An Optimized Indoor Localization Algorithm Based On Wi-Fi Hybrid Fingerprint Map

With the popularity of wireless communication technology,the demands for location-based services from outdoor to indoor increase greatly.Because GPS provides absolute accurate location services to outdoor localization,but to indoor areas,the signal transmission will be disturbed by people activities,multipath effect,shadow effect,temperature,humidity and other factors.Indoor positioning faced with a big problem called "last one hundred meters' location".With the router,which also called wireless signal access point(AP),widely distributed in people's lives,smart phones,tablet PCs,laptops can receive Wi-Fi signals.Wi-Fi positioning does not need to add additional equipment,it has become a low-cost positioning technology,and becomes a research hotspot.Based on RSS(Received Signal Strength)of WLAN fingerprint indoor positioning technology divided into offline and online stages,their ways to collect signal data and to calculate the position,both can affect the precision of the location result.At present,many experts and scholars have focused on the improvement of online matching algorithm,but the construction,utilization and update of the fingerprint atlas also determine the positioning accuracy.The paper analysed the status of fingerprint atlas such as the heavy manual workload,low utilization rate,low efficiency,and the low positioning accuracy.Then it gave some steps to optimization in detail.Firstly,in the off-line stage,we collected the vector information in target area,combined with artificial neural network and data mining clustering algorithm,training the virtual reference point,and then established a high density mixed fingerprint library organization.Secondly,using the geographical coordinates and the fingerprint altas to identify the indoor environment set the positioning results of boundary conditions.Online positioning,modified WKNN(Weighted K-Nearest Neighbor)algorithm fusion positioning results correcting map information,output a high quality positioning result.Thirdly,implemented the hybrid fingerprint database positioning system.Above all,this algorithm established a high quality and high density mixed fingerprint map,improved the utilization rate of the RSS sample information,effective data mining and artificial neural network method into fingerprint positioning process.Experimental results shown that the hybrid fingerprint image can achieve and provide a more accurate and highly organized fingerprint database,and accelerate the convergence of online localization algorithm.Therefore,the algorithm has good universality and popularization value.