Research On Fingerprint Indoor Localization Algorithm Based On The Fusion Of RSS And DOA

As the digital era gathers momentum,people rely more and more on localization based services.Among various indoor positioning technologies,the fingerprint localization method has attracted much attention due to its powerful immunity to interference and high positioning precision.Most of existing researches on fingerprint localization algorithm use Received Signal Strength(RSS)as fingerprint.However,RSS is easily influenced by indoor environments,which leads to degradation of localization accuracy.Therefore,to minimize the positioning error from RSS fluctuation,it is an effective method to adopt the fusion of multiple fingerprints for accurate localization.In addition,another challenge of fingerprint localization is the high cost of offline training.The cost includes the cost of establishing the fingerprint database and the cost of maintaining the database;meanwhile,the indoor environment in which the fingerprint positioning system is located may change during long-term use.In order to ensure the validity of the fingerprint database,it is imperative to renew the fingerprint database in a time manner and minimize the burden of renewing the fingerprint database.Therefore,to improve the accuracy of fingerprint positioning and minimize the expense of fingerprint localization,the following research work is carried out based on the existing researches:(1)A fingerprint indoor localization algorithm with the fusion of RSS and Direction Of Arrival(DOA)is proposed based on circular reference points layout.The proposed algorithm abandons the traditional fingerprint reference point distribution in a lattice pattern and uses a circular distribution to set the reference points to reduce the count of reference points.In the offline phase,RSS and DOA information is gathered at the reference points to make the fingerprint database respectively.During the online positioning,the matching loop is determined by the RSS average of each loop to narrow the matching range,and then using DOA for position estimation.The proposed algorithm avoids the fusion of training weights for multiple fingerprints,and it reduces the online matching range.Simulation results indicate that for the same reference points interval,the number of reference points required by circular layout is less than that of lattice layout.In addition,the proposed algorithm reduces the average localization error by 80% compared with the traditional RSS fingerprint localization algorithm with lattice reference points layout and by 33.5% compared with the RSS and DOA fusion fingerprint localization algorithm with lattice reference point layout.(2)When the indoor layout is suddenly changed or other irresistible factors occur,the Anchor Node(AN)is moved and the fingerprint database cannot accurately reflect the changed indoor environment,thus the fingerprint database needs to be rebuilt.In addition,when the AN is not in the center of the target area,the RSS fluctuates greatly in the corners away from the AN,resulting in an increase in the positioning error.To ensure the positioning performance and lower the offline cost of reconstructing the fusion fingerprint database,an update mechanism for the fused fingerprint positioning algorithm is proposed using the proposed fusion fingerprint localization algorithm.In the offline phase,the position coordinates of existing reference points are updated using geometric relationship,and there is no need to re-capture the fingerprint information of these reference points;only the fingerprint information of the newly introduced reference points need to be collected to reconstruct the whole fingerprint database.The online phase relies on more stable DOA fingerprint,narrowing the online matching range to a certain angular range and allowing RSS fingerprints to assist in determining the location of the target node at the corner.Simulation results show that in the corner area far from the AN,the average localization accuracy using the improved fingerprint localization algorithm is higher than that of the original fusion fingerprint localization algorithm.