Research On Indoor Fingerprint Localization Based On Access Point Analysis

The signal reflection,refraction and attenuation effects caused by the complex environment layout,obstacles and crowd activities in the indoor environment make the Access Point(AP)distributed in the scene show great differences in the positioning process.The paper takes the contribution of AP participation in positioning as the value orientation,and analyzes the internal relationship between its signal distribution characteristics and spatial location in detail.Based on this,the indoor fingerprint positioning algorithm is studied to achieve efficient database construction at low cost and low error positioning at high efficiency.The specific research contents are as follows :(1)In order to solve the problem of increased computational complexity and decreased positioning accuracy caused by the redundancy of fingerprint data information caused by a large number of indoor detectable AP sources,a multi-scale screening AP strategy based on compressed sensing algorithm is proposed.In the offline phase,the algorithm uses the reference point(RP)spatial location feature to divide the positioning scene into a regional subset,and designs a fuzzy discriminant criterion to realize the clustering boundary intersection.In this paper,the sparse basis establishment and measurement matrix design in the field of compressed sensing are related to the fingerprint database construction and AP screening of indoor positioning,and the multi-scale AP screening strategy is designed to condense the effective positioning information in fingerprint data.In the online stage,the sparse Bayesian algorithm is selected to weaken the fingerprint correlation to accurately recover the target position vector,and the position weight coefficient is reconstructed.Verified by the actual scene,the algorithm has a good positioning accuracy improvement compared with the same field and traditional positioning algorithms.(2)Considering that the collection and processing of offline fingerprint data consumes a lot of manpower and time costs in large-scale positioning scenarios,this paper proposes an offline fingerprint database construction algorithm driven by the dynamic position of the source.The paper processes the outliers in the measurement data through the absolute median deviation algorithm.In the offline phase,the paper fits the AP position distribution based on geometric ideas and least squares,and removes the outliers to determine the AP dynamic position.By analyzing the AP signal coverage characteristics,the paper introduces the signal distance weight to interpolate and predict the adjacent sampling points to complete the fingerprint database expansion.In the online stage,the fluctuation range of AP signal is quantified to establish a ring-shaped adjacent domain,and the adjacent RP set is searched in echelon.Through comparative experiments,the algorithm can achieve the purpose of fingerprint database interpolation filling while ensuring the interpolation quality,and the average positioning error is reduced by more than 17.4 % compared with similar algorithms.(3)Complex positioning algorithms usually sacrifice certain efficiency to pursue lower positioning errors,making it difficult to balance matching positioning efficiency and positioning error accuracy.Aiming at this problem,this paper proposes an offline two-dimensional partition localization algorithm that introduces source location features.The algorithm divides APs with similar spatial location and signal distribution into regional AP groups.Taking this as the anchor point,the entropy weight of the reference point to the AP ’s attribution information is calculated to determine the RP ’s region,and a small-scale fingerprint database set with rich positioning information is established.In the online stage,the algorithm proposes a similarity discrimination method based on the propagation distance to match the nearest neighbor points,and the misjudgment points are eliminated by the sum of the distances.Compared with other algorithms,the proposed algorithm has excellent performance in positioning time consumption and accuracy.