Publicação
FastGraph - unsupervised location and mapping in wireless networks
| Resumo: | Wi-Fi Based Indoor Positioning solutions normally require complex and time-consuming deployment processes or have limited accuracy. Fingerprinting Matching is one of most used techniques for indoor positioning, which relies on a radio map that is normally created offline in a calibration phase by manual site survey. Wi-Fi Fingerprint can be used to locate regular mobile devices, such as smartphones, using only software and can be used in any indoor environment without being necessary to deploy additional infrastructure, relying only on the existing Wi-Fi infrastructure. However, in most cases the manual site survey is unpractical and involves a significant effort, even for small scale spaces. Moreover, due to the dynamic nature of radio environments, to maintain the system performance, the site survey has to be repeated often to keep the radio map updated. This process is not feasible in large spaces, and compromises the scalability of this type of approach. Solutions have been proposed to reduce the calibration effort, using collaborative site survey to create and maintain the radio maps, or by using Model-Based methods to approximate a radio map. However, the reduced calibration effort usually implies a lower positioning accuracy and higher computational requirements. In this context, FastGraph is proposed as a new solution able to provide unsupervised positioning using different devices, such as smartphones or autonomous machines, while automatically creating and maintaining a Radio Map. A 3D Force-Directed Graph is used to rapidly model the radio environment. The 3D Graph is iteratively constructed with data collaboratively collected by several devices. Orientation and motion information, obtained from different sensors, can be used to improve the Graph constrains. FastGraph is able to operate shortly after its deployment, without previous knowledge about the environment. The proposed solution uses a novel algorithm to automatically provide location while simultaneously updating the radio map; and estimate the position of the Access Points (APs) and location-specific radio propagation parameters. In addition, FastGraph does not rely on expensive hardware or requires high computational effort. The FastGraph approach may be used in different contexts. In addition to the indoor positioning, the radio maps created by FastGraph include supplementary information that can be used to automatically map the interference in Wi-Fi networks and even to automatically map the physical space. The described solution was deployed and evaluated in two very distinct real world spaces, an industrial environment and a regular office building. The experiments, in these two spaces, evaluated the several aspects of FastGraph, and considered scenarios where only Wi-Fi data is available and when the Wi-Fi can be combined with data from other sensors. The results suggest that the proposed solution has potential to provide interference information in wireless networks and provide positioning in different indoor scenarios, from regular buildings, to autonomous vehicles in industrial environments, with the possibility of being also extended to outdoor spaces using data from cellular networks, especially considering the characteristics of the upcoming 5G networks. |
|---|---|
| Autores principais: | Pendão, Cristiano Gonçalves |
| Assunto: | Engenharia e Tecnologia::Engenharia Eletrotécnica, Eletrónica e Informática |
| Ano: | 2019 |
| País: | Portugal |
| Tipo de documento: | tese de doutoramento |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade do Minho |
| Idioma: | inglês |
| Origem: | RepositóriUM - Universidade do Minho |
| Resumo: | Wi-Fi Based Indoor Positioning solutions normally require complex and time-consuming deployment processes or have limited accuracy. Fingerprinting Matching is one of most used techniques for indoor positioning, which relies on a radio map that is normally created offline in a calibration phase by manual site survey. Wi-Fi Fingerprint can be used to locate regular mobile devices, such as smartphones, using only software and can be used in any indoor environment without being necessary to deploy additional infrastructure, relying only on the existing Wi-Fi infrastructure. However, in most cases the manual site survey is unpractical and involves a significant effort, even for small scale spaces. Moreover, due to the dynamic nature of radio environments, to maintain the system performance, the site survey has to be repeated often to keep the radio map updated. This process is not feasible in large spaces, and compromises the scalability of this type of approach. Solutions have been proposed to reduce the calibration effort, using collaborative site survey to create and maintain the radio maps, or by using Model-Based methods to approximate a radio map. However, the reduced calibration effort usually implies a lower positioning accuracy and higher computational requirements. In this context, FastGraph is proposed as a new solution able to provide unsupervised positioning using different devices, such as smartphones or autonomous machines, while automatically creating and maintaining a Radio Map. A 3D Force-Directed Graph is used to rapidly model the radio environment. The 3D Graph is iteratively constructed with data collaboratively collected by several devices. Orientation and motion information, obtained from different sensors, can be used to improve the Graph constrains. FastGraph is able to operate shortly after its deployment, without previous knowledge about the environment. The proposed solution uses a novel algorithm to automatically provide location while simultaneously updating the radio map; and estimate the position of the Access Points (APs) and location-specific radio propagation parameters. In addition, FastGraph does not rely on expensive hardware or requires high computational effort. The FastGraph approach may be used in different contexts. In addition to the indoor positioning, the radio maps created by FastGraph include supplementary information that can be used to automatically map the interference in Wi-Fi networks and even to automatically map the physical space. The described solution was deployed and evaluated in two very distinct real world spaces, an industrial environment and a regular office building. The experiments, in these two spaces, evaluated the several aspects of FastGraph, and considered scenarios where only Wi-Fi data is available and when the Wi-Fi can be combined with data from other sensors. The results suggest that the proposed solution has potential to provide interference information in wireless networks and provide positioning in different indoor scenarios, from regular buildings, to autonomous vehicles in industrial environments, with the possibility of being also extended to outdoor spaces using data from cellular networks, especially considering the characteristics of the upcoming 5G networks. |
|---|