An important component of any ITS system is the network of sensors used to monitor the traffic performance throughout the freeway system. These freeway sensors are used to alert Traffic Management Center (TMC) dispatchers to incidents and to predict travel times for roadway users. Data quality is essential to maintain peak TMC operational efficiency and to maintain the public’s confidence in the information. The large number of sensors and data produced on a daily basis makes the use of human groundtruthing nearly impossible. Therefore an automated ongoing sensor data quality monitoring process must be implemented to identify the sensors in most need of attention.

This project proposes a system-wide heuristic approach to station health monitoring based on the principles of the “Six Sigma Process” and DMAIC Model for error identification and control. A test location on I-65 was outfitted with three different sensors; two side-fire radar sensors and 3M Microloop sensors. Data was collected and analyzed to assess the quality of sensor data, using performance metrics based on volume, speed, occupancy and Average Effective Vehicle Length comparison.

This study recommends combining sensor outputs into the single Average Effective Vehicle Length (AEVL) metric. Combined with the use of historical values and heuristic site knowledge the AEVL metric can provide a good tool for initial data quality control monitoring. Additional control efforts involve the use of portable side-fire radar units for temporarysensor co-location.

Report Number



Sensor Data Quality, Sensors, Performance Metrics, Average Effective Vehicle Length, Sensor Co-Location, SPR-3026

SPR Number


Date of this Version