| dc.description.abstract | As the communication services are increasingly demanding more access for higher frequencies up to Ka-band and beyond, a model to predict the propagation through rain is required in order to estimate the link budget and the communication performance. The rain drop size distribution (DSD) is the most important parameter in the rain attenuation prediction model. In this paper, we establish the DSD model from measurements, followed by presenting a rain attenuation model using the DSD model.
A two year observation (2002-2003) of rain drop size distribution using two two-dimensional optical distrometers at different location were recorded. The DSD were measured and analyzed for different seasons under various rain rates. The variability of DSD in both space and time was clearly shown even in the not so large area of North Taiwan. It follows that a relationship between rain rate and DSD was established. By applying statistical regression, it was also found that, in most cases, the DSD follows the Gamma distribution best. By applying three parameters (μ:mean, σ:standard deviation, : drop numbers per unit volume) into the Gamma model, the DSD model can be established.
Long-term rain attenuation measurements using a Ka band (28GHz) CW system at vertical polarization were conducted in northern Taiwan. An optical rain gauge, which has resolution of 0.01mm and can collect the rain rate every 5 seconds, measured the rain rate at the same location as the Ka band CW system. The attenuation due to the rain can be estimated by calculating the extinction coefficient over all of the rain drops within the antenna beam volume. Two methods were used to estimate the extinction coefficient. First, assuming that the scattering mechanism follows the Mie scattering approximation, and the rain drops are all sphere. Second, using T-Matrix method and the oblateness of the rain drops that vary from 1.0 to 0.8. Making use of the DSD model, a semi-empirical rain attenuation model was then developed. To validate the model, we compared it with the measured data. We further compared the results to the ITU-R and Crane models, and the comparisons show that the proposed model matched very well with in-situ measurement from a two-year data set. Both the Crane model and the ITU-R model were inadequate, as expected, for a correct interpretation of the accumulated measurement data producing overestimates. | en_US |