The adoption of digital sound broadcasting systems, which are under testing in the country, allows new studies aimed a better planning for the implementation of new stations, which means to reassess the major existing radio propagation models or propose new alternatives to meet demands inherent in digital systems. The current models, as Recommendations ITU-R P. 1546 and ITU-R P. 1812, do not match closely with the reality of some regions of Brazil, especially in the tropical regions, such as the Amazon Region, due to the high rainfall and the vast existing flora. Using models suited to the propagation channel, it becomes feasible to develop planning tools covering most accurate and efficient. The use of these tools is applicable both to ANATEL, for the elaboration of the basic plans, as distribution channels for broadcasters.
This paper presents a methodology using a computational intelligence based in Bayesian Networks for prediction of electric field intensity, which can be applied to planning or expanding coverage areas in broadcasting systems for frequencies in the range of medium wave (300 kHz to 3 MHz). This methodology generates electric field values estimated from the values of terrain altitude (through analysis of conditional probability tables) and provides a comparison of these values with the measured electric field.
The data used in this study were collected in Brazil’s central region, nearby the city of Brasilia. The transmitted signal was an AM radio signal transmitted at a frequency of 980 kHz. With the data collected during the measurement campaigns, simulations were performed using conditional probability tables generated by Bayesian Networks.
Thus, it’s proposed a method for predicting values of electric field based on the correlation between the measured electric field and the altitude through the use of artificial intelligence. Compared to numerous studies in the literature that have the same goal, the results found in this study validate the use of the methodology to determine the electric field in medium wave radio broadcasting using Bayesian Networks.
|
