With recent developments in the Global Satellite Navigation Systems (GNSS), the applications and services of positioning and navigation have developed rapidly worldwide. Location-based services (LBS) have become a big application which provide position related services to the mass market. As LBS applications become more popular, positioning services and capacity are demanded to cover all types of environment with improved accuracy and reliability. While GNSS can provide promising positioning and navigation solutions in open outdoor environments, it does not work well when inside buildings, in tunnels or under canopy. Positioning in such difficult environments have been known as the indoor positioning problem. Although the problem has been looked into for more than a decade, there currently no solution that can compare to the performance of GNSS in outdoor environments. This thesis introduces a collaborative indoor positioning solution based on particle filtering which integrates multiple sensors, e.g. inertial sensors, Wi-Fi signals, map information etc., and multiple local users which provide peer-to-peer (P2P) relative ranging measurements. This solution addresses three current problems of indoor positioning. First of all is the positioning accuracy, which is limited by the availability of sensors and the quality of their signals in the environment. The collaborative positioning solution integrates a number of sensors and users to provide better measurements and restrict measurement error from growing. Secondly, the reliability of the positioning solutions, which is also affected by the signal quality. The unpredictable behaviour of positioning signals and data could lead to many uncertainties in the final positioning result. A successful positioning system should be able to deal with changes in the signal and provide reliable positioning results using different data processing strategies. Thirdly, the continuity and robustness of positioning solutions. While the indoor environment can be very different from one another, hence applicable signals are also different, the positioning solution should take into account the uniqueness of different situations and provide continuous positioning result regardless of the changing datWith recent developments in the Global Satellite Navigation Systems (GNSS), the applications and services of positioning and navigation have developed rapidly worldwide. Location based services (LBS) have become a big application which provide position related services to the mass market. As LBS applications become more popular, positioning services and capacity are demanded to cover all types of environment with improved accuracy and reliability. While GNSS can provide promising positioning and navigation solutions in open outdoor environments, it does not work well when inside buildings, in tunnels or under canopy. Positioning in such difficult environments have been known as the indoor positioning problem. Although the problem has been looked into for more than a decade, there currently no solution that can compare to the performance of GNSS in outdoor environments. This thesis introduces a collaborative indoor positioning solution based on particle filtering which integrates multiple sensors, e.g. inertial sensors, Wi-Fi signals, map information etc., and multiple local users which provide peer-to-peer (P2P) relative ranging measurements. This solution addresses three current problems of indoor positioning. First of all is the positioning accuracy, which is limited by the availability of sensors and the quality of their signals in the environment. The collaborative positioning solution integrates a number of sensors and users to provide better measurements and restrict measurement error from growing. Secondly, the reliability of the positioning solutions, which is also affected by the signal quality. The unpredictable behaviour of positioning signals and data could lead to many uncertainties in the final positioning result. A successful positioning system should be able to deal with changes in the signal and provide reliable positioning results using different data processing strategies. Thirdly, the continuity and robustness of positioning solutions. While the indoor environment can be very different from one another, hence applicable signals are also different, the positioning solution should take into account the uniqueness of different situations and provide continuous positioning result regardless of the changing data. The collaborative positioning aspect is examined from three aspects, the network geometry, the network size and the P2P ranging measurement accuracy. Both theoretical and experimental results indicate that a collaborative network with a low dilution of precision (DOP) value could achieve better positioning accuracy. While increasing sensors and users will reduce DOP, it will also increase computation load which is already a disadvantage of particle filters. The most effective collaborative positioning network size is thus identified and applied. While the positioning system measurement error is constrained by the accuracy of the P2P ranging constraint, the work in this thesis shows that even low accuracy measurements can provide effective constraint as long as the system is able to identify the different qualities of the measurements. The proposed collaborative positioning algorithm constrains both inertial measurements and Wi-Fi fingerprinting to enhance the stability and accuracy of positioning result, achieving metre-level accuracy. The application of collaborative constraints also eliminate the requirement for indoor map matching which had been a very useful tool in particle filters for indoor positioning purposes. The wall constraint can be replaced flexibly and easily with relative constraint. Simulations and indoor trials are carried out to evaluate the algorithms. Results indicate that metre-level positioning accuracy could be achieved and collaborative positioning also gives the system more flexibility to adapt to different situations when Wi-Fi or collaborative ranging is unavailable. The collaborative positioning aspect is examined from three aspects, the network geometry, the network size and the P2P ranging measurement accuracy. Both theoretical and experimental results indicate that a collaborative network with a low dilution of precision (DOP) value could achieve better positioning accuracy. While increasing sensors and users will reduce DOP, it will also increase computation load which is already a disadvantage of particle filters. The most effective collaborative positioning network size is thus identified and applied. While the positioning system measurement error is constrained by the accuracy of the P2P ranging constraint, the work in this thesis shows that even low accuracy measurements can provide effective constraint as long as the system is able to identify the different qualities of the measurements. The proposed collaborative positioning algorithm constrains both inertial measurements and Wi-Fi fingerprinting to enhance the stability and accuracy of positioning result, achieving metre-level accuracy. The application of collaborative constraints also eliminate the requirement for indoor map matching which had been a very useful tool in particle filters for indoor positioning purposes. The wall constraint can be replaced flexibly and easily with relative constraint. Simulations and indoor trials are carried out to evaluate the algorithms. Results indicate that metre-level positioning accuracy could be achieved and collaborative positioning also gives the system more flexibility to adapt to different situations when Wi-Fi or collaborative ranging is unavailable.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:684878 |
| Date | January 2015 |
| Creators | Jing, Hao |
| Publisher | University of Nottingham |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://eprints.nottingham.ac.uk/30475/ |
Page generated in 0.0025 seconds