Antenna Optimization Based on Neural Network for M2I Underground Wireless Sensor Network
摘要
Underground wireless sensors seem to be the most promising technique applied in mine safety monitoring, agricultural automation, etc. However, traditional EM wave-based transmission technique suffers from large propagation loss in complicated underground scenarios. Recent studies show that metamaterial magnetic induction (MI) technique can provide a reliable transmission in hostile underground medium, but how to implement an artificial electromagnetic metamaterial is still a time-consuming project. In this article, we design an optimized metamaterial slab (M-slab) to improve the energy-flux density of M2I transceiver antenna in destination direction. The M2I transceiver antenna (combination of coil and M-slab) was modeled by COMSOL Multiphysics in earth, and induction voltage at receiver coil is used to measure the enhancement by installing M-slab; the optimal permeability value of the metamaterial is proven to be − 20 through simulations, and the optimal parameters of the metamaterial slab structure are fitted using back propagation neural network algorithm in MATLAB; the metamaterial slab structure is designed and verified by HFSS simulations; the induction voltage at receiver coil is measured by experiments while M-slab installed before transmitter coil or/and receiver coil with optimal coil-slab gap (3 mm). The experimental results show that the designed M-slab can evidently improve the receiving signal strength, achieving at least 15.3 dB enhancement in voltage while M-slab placed in both Tx and Rx coil compared to that of no M-slab within its 3dB bandwidth while the coil distance is 0.5 m.