Implementation of machine learning for the design of spiral shaped multiband monopole antenna for MICS/IEEE802.11a/IEEE802.11b applications.

This paper presents a method to reduce the size of a compact antenna for MICS and IEEE 802.11a/b applications. Initially a monopole antenna (45 × 40 × 1.6 mm³) is designed for the ISM band (2.4–2.5 GHz). A spiral meandered single patch is incorporated to lower the operational frequency to 600 MHz assuring significant antenna size reduction. Further enhancements include a two-sided spiral extension and a T-shaped arm around the microstrip feed, enabling operation across three frequency bands achieving an overall 84% size reduction with improved gain. The prototype meets FCC standards for Specific Absorption Rate (SAR). To optimize gain, machine learning models along-with LASSO, Ridge, and Random Forest regression algorithm are used. The LASSO model proves most effective, achieving gains of 6.6629 dBi at 400 MHz, 7.6225 dBi at 2.45 GHz, and 8.7569 dBi at 5.5 GHz, with fractional bandwidths of 22%, 27%, and 8% respectively. [ABSTRACT FROM AUTHOR]