TrafficScience produces reliable, high-quality real-time and historical traffic information, suitable for a range of applications and users. Our patented FVD^® technology gives us a distinct advantage in providing detailed information that covers wide geographic territories - often nationally - at a lower cost than was previously possible. To our partners, our software platform is the basis for their applications, generating new and increased revenue streams.

(Click image for larger version in new window)

TrafficScience operates on a COLLECT » DETECT » PUBLISH model.

• COLLECT: Taking data from multiple sources
  Data about the movement of mobile phones and vehicles is sourced from cellular networks, vehicle fleets, fixed sensor networks and government agencies. ITIS pioneered the use of Floating Vehicle Data® (FVD) from cellular networks, and it is key to us being able to offer traffic information with the greatest geographic coverage.


• DETECT: Adding value to the data
  The complex analysis performed within TrafficScience results in comprehensive high-quality information about traffic flows – both real-time and historical. Accurate prediction of traffic flows is vital in providing information that users trust.


• PUBLISH: Providing data in the form needed by diverse applications
  We work with public and commercial organisations to provide data to them in precisely the way they need to be able to further increase the value of the data in the applications those organisations build and use to deliver enhanced services to their customers. We work with partners to help them develop applications that use TrafficScience information, and in some cases develop applications directly for end customers.

Source data is progressively converted into high-quality information as it is moved through the software components of TrafficScience. In this simplified product architecture, the software components can be seen, and these are the actions that they take:

The first stages are used to process incoming Floating Vehicle Data from GPS devices and mobile phone networks.

Component	Action
Translator	Data can be sourced from a wide variety of network mediators and in this component, the data is first translated into a standard ITIS format. This will then enable events within the data to be identified, including the locations of devices and traffic incidents.
Mediator	In this component, all the data is completely de-personalised. Feeds from a number of sources can be combined here, and the performance of the feeds over time is tracked.
Dispatcher	Irrelevant events are filtered out of the data stream. The distribution of the data is assessed and outlying data that exceeds a threshold is dropped. An initial assessment of the event's geographical footprint takes place to filter out unusual events that cannot be usefully analysed. As the data is passed forward, load balancing between multiple Observers is carried out.
Observer	By examining the stream of associated events and their associated geographical footprints, it is possible to select the data that needs to be fully analysed. For example, data that is not coming from devices actually on the road network can be dropped. Events are associated so that the path of travel can be determined, and travel time over individual road segments can deduced from sequences of timed location related events.

At this point, data can be combined with incoming fixed-sensor data, and journalistic data which reports incidents and their causes.

Component	Action
Correlator	The data from several sources is correlated from several Observers so that event data from cellular and GPS devices can be linked to relevant incident data reported from journalistic sources, or fixed-sensor. By combining the data, it is possible to make more complete deductions about the meaning of what is being observed in the traffic flows on the road network.
Evaluator	The correlated data is now analysed in the context of the eventual use of the information, and in relation to the geographic (mapping) data. For example, if the data is to be used in historical analysis within a data warehousing application, larger groups of events may be gathered together in order for evaluation, whereas if the information is required by a real-time application, smaller groups of events may be fed forward into the next stage, in order to allow rapid analysis. By considering the temporal and geographical context, traffic incidents can be detected, and issues that are likely to originate from a common cause can be identified.
Predictor	This component makes a short term prediction of the traffic pattern, dependent on the intended application use, and feeds it forward to the web service application interface. Both the current situation and overall historical trends can be used in produsing the prediction.
Web Service	The data is presented to the application in the format required.

Geographic support comprises digitised maps of the road topology, correlated to information about the layout of the cellular phone network.

Sources of Data

TrafficScience uses ITIS’ patented Floating Vehicle Data (FVD^®) technology, which collects anonymous location data from devices travelling within mobile phone networks and GPS enabled fleets. This data can be collected over a wide area – typically nationally – and allows road network operators to monitor real time traffic information over far greater geographic areas and monitor much smaller classes of road than can be done with fixed-sensor systems.

While FVD^® technology is accurate enough to be used by itself, TrafficScience can combine FVD^® data from other sources such as traditional fixed sensor equipment and journalistic information from eye witness reports. This rich data set can be analysed to produce traffic information of great precision and accuracy.

ITIS’ FVD^® data collection does not require any fixed roadside equipment and as a result, the cost to install and maintain TrafficScience is significantly reduced vs. fixed equipment. The timeliness and quality of traffic information for road users and road network operators is significantly improved.