A phase-controlled microstrip antenna array using neural networks

An advanced microstrip patch antenna array with beam steering capability is presented. Characteristics of a microstrip antenna are investigated using basic theories and a 3D full-wave simulation tool. Neural networks are used to approximate antenna characteristics such as input return loss. The beam steering of the array can be achieved by using far-field phase shift characteristics obtained by placing a shorting pin in different locations or varying the impedance of a load embedded on a microstrip patch antenna element. Far-field phase shift capability according to the location of a shorting pin is investigated. Beam steering examples using the shorting pin for a patch antenna array are presented. A microstrip antenna with an embedded load is studied. Antenna characteristics such as input return loss, antenna gain, directivity, far-field magnitude, and far-field phase are investigated for different load resistance and reactance values. A varactor diode can be used as the embedded load whose impedance can be controlled by a proper input voltage. Neural networks are used to control the far-field phase angles by supplying a voltage to the varactor diode. A phased array is designed using embedded microstrip antenna elements. Several different methods including linear equations and neural networks are used to implement the beam controller. It is shown that the beam pointing error can be significantly reduced by using neural networks instead of a linear controller. Neural controllers also enable us to better predict a beam shape. Beam steering capability of 360° in a horizontal plane is achieved by using neural networks for a circularly oriented 4-element array.