4.3.1 Centralized Versus Distributed Architecture Approach

The classical ATMS is generally developed using a centralized processing architecture as a template. All traffic sensor data is processed and archived by equipment at a central operations center, and field devices are managed by equipment at the center. ATMS users, regardless of physical location, get their information from the center. A secondary operations center with nearly identical capability generally is designated as a back-up site.

This is a relatively simple architecture to implement but often does not scale well as the ATMS grows. The operations center typically runs out of space and failure recovery is typically slow. The CSC/PBFI Team has concluded that a distributed approach—shown by figure 4-1—offers compelling advantages for information distribution and communication efficiency, and offers superior ATMS availability relative to a centralized approach.

One of the most compelling advantages is ATMS survivability. Each distributed processing site is a logical microcosm of the total ATMS. Failure of a single site does not result in extended failure of the system because control of field devices is switched to an alternate operations center. This allows ATMS users to recover operating capability quickly. Further, any processing site can assume responsibilities of the central operations center back-up.

A geographically distributed ATMS architecture also lends itself to efficiencies of parallel operation and communications. For instance, instead of using expensive communications bandwidth to send raw traffic sensor data to the central operations center for processing, data can be processed and stored locally in a distributed database. It can then be sent to the central operations center when an incident or change in traffic flow is detected. When data is needed for further analysis, it can be retrieved by distributed query.