Generalization Capabilities of Neural Network- Based Smart Antenna for Doa Estimation

We are concerned with neural network-basedsmart antenna application in direction of arrival estimation.We have separated and presented the types of generalizationsthat the network should be able to perform in this type of aproblem. The basic concept of the network is presented,followed by steering information for detailed analyze. Usingcomputer simulations, the suitable examples are presentedand comments are given. After conclusions, the informationfor the necessary future investigation is noted. GENERALIZATION CAPABILITIES OF NEURALNETWORK-BASED SMART ANTENNA FOR DOAESTIMATION, M. Sarevska, B. Milovanovi , Z. StankoviFigure 5. Simulation results for antenna array with M=10 el.,SNR=20 dB (3 users at distance of 6°, 5.5°, 5°, 4.5°, and 4°) Figure 6. Simulation results for antenna array with M=10elements, SNR=30 dB (2 and 3 users at distance of 3°)