Integrated Inertial Positioning With RSSI Fingerprinting Indoor Localization System

The rapid development of wireless communication technology and smart mobile devices has resulted in a great awareness on the research of localization.Indoor localization has significant importance for an increasing variety of emerging pervasive commercial,consumer and enterprise applications.While the satellite-based positioning technique(used by GPS,GLONASS,GALILEO and BEIDOU)becomes the standard for out-door civil positioning,indoor localization is still a battle field that various technology options such as infrared(IR),ultrasound,radio-frequency identification(RFID),wireless local area network(WLAN),Bluetooth,ultra-wideband(UWB),ZigBee are available for building solutions.Each technology needs appropriate methods and algorithms to locate the target efficiently.Generally,localization algorithms can be classified into 2 categories: signal-propagation-based and fingerprint-based systems.Fingerprinting is the widely adopted approach in Received Signal Strength(RSS)based systems.In this work,we build an indoor localization system which integrates inertial positioning with RSSI fingerprinting method.The system takes advantages of both two methods.Although the inertial based localization suffers from cumulative errors,it accurately and quickly estimates the shift distance from the previous position.In order to correct cumulative error,we combine the RSSI fingerprinting method.In the meanwhile,we proposed a dynamic active area clustering(DAAC)to reduce the computational overhead of fingerprinting.We evaluate the performance of indoor positioning system and clustering algorithm by deploying them in the real indoor environment.Crowdsensing is a new and promising paradigm which harnesses smartphones to collect and share data.There have been proposed some work which use crowdsensing to collect RSSI and build fingerprint database for indoor localization.In this work,we study the sensing task allocation problem in such system.We formally formulate this problem and prove this problem is NP-hard.We propose two hybrid algorithms which combine a heuristic algorithm and two meta-heuristic algorithms respectively.The extensive simulation results show that the proposed hybrid algorithms outperform the meta-heuristic algorithms.