Objective

Minneapolis has extensive construction interfering with traffic.  As part of Salander’s (a leader in implementing intelligent work zone systems) contract to manage construction zones in Minneapolis, SMATS deployed its TrafficXHub sensors on Highway 35W to test its performance and in particular MAC address matching rate.

Current Technology Approaches

Bluetooth travel time sensors, that detect discoverable Bluetooth devices, have been an industry standard. The detection rate of these sensors has been a main concern of the traffic practitioners and has been reported to be between 2 to 5% of the traffic. However, in the past few years, the number of on-board electronic devices, operating in Bluetooth Discovery mode by default, has been getting lower and the detection rate has been dropping over time. The low detection rates can result in a statistically unreliable data sample for travel time measurement and estimation of origin-destination matrix. With poor sample sizes the travelling public is getting poor travel information.

SMATS Innovation

SMATS TrafficXHub™ pole-mount sensor is capable of detecting Bluetooth signals in Discovery and Connected modes, as well as WiFi signals and consequently produces a larger sample size. This becomes more critical especially when the variance of travel time increases.

Two TrafficXHub™ sensors were installed on Highway 35W in Minneapolis, Minnesota in the fall of 2018 in collaboration with Salander LLC . The two sensors were installed 1.7 miles apart. Data was collected for 6 days. The TrafficXHub™ was used to target Bluetooth Connected, WiFi, and Bluetooth Discoverable devices passing by in the traffic. The sensors were connected to the SMATS iNode™ data analytics cloud application through GSM communication. MAC address matching and outlier filtering were running on iNode™ in real-time during the project. The Houston Radar SpeedLanePro multi-lane traffic monitor, counter and classifier sensor was used to measure the ground-truth traffic volume to measure the detection rate.

Sensor's and link locations on the iNode map

Sensor's status page on iNode

 

Results

SMATS delivered the following test results:

  • iNode™ successfully filtered outliers and captured abrupt changes to the travel time during this period. The links travel time graphs and histograms were illustrated on iNode™. The travel time showed a near normal distribution pattern.

 

 

iNode link analysis interface
  • The detection rate of the TrafficXHub™ senor was measured based on the number of MAC addresses matched and filtered for each signal type divided by the actual traffic counts.  WiFi detection rate was 5%, the Bluetooth Discovery mode detection rate was recorded as 2%, and the Bluetooth Connected mode detection rate was 17%. These results clearly indicate that sample size for Bluetooth Discovery mode is low and relying only on this as the sole source of data can potentially result in statistically unreliable data. The total detection of the TrafficXHub™ sensor was found to be 24%.

TrafficXHub provides an innovation in ITS, ensuring travellers get accurate travel time information.

TrafficXHub™ sensor detection rate