Location Problem Based On Simultaneous Localization And Mapping And Visible Light Signal

With the arrival of the fifth generation(5G)networks and the rapid update of the intelligent terminal,more and more devices are involved in wireless sensor networks to satisfy the massive connectivity and diverse real-time services.Light emitting diodes(LEDs)as new lighting equipment have replaced incandescent and fluorescent lamps.VLC is an attractive approach for short-range communications,which complements radio and can be used in the environment of the sixth generation(6G)networks.Meanwhile,under the indoor environment,the cooperation between VLC and visible light positioning(VLP)using white LEDs has caused extensive concerns.Recent years,visible light has achieved high positioning accuracy in indoor positioning due to its fine-grained characteristics,but it often needs high calibration cost in training ranging model.In this work,we propose a system named Crowd Light to leverage crowdsourcing traces from multiple users to construct radio maps for VLP.We design visible light fingerprints based on external frequency control and FFT algorithm.The crowdsourcing traces are merged based on an enhanced ICP algorithm comprised of two main steps,coarse-grained trace merging and fine-grained trace merging.The positioning accuracy is calculated by KNN method.The contributions of this work are summarized as follows.(1)Constructing coarse-grained radio map.For crowdsourcing trace merging with the measured visible light RSSI,after preprocessing and trace clustering,we propose a com Similar algorithm to detect the matching segments via the visible light RSSI and filter false positive ones by adding window function,detecting the consistency of trace shape and detecting the consistency of RSSI vector.Then according to the merging orders decided by multi-tree structure,the ICP algorithm is applied to the traces to construct coarse-grained radio map.(2)Constructing fine-grained radio map.Since the coarse-grained trace merging treats a trace as a rigid body,to compensate the accumulative errors inside the traces caused by pedestrian dead reckoning(PDR)algorithm,a fine-grained radio map is constructed via the physical coordinates.After splitting the obtained traces with turn points and merging the close traces,we use the Canny edge detection method to find the connected domain and remove the outlier to construct a fine-grained radio map.(3)Using the proposed Crowd Light approach,we build a crowdsourcing visible light radio map and design a fingerprinting system.We evaluate this system in a comprehensive indoor environment through the real measured data and our specially designed equipment.The evaluation results show that our system achieves a mean accuracy of 0.65 m for the merged traces and 0.83 m for KNN indoor positioning.