The design of a microwave channel probing analyzer is presented and discussed. Special attention is paid to the characteristics of the transmission path to be measured; the choice of spread spectrum probing signal; the signal processing methods used; modulation, demodulation, and synchronization at the transmitter and receiver sites; the path model to be fitted to the measured data and finally, the impact of all of these issues upon the necessary hardware and software. Following a formulation of the channel-probing problem, transfer function regression techniques are reviewed to illustrate the advantages of periodic averaging over established correlation methods. The construction of an experimental sounder based on these methods is described. The equipment has been used to measure the time-variant frequency-selective channel dynamics of a short (158 km) troposcatter transhorizon link at a frequency of 11.647 GHz over a 31.25-MHz bandwidth. Off-line, nonlinear spectrum analysis, based on Prony's method and singular value decomposition, is used to estimate parameters of a high-resolution delay model that accurately reflects the underlying ray structure of the transmission path.

The authors describe the application of the singular value decomposition Prony (SVDP) algorithm to the analysis of the time-variant structure of a microwave line-of-sight multipath channel. The algorithm deduces the multipath ray delays and associated amplitudes from measurements made in the frequency domain. The capability of this algorithm to resolve closely spaced rays is demonstrated by analyzing both computer simulated data and field measurements. The authors present experimental results of the time-variant structure of the transmittance H(jw,t) and include the evolution in time of the amplitude, group delay, and the constituting rays thus giving direct insight into the underlying multipath mechanism of line-of-sight microwave links.>

This study is concerned with using modern spectral analysis techniques to derive estimates of multipath ray parameters (delays and amplitudes) from measurements of selective fading over a wide band at microwave frequencies. The applicability of the general category of autoregressive methods is addressed, and the specific method adopted, the Prony algorithm, is described. An illustrative example in which this is applied to a typical field measurement is followed by an investigation into the capabilities and limitations of the algorithm. A statistical analysis of the method has been carried out using a database of laboratory measurements. The Prony algorithm's ability to resolve delays to better than half the Fourier limit is shown, and its tendency to yield biased estimates is studied.>

The joint statistical analysis of duration D and intensity R of rain exceedances relevant to the design of microwave communication systems is considered. The data analyzed are derived from point precipitation-rate observations recorded over a period of 49 years in the city of Barcelona, Spain, using a Jardi gauge. The concept of event and the duration of a single exceedance of a threshold R within that event is introduced. The analysis shows the parametric R dependence of the annual average number of exceedances and of the average duration.>

This paper is concerned with the statistical investigation of a massive data bank of 49 years of rainfall rateRcontinuously recorded in Barcelona using a rain-rate gauge with ten seconds response time. With radio communications in mind, the paper addresses and reviews in detail: 1) reliable statistical model forR, 2) number of years required to obtain a database from which to derive a reliableR-distribution, and 3) the CCIR worst-month concept. The research has shown that the generalized Paretoaexp(-\muR)/R^{b}gives nearly perfect fit for all ranges ofRfollowed closely by the gamma distribution, and the simpler square root (R^{1/2}) normal distribution gives excellent fit too. The log-normal distribution was unsatisfactory forR \geq 60mm/h. The spread of the yearly distribution ofP(R)is cube root normally distributed([P(R)]^{1/3})and between 7 and 10 years are required before a reliable average distributionP(R)can be obtained. The study of theP(R)return time in years is also presented. High resolution ofP(R)is presented looking at the evolution of the annualP(R)in terms of the hourly and monthly contributing parts revealing statistical features such as the location in time of rain rates above 50 mm/h. Finally, the study shows that the calendar month contribution toP(R)remains at all times well below the synthetic CCIR worst month and recommendations are then given about its use.