Evaluation of position estimation of a human body around a vehicle by FDTD propagation analysis

現在車両の所有者に対するサービス向上を目的として車両周辺の人体位置を電波により検出する技術が注目されている．本稿では，車両側の単一アンテナから電波を送信し，人体からの反射波を含む受信信号を車両側にも設置した複数アンテナで受信することを想定している．受信信号に基づいて，人間の位置を推定する．推定方法としては，近距離Beamformer法と近距離MUSIC法を用いた．推定の分析の結果，受信アレーの素子数が3のように少ない場合でも，MUSIC法を適用すると角度方向に対する分解能が高いことを示した．しかしながら，推定方法の実用的な使用を考慮すると，距離方向に対する分解能は低い．そこで，推定精度を向上させるために、アレーアンテナの配置を変えることを考える．同様の素子数で開口長を拡大したアレーアンテナ配置では推定誤差が1m以下になる確率を50％から90％に改善できることを示す．As a method of estimating position of a human body around a vehicle, a sensing technique using radio wave is considered. In this paper, we assume radio wave is transmitted from a transmitting antenna on the side of a vehicle and the received signals including the reflected waves from a human body are received by multiple receiving antennas mounted also on the side of the vehicle. Based on the array of the received signals, the position of the human is estimated. We calculated the propagation characteristics of the received signals by FDTD. We used the near-field Beamformer and the near-field MUSIC methods as the estimation methods. We showed as a result of the analysis on the estimation that even when the number of the elements of the receiving array is small such as 3, the resolution in the angular domain is high when MUSIC is applied. However, the range resolution is unacceptably low considering the practical usage of the estimation method. We therefore consider a different arrangement of the array antennas to improve the estimation accuracy. We show that the probability that the estimation error becomes 1 m or less can be improved from 50% to 90%