echoes in terms of their spike-time representations. The precision of delay registration is at least 0.5 microseconds at reasonable SNRs, which locates the leading glint in each target to within about 0.2 mm in range and about 0.5 degrees in azimuth. This solves the second problem of providing a sufficiently accurate scale of delay to support subsequent representation of target shape. Information about the rest of the target is carried by the frequency spectrum of the echoes, which is deconvolved by a parallel delay-line and coincidence-detector system that locates other glints in the target on the range-crossrange plane. The fact that this parallel system is working in relative range-crossrange units with respect to the leading glint means that rotations of the target derived from different looks (i.e., broadcasts) can easily be combined to build up a picture of target shape from widely-separated locations along the bat's flight path. This solves the third problem of selecting a center of rotation within each target that serves as the reference point for testing whether successive views of the object show it as having a constant shape in spite of the rotated views.

• Chicken Cordon Bleu
• Baked Scrod with Lemon Butter

All selections are $18.00 with payment at the door. Students are half price.

(Required to properly allow set-up at restaurant)

Frank A. Tito or Sally Sutherland-Pietrzak

Office: 401-832-5090 Office: 401-832-8735

E-mail: titofa@npt.nuwc.navy.mil E-mail: sutherlandsa@npt.nuwc.navy.mil