Smart electromagnetic structures: The neural antenna

Smart electromagnetic structures (SEMS) are defined as structures capable of interacting with their surrounding electromagnetic fields and either influencing the field or sensing and adapting to its presence. A structure is smart when it integrates sensing elements (e.g., antennas), processing elements (neural networks) and control elements (diodes) autonomously. SEMS provide an adaptive electromagnetic (EM) environment for the structure on which they are mounted. The ability to adapt derives from the closed loop nature of the SEMS. The speed of adaptation is determined by the speed of the loop, which is set by the computational elements. Our experiments shown the time required for a response is about fifteen gate delays. The integration of artificial neural processors with tuneable antennas was proposed several years ago by our group. This synergy over the past three years was studied. The control of the operating frequency of a microstrip patch antenna was demonstrated. We believe that ours is a unique program offering great potential for payoff in the area of electromagnetic smart skins. Our main goal in this program was to '...determine the feasibility of a neural network controlled antenna and to quantify the ability of the antenna and the NN to learn to tune automatically to the center frequency of a received signal.' These goals were achieved and spun off neural networks that have application of radar and communications systems. Because of its self adaptability the closed loop neural control of an antenna element, provides the potential for design of an easily manufacturable antenna which is immune to typical siting problems, and is tolerant to moderate external damage.