Get access

Transionospheric signal recognition by joint time-frequency representation

Authors

  • Shie Qian,

  • Mark E. Dunham,

  • Matthew J. Freeman


Abstract

A challenging problem in signal processing is the detection of impulsive signals which are dispersed by the ionosphere and received by RF satellite sensors. One example of this type of signal is the so-called transionospheric pulse pair (TIPP), thought to be associated with lightning discharges. Since these dispersed events occur sporadically and typically persist for less than 100 μs, the first objective is to detect or trigger on the signal of interest, thereby reducing the amount of data relayed back to Earth. Difficulties in detecting transionospheric impulses arise from the dispersion of the original signal, which lowers the signal-to-noise ratio (SNR), and the need for continuous, real-time observation. The energy content of TIPPs has been measured from 25 MHz to more than 100 MHz, requiring sampling rates of at least 150 MHz to capture the events. Although these transionospheric signals are easily recognized in the joint time-frequency domain, the computation involved in time-frequency transformations prohibits application to a real-time detector with this sampling rate. Using a joint time-frequency representation allows us to derive a robust time convolution template to fit transionospheric signals. This template can be used to design matched chirp filter banks, which are optimum with respect to SNR, for detection of transionospheric signals. We believe that a realistic matched filter bank could have a sparse spacing while maintaining good detection ability. This type of detector is also amenable to real-time analog hardware implementation in conventional finite impulse response (FIR) filter structures. Our work shows that transionospheric signals encountered in real data are effectively detected by matched filter banks. While a specific application is treated here, the methods described are general, and applicable to many problems in signal detection and recognition which treat nonlinearly dispersed signals.

Get access to the full text of this article

Ancillary