Investigation into Performance of Intelligent Transportation Systems, 10-R9634Printer Friendly Version
Inclusive Dates: 06/12/06 10/12/06
Background - A typical intelligent transportation system (ITS) that is developed and deployed by the Institute will commonly have well in excess of a million lines of source code. As software functionality increases, the performance requirements for the Graphical User Interface (GUI) become higher. New advances in computer processor technology provide the ability for the user interfaces to exploit dual processor technology. This quick-look effort investigated the performance of an Intel "dual core" technology-based workstation when executing the SwRI-developed advanced traffic management systems GUI software, as well as vendor-supplied video decoding software on the workstation (this software is used to display traffic video). Additionally, the project performed scalability testing of the ITS software to determine how well it functions when loaded with thousands of traffic speed detectors (plan to test 1,000 to 4,000 detectors).
Approach - To perform the testing, two different ITS applications (with client approval) were used to "inject" data into an operational environment. One application simulated a roadway detector and the other application simulated a control center; these two data sources can be combined to create a data stream of any desired size. The data stream was set up so that large numbers of detectors can be simulated. After the test environment was established, the performance of the existing GUIs was tested on both single central processing unit workstations and dual core workstations. The testing included both browser-based GUI and a Windows-based GUI. Throughout the testing, software decoding of video streams was also performed on the workstations. The tests were structured so that both the parallelism of the application and that of the operating system could be exploited to the fullest extent.
Accomplishments - The data from the testing demonstrated that software video decoding did not have a significant impact on the performance of the GUIs. It was realized that architecturally, in particular with the browser-based GUI, that some implementation alternatives were available that could allow the browser performance to be improved by an order of magnitude, which is very significant in the ITS environment.