| Indoor positioning technology is playing a pivotal role in the rapidly developing Io T applications,and its social demand and commercial value are increasing.Fingerprint indoor positioning technology based on Wi Fi received signal strength has been of great interest and exploration by researchers due to its low hardware cost,good positioning performance and high universality.However,the technology is limited by the complexity of the indoor target space and the sensitivity of the signal itself.Therefore,according to the idea of "divide and conquer",the work is compatible with the advantages of fuzzy division of space,and reduces the amount of matching calculation as much as possible with a controllable error range.Based on the spatial distribution characteristics of the source fingerprint,the best access point set in the subspace is matched to highlight the performance of the simplified access point in each subspace.The iterative update of the reference point is adopted to match the optimal neighbor point set for the estimated point to improve the positioning performance of the system.The fuzzy division of the target space not only ensures the timeliness of positioning,also plays the multi-matching value of the spatial access point and the location of the nearest neighbor with the fuzzy subspace with more learning capability.The specific research content and innovation points are as follows:(1)To address the problem that the localization efficiency of large-area targets is limited and the traditional clustering method cannot guarantee the effective division of regions,a spatially adaptive soft division method based on core cohesion is proposed.Firstly,the signal anomalies caused by uncertain factors such as noise and transmission are preprocessed to avoid the interference of signal acquisition errors on subsequent positioning.Secondly,based on the idea of division and judgment,the core cohesiveness is used to determine the clustering center,and the target space is automatically divided into several fuzzy subspaces with similar characteristics under the effect of the complementary benefits of fingerprint two-way distance information.After executing coarse subspace matching,the spatial search range of the target to be located is narrowed by alternately updating the neighbor fingerprint centroid,completing the detailed matching of the estimated points,and feedback the best positioning results of users.Through experimental validation in both regular and non-regular spaces,it is demonstrated that the proposed algorithm reduces the localization cost and improves the localization accuracy with more streamlined fingerprint samples while ensuring the subspace discrimination capability,maintains a strong universal performance and flexible adaptation to different indoor environments.(2)Aiming at the problem of unreliable access point information introduced by spatial structure barrier,signal weak penetration and other factors,a spatial dual-refinement strategy is proposed to achieve the filtering of redundant sources at different locations.The algorithm analyzes the frequency of the strongest received signal from a single source and the physical characteristics of the actual spatial clustering for horizontal refinement of reference points.On the basis of this reduced-dimensional subspace,the best set of refinement access points with sufficient contribution is vertically filtered for each subspace by the dual-scale measurement of spatial differentiation and coverage reliability,which solves the redundancy problem of full-scene matching and at the same time builds a high-value fingerprint library for online positioning.The proposed horizontal streamlining strategy has been tested in practical road shows that the clustering of the proposed horizontal streamlining strategy is also more in line with the constraints of the scene structure,and the vertical streamlining strategy improves the average positioning accuracy by at least 17% compared with the traditional access point matching algorithm.(3)In order to solve the problem of cross-interference of blurred boundaries between clusters,environment and signal dynamic characteristics on fingerprint positioning results,an indoor positioning method with self-renewing matching of nearest neighbour sets in fuzzy space is proposed.While compatible with the advantages of overlapping spatial divisions,the negative effect of absolute discriminations between zones is alleviated and the generalization ability of localization matching is improved.The distance metric in the signal domain between the reference point and the point to be positioned is transformed into a dimensionless ranking to indirectly map the similarity between the point to be positioned and the reference point by integrating the signal domain and the spatial domain to iteratively constrain the reference point to achieve the dynamic selection and clustering effect of the target nearest neighbour set,thereby overcoming the interference of environmental changes and signal fluctuations.The proposed algorithm outperforms similar zoning algorithms by 4.7%-11.8%in terms of localisation accuracy and reduces the average localisation error by 0.422 m when compared with global matching methods. |
