3D Fingerprint-Based Localization for Low Altitude Stereo Coverage Scenarios
Traditional search-based fingerprint matching algorithms require huge overheads in wide-area low-al-titude stereo coverage scenarios.Therefore,a three-dimensional(3D)fingerprint localization method based on a conv-olutional neural network is proposed.First,the angle-space domain channel power matrix containing rich scattering information is proposed as a fingerprint for cell-free massive multiple-input multiple-output(MIMO)systems.Then,an improved Z-Score approach is employed to normalize the fingerprint which unifies the data magnitude without changing the feature distribution.Finally,a convolutional neural network is designed to learn the fingerprint-position mapping relationship for position estimation.The simulations verify that the proposed method can achieve high-preci-sion localization with low overhead.
万方数据
维普
