DESIGN OF MIMO ANTENNA FOR ISM BAND USING ANN ENSEMBLE by Kamaldeep Kaur
In this era of wireless communication, the antenna designers have been challenged to come up with compact size, high- performing, easy to fabricate antennas with increase in radiation efficiency to keep up with the day to day increasing demands. As the size of communication devices is reducing with the time, so the demand for compact size antennas are growing. Now days the latest trend is to reduce the size and increase the bandwidth of antenna. An ensemble hybrid algorithm is formed by using the artificial neural networks. If we use single ANN models that have some limitations such as the generalisation of the different networks is not unique and also the performance is not up to the mark when we deal with insufficient training data and complex data sets. In this work, new fractal antenna for MIMO application is proposed. For designing this fractal geometry we use Koch and Minkowski curves placed on the boundary of inverted U patch. The proposed antenna design is on RT-Duroid substrate having thickness 3.175 mm. Antenna is required to resonant for ISM Band i.e. 2.45GHz. The antenna is also checked and verified experimentally for the above substrate, the simulated and experimental results are in good harmony. The fabricated fractal antenna has good return loss value of -18.691dB which shows the signs of good antenna. The simulation of MIMO antenna is done by using IE3D software in which data set is calculated for different Length width and resonant frequency is calculated for both Simple and Fractal Geometry and then testing and training is done using ANN and for 2.45GHz frequency particular Length & Width through Ensemble is calculated and verified again using IE3D software. The MIMO antenna results for single and two element antenna are same i.e. the resonant frequencies comes to be same. The MIMO antenna is fabricated on the RT-Duroid substrate and same is verified experimentally with the help of Vector Network Analyzer. There is similarity between measurement and simulation results.