Machine learning-optimized compact wearable frequency reconfigurable antenna for sub-6 GHz/mm-wave 5G integration

Area of Science:

• Electromagnetics and Wave Propagation
• Wireless Communication Systems
• Machine Learning Applications in Engineering

Background:

• Future 5G wireless systems face integration challenges between sub-6 GHz and millimeter-wave (mmWave) bands due to vast frequency differences.
• Existing antenna solutions often struggle to provide efficient and reconfigurable operation across these disparate frequency ranges for wearable applications.

Purpose of the Study:

• To propose and validate a machine learning-optimized, compact, wearable, frequency-reconfigurable antenna for seamless sub-6 GHz/mmWave 5G integration.
• To demonstrate dual-band (3.5 GHz and 28 GHz) operation with high performance metrics and compliance with safety regulations.

Main Methods:

• Fabrication of a reconfigurable antenna on a flexible substrate, initially resonating at 28 GHz, with an H-shaped slot for dual-band capability.
• Integration of a PIN diode for frequency reconfiguration between ON (3.5 GHz and 28 GHz) and OFF (28 GHz) states.
• Utilized a supervised machine learning regression framework, specifically a decision tree algorithm, to predict antenna performance (S11).

Main Results:

• Achieved dual-band operation at 3.5 GHz and 28 GHz with significant bandwidths (25.4%, 73.2%), gains (3.63 dBi, 5.25 dBi), and high radiation efficiencies (90.5%, 88%) in the ON state.
• In the OFF state, the antenna maintained a 72.9% bandwidth and 6.2 dBi gain at 28 GHz with 89% efficiency.
• The decision tree model demonstrated state-of-the-art accuracy (R²: 97.80%) in predicting S11, with minimal error metrics.

Conclusions:

• The proposed frequency-reconfigurable antenna effectively integrates sub-6 GHz and mmWave bands for 5G applications, offering stable on-body performance.
• The machine learning optimization significantly enhances prediction accuracy for antenna parameters, streamlining the design process.
• The antenna meets FCC and ICNIRP safety standards for specific absorption rate (SAR), making it suitable for future 5G wearable devices.