The symbolic models of the world are inspired in human mental models of space as studied in psychology. In the scope of this research project, the focus was on a graph-based symbolic world model built upon four basic components: objects, object attributes, relations and relation attributes. Figure 1 shows an example of such a model.
This model is meant to be distributed over a federation of servers. Each user may build their own personal local model and insert relations to other available models. As the number of interrelated objects grows, a wider-ranging global model starts to emerge.
Figure 2 shows our system architecture.
The main topics of this research were:
- Modelling of the space dynamics
- What is the relation between the dynamics that may be observed in situ and the dynamics that can be captured by the deployed sensor networks (e.g., WiFi network)
- WiFi data analysis based on:
- The number of users per access point
- Lists of MAC addresses per access point
- Anomaly detection techniques
- Utilization level difference
- Pearson correlation
- Spearman rank correlation
- Development of a set of processing modules for automatic model feeding:
- Classification of access points based on the utilization rate and the average number of users;
- Correlation between pairs of access points
- Hotspots detection based on the data generated by the two previous modules and on the distance between correlated access points
- Tenant level calculation which indicates if a place is visited mostly by its tenants, by frequent visitors or by strangers
- Detection of groups of users
- Development of an editing application for manual model feeding (SC Space Editor)
- Development of a pedestrian navigation application (Route Finder)
The prototypes of our applications may be found on the following links:
- SC Space Editor: http://local.dsi.uminho.pt/scse
- SC Browser: http://local.dsi.uminho.pt/scb
- Route Finder: http://local.dsi.uminho.pt/route
This research project resulted in the following publications:
- K. Baras, A. Moreira, “Anomaly Detection in University Campus WiFi Zones”, in Proceedings of the 7th Workshop on Managing Ubiquitous Communications and Services, Mannheim, 2010.
- K. Baras, A. Moreira, “Symbolic Space Modeling Based on WiFi Network Data Analysis”, in 7th International Conference on Networked Sensing Systems (INSS’10), Kassel, 2010.
- K. Baras, A. Moreira and F. Meneses, “RouteFinder – Symbolic World Model Assisted Navigation”, International Conference on Indoor Positioning and Indoor Navigation (IPIN), Zurich, 2010.
- A. Moreira, F. Meneses and K.Baras, “Self-location and navigation without maps”, demonstration, International Conference on Indoor Positioning and Indoor Navigation (IPIN), Zurich, 2010.
- K. Baras, A. Moreira, “Groups and frequent visitors shaping the space dynamics”, The 4th Conference on Smart Spaces ruSMART 2011, S. Petersburg, 2011.
- K.Baras, “Dynamic world models for context-aware environments”, Thesis (PhD), University of Minho, 2011.