Neural Networks and Smart Antennae : A Case Study

Abstract: This dissertation evaluates the artificial neural technique for evolving a smart antenna system. The AI techniques pose a challenging research in the field of communication. As such the antennas help to communicate with the digital processor to choose the desired signals and reject the others. It makes its own decision even to find the level of interferences and noises to be discarded by amplitude elimination process through the use of perceptron optimization algorithms like LMS (Least Mean Squares). This method helps to enhance the performance of signal processing efficiently. The design of hardware and software are quite complex. This is due to the fact, that the behaviour of the system is not fully understood being a real-time dependent system. This research work is carried only on software with certain simulated activity on beam-formation algorithm and as well, the system responses before and after using the adaptive algorithm. In this report, we try to concentrate to work on the method of adaptivity to make antenna adaptable to a virtual form of real-time environment. For, this reason a two-element antenna is used for simulation testing, as it is the most commonly used antenna for all purposes in communication. It is also tested on various scanning levels of rotation to determine the learning rate (a parameter that has no effect on the radiation output after using LMS) mean-square error rates and convergence analysis. For the purpose of above mentioned tests, three hypotheses are framed in relation to side-lobe reduction level above 5 decibels, the narrowing of the beam after adaptivity and finally the response of the main beam output for varying values of learning rate, respectivelty. The given research work, may comprehend good practical use of this LMS algorithm and also to indicate antenna patterns and the responses to adaptivity conditions through clarity in graphical format. The method is influenced to reduce computational complexity and bring simplicity to the functionality of the antenna with more efficient and effective adaptivness. An effort to test theoretical concepts in practice is also been made in this thesis work. The results show that the antenna system can be made to evolve itself through the process of adaptation with simple behaviour by relying on artificial intelligence technique which ensures little supervision and human intereference. Eventually, it is understood that the reader should have possessed prior concepts, related to antennas, digital signal processing and its practical usage in artificially intelligent systems and as well the exceptions in it, since the work is explained in the direct level assuming so.