Distributions of underdense meteor trail amplitudes and its application to meteor scatter communication system design

Abstract

Analysis of data from recent experiments leads to the observation that distributions of underdense meteor trail peak signal amplitudes differ from classic predictions. In this paper the distribution of trail amplitudes in decibels relative to 1 W (dBw) is considered, and it is shown that Lindberg's theorem can be used to apply central limit arguments to this problem. It is illustrated that a Gaussian model for the distribution of the logarithm of the peak received signal level of underdense trails provides a better fit to data than classic approaches. Distributions of underdense meteor trail amplitudes at five frequencies are compared to a Gaussian distribution and the classic model. Implications of the Gaussian assumption on the design of communication systems are discussed.

Distribution of underdense meteor trail durations and duty cycle and applications to meteor scatter communication system design

“…In previous work, Weitzen et al [1991] showed that the distribution for peak received power (in decibel Watts) from underdense trail reflections could be modeled by a normal distribution; for example,…”

“…The variance of log(c) is given in appendix A, and ranges between 1.1 and 2.0 for all ranges and frequencies considered. The variance of log(P) is tedious to determine analytically, but has been computed numerically for typical values of P dB and CrVdB, as provided byWeitzen et al [1991], and ranges between 0.001 and 0.002. Hence the distribution of log(t s) should be approximately normal, and the distribution for t s should therefore be lognormal.…”

“…Goodness of fit was measured using the Kolmogorov-Smirnoff (K-S) bound at an alpha risk of 0.05, described by Neter et al[1988]. (See alsoWeitzen et al [1991] for a general description of the process).Table 1provides a sum_mary of the best fit estimate of the distribution parameters for tB and ob at frequencies of 45, 65, 85, and 104 MHz, and for thresholds (P•) of-126, -116, and -106 dBm. The P gn• parameter is an estimate of the tail probability, since the data set was truncated, and only included shorter duration trails.…”

Meteor burst link and network simulator

“…The distribution of vpeak becomes approximately Gaussian [8]. The underdense time constant, t, is exponentially distributed.…”