Research Of Fingerprint-based Indoor Localization In Ultra Dense Networks

In recent years,with the rapid development of the Internet of Things,the demand for location-aware applications has proliferated,such as positioning and guidance in hospitals,airports and large shopping malls.With the wide deployment of Wi Fi infrastructure,Wi Fi-fingerprint-based localization technology has become one of the most promising localization technologies.On the one hand,Wi Fi-fingerprint-based localization technology can work in line-of-sight(LOS)or nonline-of-sight(NLOS)environments,so it is more suitable for practical complex indoor environments.On the other hand,the collection of received signal strength(RSS)requires no additional infrastructure and can be implemented with only a few basic devices(eg,smartphones,tablets).However,actual indoor environments of ultra dense network exist the following problems and challenges.First,dense access points(APs)cause the construction of fingerprint database to be very cumbersome.Furthermore,the fingerprint database has a very large amount of RSS datas due to dense APs,which increases the matching time of online location estimation.Second,APs deployed intensively cause environmental dynamics problems(such as dynamic changes of APs,RSS time-varying problems),which reduces the effectiveness of the fingerprint database,making it difficult to estimate location accurately.Third,due to the different Wi Fi chips installed in different mobile device,absolute RSS values collected exist some differences(even in the same location).The use of different devices to collect the RSS values of offline and online phases can cause device heterogeneity problem and the discrepant absolute RSS values is difficult to calibrate.In view of the problems in the above indoor environments of ultra dense network,this paper focuses on the indoor localization approachs,which is applicable to ultra dense networks.The main work is as follows:1)This paper proposes a robust indoor localization approach based on fingerprint similarity,which includes the following key points.First,In view of the problem of too much fingerprint database datas,a simple and effective AP selection method is designed to select the APs that can effectively characterize RSSs.The method can remove a large amount of redundant datas,which can reduce the size of the fingerprint database by 76%,in the case of ensuring that the localization accuracy meets the requirements.Second,because absolute values of RSS collected between different mobile devices has some differences,a fingerprint database construction method based on RSS relative values is proposed,which can solve the heterogeneity problem of devices.Third,in order to mitigate the impact of environmental dynamics,the localization matching method based on score mechanism is designed in the online phase,which could deal with the positioning problems caused by APs' dynamic changes and time-varying RSSs.The experimental results show that the designed approach can improve the localization accuracy,especially in the case of many APs missing,which can provide more robust localization accuracy.2)Although the above AP selection method can reduce the size of the fingerprint database,it still requires to spend a lot of time to collect RSS samples.In order to avoid site collection of RSSs,a collection-free indoor localization approach is designed based on the previous approach.The localization approach constructs fingerprint database by using the distance between APs and reference points.Therefore,the approach does not require any on-site measurement and only needs to know the exact location of the AP to quickly construct fingerprint database,which greatly reduces the overhead of the offline phase.The simulation results show that in the two typical indoor scenarios,conference rooms and corridors,compared with the Selective-AP method,the median localization error of the method is reduced by 55.53% and 31.20%,and the corresponding average error localization is decreased by 49.17% and 43.18%.Based on the simulation results,the collection-free indoor localization approach can effectively improve the localization accuracy.