法規制による送信電力の制限や伝送損失により、マイクロ波給電に基づくバッテリレスデバイスの動作範囲は数 m程度に限られる。そこで同一の電力密度からより多くの電力を取り出すため、受信器の整流器が有する非線形性を考慮し、整流効率を向上する送信波形生成手法が提案されてきた。しかし、送信波形の帯域が大きい場合、マルチパスにより信号が歪み、整流効率が低下する。そこで、ベイズ最適化を用いて少ない学習で伝搬経路に適した波形を効率よく探索し、整流効率を向上した。

The working range of battery-less devices powered via microwave power transfer could not go beyond several meters away from the source due to the output power limited by regulations and transmission loss. Recently, power optimized waveforms have been proposed to increase power conversion efficiency at the receiver side by generating high peak to average power ratio without increasing transmit power. However, transmit signal with large bandwidth suffers from multipath effects. In this work, we introduced bayesian optimization to efficiently learn channel response and to generate transmit signal that has better power conversion efficiency depending on the channel response.

Takashi Ikeuchi, Yoshihiro Kawahara, Joshua R. Smith

Contact: ikeuchi@akg.t.u-tokyo.ac.jp