Abstract
Generally, for communication experiments, it is very important to build a channel model as close as possible to the real situation first. In past few years, many members of our team have already built a prototype of Ka-band channel model successfully and made many researches to upgrade the system performance by various coding schemes also. But, after starting the direct broadcast experiments via ROCSAT-1, we find that there are some discrepancies between our predicted channel model and the real channel. Hence, making the predicted model more realistic and increasing the coding performance is the main objective of the thesis. Because the original values of noise figure in our model are given by the constructor specification. We should modify the noise figure in RGT system by the actual measured values first. The difference between the original values and new values is about 20 dB. Second, since the satellite signals always suffer serious impairments and noises, we construct a new powerful FEC coding scheme, turbo codes, to enhance the coding gain from the old FEC coding structure. Although turbo codes have better performance, its decoding structure demands more complexity and takes more time. Hence, a modified structure in turbo codes is proposed to our Ka-band channel model. Third, after conducting experiments for a few months, we can make some discussions about data flow, system status, etc. from received data. Some suggestions and summary are also given in order to make the subsequent experiments more successful. Finally, we give an example to show the complete simulation and experiment flow and also present the comparison between measured power level and predicted power level. Conclusions and suggestions for future experiments will be given from all these results.