Current Research Projects

  Visual Analytics for Mobile Communication Networks

This applied research project proposal is a cooperation between NCVA (National Center for Visual Analytics), Linköping University and Ericsson Research, Linköping, facilitating a strong synergy between world-leading visualization technology and management of mobile communication networks. The project targets Visual Analytics (VA), whereThe project targets Visual Analytics (VA), where novel techniques will be developed and applied to smart network management prototypes for mobile communication networks. Research will focus on network management applications for mobile communication networks in 3GPP defined systems, such as Long Term Evolution (LTE). Additionally, the project targets operator-promoted Self-Organizing Networks (SON) features, which manage the complexity in large networks. The overall objective of the project is to prototype smarter network management applications for mobile communication networks. The system shall assist the operators in a smart way. The input to the system is a voluminous real-time flow of spatial-temporal and multivariate information with an objective to high-light the important problems to the operator, utilizing advanced visualization and data-mining techniques. Thus, the information is interpreted and transferred to the operator in a suitable way, which improves the comprehensibility and reliability. Particularly, it is important to provide operators with a system that ensure trust in automatic SON features in the mobile communication network, which are means to cope with the high complexity of today’s networks. The system shall be scalable from smaller networks up to networks with thousands of nodes. To the operator, the application shall be perceived as one integrated solution, although the network may comprise a multitude of radio-access-technologies (multi-RAT), like GSM, WCDMA and LTE, standardized by 3GPP. 

Choropleth Map with a Map Layer Architecture 

The choropleth map component facilitates a novel 100% Web-enabled implementation of a layered architecture for simultaneous transparent views of multiple map layers. It means that each class of spatial information is represented by its own layer, e.g. glyph, shaded map and Google map. These layers can then be combined and controlled to be displayed, hidden or transparent depending on the needs of the user.

Figure: GAV Flash Map Layer Component with four map layers pie glyphs, shaded map, country and Google map
You can test this Demonstrator at:

http://vitagate.itn.liu.se/GAV/eXplorer/OECDRegional/

Spatio-temporal and multivariate statistical data explorative visualization

eXplorer helps the analyst see patterns of events, relationships, and interactions over time within a geospatial context. The space-time-attribute data cube is used to conceptually explain the methodology in Explorer’s data handling. The data cube has three dimensions: geography (OECD regions), time (years), and attributes (indicators such as age groups, education etc). Each cell in the cube is defined by a specific spatial object (region), a specific time (year 2007), and a selected indicator (age group 65+). The value at that cell is the indicator value (Liguria, 1999, age 65+). Each region is represented by a horizontal slice in the data cube. Selected regions are represented as “time profiles” in the Time Graph above for a given indicator.

Figure: eXplorer is developed from GAV Flash components customized and optimized to sustain real-time coordinated time-linked views that are simultaneously updated with changing regional statistics data for every new time step. See also web site for demonstration: http://vitagate.itn.liu.se/GAV/eXplorer/OECDRegional/

Synergy between Choropleth Map and Tree Map for regional hierarchical data

This research demonstrates and reflects upon the potential synergy between the use of a squarified treemap dynamically linked to a choropleth map to facilitate visualization of social science regional data at several hierarchical levels aiming to discover statistical patterns that relate to significant characteristics of regions under study.

Figure: A squarified treemap ordered by population size and NUTS1 regions, dynamically linked to a choropleth map, both coloured by population for age group 65+, and applied on a limited (Italy) OECD regional hierarchical dataset. The hierarchical data structure is based on 5 levels continent, country, NUTS1, NUTS2 and NUTS3.

We are interested in finding answers about what methods and tasks are important when exploring demographical hierarchical data, such as general overview, trends over time, geographical patterns, indicator correlation, outliers and simultaneously mapping two dependent indicators such as age group and total population. For example, for a choropleth map screen space is always allocated depending on geographical area rather than an indicator of interest? Can the treemap compensate this weakness and are its known strengths and weaknesses applicable to the demographics data domain?

Figure: Linked treemap and choropleth map showing the ratio of children in the European OECD member countries. The colour of each region represents the percentage of the total population that falls within the 0-14 age group. Size in the Treemap shows the size of the total population.

Red areas have a high ratio of children and blue areas have a low ratio. Cell size in the treemap represents the total number of people in each region.
Four extreme clusters of regions immediately stand out in the choropleth map. South-eastern Turkey is one of them, with children in some parts making up almost half the population. On the opposite end of the distribution, with ratios down to nine percent, lie former East Germany, north-eastern Spain and northern Italy.
The additional information provided by the treemap that the choropleth map lacks is all based on the introduction of a second indicator; in this case the size of each region’s population. For instance, it becomes apparent that despite that Sweden is the third largest of the included countries measured in physical area, it only has a population of nine million – less than Turkey’s Istanbul region alone. It is also possible to see that while Germany and Turkey currently have almost identical total population numbers, Turkey has far more children. Over the next few decades Turkey is therefore likely to overtake Germany as the European OECD member country with the highest population.
 

Hi-res image: Ratio of 0-14 yrs, Map + Treemap

Hi-res image: Population change in oecd countries

Interactive Quantification of Categorical Variables in Mixed Data Sets

Data sets containing a combination of categorical and continuous variables (mixed data sets) are difficult to analyse since no generalized similarity measure exists for categorical variables. Quantification of categorical variables makes it possible to represent and analyse this type of data using techniques designed for numerical data.
Within this research project a quantification process of categorical data in mixed data sets has been developed. The research aims to utilize the efficiency of statistical data analysis as well as making use of the domain knowledge of an expert user. The quantification process uses statistical analysis to find relationships within the categorical variables. Information on relationships among the continuous variables is incorporated into the process using clustering methods. The process is carried out in an interactive environment using parallel coordinates as a visual interface. Within this environment the user is able to control the quantification process, analyse the result and modify it according to his or her knowledge.
For more information contact Sara Johansson