Research On Improvement Of Indoor Location Algorithm Based On IBeacon

With the increasingly development of the technologies of Internet of Things,various smart communities and smart campuses,location based services has gradually become an indispensable part of people's daily life.In outdoor positioning,there are mature and universal systems,such as GPS system in the United States,Beidou system in China,etc.In the field of indoor positioning,technology based on WIFI,Bluetooth and other equipment has also been developed,but there is still no mature system.As a Bluetooth technology based on Bluetooth 4.0 protocol,iBeacon device has low power consumption and long transmission distance,which meets the needs of indoor positioning.In this paper,the indoor location algorithm based on iBeacon is studied,and the algorithms in different directions are improved.In this paper,the iBeacon-based indoor location algorithms are classified and summarized,and proposed the improvements algorithms of the range-based indoor location algorithm and fingerprint matching algorithm.In the indoor location algorithm based on ranging,the core is to convert the received signal strength(RSSI)of Bluetooth device to distance,which is relatively simple to implement,but requires high accuracy of distance.In view of this feature,an extended Gauss filtering method is proposed to further optimize the received RSSI values and weaken the multipath fading effect of indoor environment.Experiments show that this method has higher accuracy than traditional mean filter and Gauss filter.Compared with the indoor location algorithm based on ranging,although fingerprint matching algorithm needs to establish additional fingerprint database,it does not need to convert RSSI values into distance,and the systems have strong anti-jamming ability.Based on the analysis of fingerprint database,this paper points out that the database used for indoor location generally has the characteristics of high dimension,low proportion of effective information and non-linearity.However,the most advanced fingerprint matching algorithms are mainly based on the shallow machine learning hierarchical structure algorithm,which is not strong enough to represent high-dimensional non-linear fingerprint information.Therefore,in this paper,a deep neural network algorithm based on stack auto-encoder is introduced to optimize the high-dimensional fingerprint database.The deep learning method is used to extractthe higher-level and lower-dimensional feature attributes,so as to reduce the complexity of the final mobile terminal location algorithm,and to extract the high-dimensional information,at the same time,it can also reduce Influences of the noises to fingerprint database to a certain extent.In the existing fingerprint database,the machine learning algorithm and the stack-based automatic encoder based on different levels of depth neural network algorithm are compared.The experimental results show that the related algorithm has higher accuracy in real data,and the choice of network depth has an impact on the algorithm.Finally,the indoor positioning software based on iBeacon is designed and implemented on the mobile terminal.In the final part of indoor positioning,the K-nearest neighbor(KNN)algorithm is used to predict the user's position.The influence of different values of parameter K on positioning effect is tested in the experiment.The average error of positioning reaches 1.38 meters,which can meet the needs of most indoor positioning scenarios.The practicability of the system is realized.