DATA STREAM VISUALIZATION

Any visualization for stream data needs to include:

· Temporal dimension of the data

· Create a “streaming video” image of the data

· Facilitate viewing of output from multiple sensors simultaneously

· Provide a summary of the data suitable for examination by a domain expert.

No data actually exists at the Domain Expert Level. Instead visualization of the stream data and results of triggers and lookmarks are pushed to the user. A trigger may be used to notify the expert of anomalous behavior/rare events. Lookmarks, on the other hand, are events that are predefined as events of interest in the stream data.

We are aware of only one visualization technique which satisfies both of these features: Temporal Chaos Game Representation (TCGR) alternatively called a Temporal Heat Map (THM). The THM has its basis in CGRs used to visualize DNA/RNA sequences. .

Temporal Heat Map (THM) is a visualization technique for streaming data derived from multiple sensors. It is a two dimensional structure similar to an infinite table. Each row of the table is associated with one sensor value. Each column of the table is associated with a point in time. Just as stream data is infinite, the number of vertical columns is infinite. Each cell within the THM is a color representation of the sensor value. Rows may be normalized separately or together. Although different color schemes may be used, assuming a normalization of 0-1: colors with normalized sensor readings between 0 and 0.5 range from zero to blue and colors for values from 0.5 to 1 range from blue to red.

An illustration of the use of a THM for VoIP traffic (provided by Cisco Systems) data can be seen here:

Visualization of VoIP Data

In this video each row represents a type of VoIP call and the columns represent discrete time intervals. The colors indicate the frequency of calls of that type. As you look at this image the typical pattern of behavior (weekend vs. week day, most frequently used calls, etc.) can be seen. In addition the anomalous behavior of more frequent phone calls on some nights and some weekends is clearly observed.