| With the development of wireless positioning technology,the demand for Location Based Service(LBS)is becoming more and more extensive.In the traditional location fingerprint database construction approach,the Received Signal Strength(RSS)at all reference points needs to be collected in turn,which consumes a lot of manpower and time cost and is not suitable for promotion to large indoor positioning scenarios.In addition,in the indoor positioning environment,the influence of multipath effect,shielding of wall and the flow of crowd has caused irregular interference to wireless signal.The RSS sequences from Access Point(AP)received at the same reference point is of complex time-varying statistical characteristics,which makes indoor positioning accuracy decrease.In view of the above problems,based on the Bluetooth Low Energy(BLE)system platform,a multi-directional location fingerprint database construction approach is designed,through the way of combining sensor-assisted dynamic acquisition of BLE fingerprint and Radial Basis Function(RBF)interpolation.And an improved Bluetooth indoor localization algorithm is studied,which reduces the time cost of constructing fingerprint database and ensures high positioning accuracy.Firstly,a fast construction approach for multi-directional sparse location fingerprint database is designed.Several linear paths are calibrated in the indoor scene,and fingerprint data on each path is collected in a dynamic walking manner.Among them,by correcting the Pedestrian Dead Reckoning(PDR)positioning results,the coordinates of the pedestrians at each moment can be obtained accurately,and the coordinates are matched with the standard fingerprint database coordinates as Nearest Neighbor(NN).Thus,the average value of the RSS,the real-time heading angle information and the corresponding reference point coordinates of the pedestrian per second are stored in the location fingerprint database,and a multi-directional sparse location fingerprint database including multiple direction RSS sequence can be constructed.Secondly,a fingerprint database enrichment approach based on RBF interpolation is designed.The RBF is used to interpolate the multi-directional sparse location fingerprint database to construct a multi-directional location fingerprint database with high reference point density.At the same time,the optimal estimation of the RSS of the non-sampling reference point is realized by analyzing and optimizing the free shape parameter and the radius of the influence domain,and the interpolation precision and the positioning accuracy of the location fingerprint database are improved.Furthermore,an improved Bluetooth indoor localization algorithm is designed.Each unidirectional location fingerprint database is preprocessed by affine propagation clustering algorithm and outlier detection algorithm to achieve reasonable clustering of fingerprint database and reduce the fingerprint matching time for online positioning.At the same time,the localization mode combining coarse positioning and fine positioning is applied.The real-time heading angle information and RSS of pedestrian are used to determine a certain group in the unidirectional location fingerprint database,thus completing the coarse positioning.And then the Tanimoto distance is used to measure the RSS sequence between the reference points.Similarity,thus achieving fine positioning.Finally,an experimental platform is set up to test each module algorithm.The experimental results show that the dynamic multi-directional RBF Interpolated location fingerprint database,which is constructed by the proposed fast database constructing approach,the probability of errors within 3 m is 83.3% based on WKNN algorithm.Moreover,the time cost of the database construction approach designed in this thesis is smaller than that of other database construction approaches,and has a high efficiency in database construction.In addition,for the constructed multi-directional location fingerprint database,the improved localization algorithm and the traditional WKNN localization algorithm are used to locate the same group of test data.The results show that the improved localization algorithm can improve the confidence rate of small distance positioning error and reduce the large distance error. |
PDF Full Text Request