Gridlock on corridor

Arterial Performance Measures (APMs)

This is part 3 of 4 in our blog series on Traffic Insights. This post covers Arterial Performance Measures. Stay tuned for more!

In part one of this series, we covered the evolution of data-driven traffic operations. In part two, we talked about the first set of insights, Signal Performance Measures. Today’s blog covers Arterial Performance Measures (APMs). APMs are traffic insights that allow optimization of traffic corridors.

Part Three: Arterial Performance Measures

As a traffic pro, have you ever been stumped by the following questions:

  • Are my intersections in a corridor well coordinated?
  • Did that new timing plan adjustment fix my progression problems?
  • Is traffic flowing normally in my main corridors right now?

What seems like rudimentary information actually requires planning and systems. But why does it matter, you ask?

Corridor traffic flow is a critical factor in citizen satisfaction. It’s tracked by a discipline called arterial traffic performance, which are a set of metrics that measure travel times across heavy volume arteries in a city. Traditionally, legacy arterial performance systems were only able to measure performance of sequential intersections. Teams had to stitch together this data to get corridor-level views, which wasn’t always easy. But today, new tools make this much easier. At Miovision, we solve for this with APMs.

So What Do These “Insights” Look Like?

Here’s the list of out-of-the-box reports. All are customizable to your city and whatever traffic priorities top your list.

  • Point-to-Point Travel Time: A comparison of travel time between two points throughout the day, with ability for comparison to historical averages.
  • Corridor Congestion Plot: An analysis of where and when delay is occurring in a corridor.
  • Travel Time Index: The median travel time along a corridor expressed as an index relative to the free-flow travel time.
  • Planning Index: The 95th percentile travel time presented as an index relative to the free flow travel time. This metric indexes the typical worst-case scenario that a traveler should plan for.
  • Buffer Index: The difference between the Planning Index and Travel Time Index. This provides an indication of the perceived extra time that a traveler should plan for above and beyond the average travel time.  This metric is a good indicator of Travel Time Variability.
  • Progression SPM: Use of various Signal Performance Measures for the collection of intersection data along a corridor. It’s presented in ways that help evaluate the quality of progression along the corridor.

How Are These APMs Generated?

The process is called “wireless vehicle re-identification,” which counts vehicles by tracking MAC addresses from mobile phones. Here’s how it works within Spectrum.

Collect: Included with the Spectrum hardware at each traffic cabinet is the antenna used to transmit data back to the traffic management center using cellular LTE. This antenna supports Wi-Fi signal discovery, and is used to monitor the presence of Wi-Fi devices passing thru the intersection, such as mobile phones or even “smart” vehicles. The Spectrum antenna scans the intersection for Wi-Fi devices in its vicinity, and reads and records addresses within the active range.

Sort: By scanning continuously at all intersections, Spectrum can recognize or “re-identify” a MAC address as it enters and then exits the readable range. Comparing the multiple identifications of the same device, Spectrum can calculate how long it takes for vehicles to travel between two intersections, otherwise known as the “travel time”. Spectrum is able to calculate travel times for 5% to 10% of all vehicles travelling along a route, which produces a statistically accurate model of the true traffic conditions.

Analyze: Spectrum then performs sophisticated data analyses and filtering to combine the individual records to produce a total Travel Time Analysis for an arterial. This analysis indicates how the commute time, delay and congestion varied throughout the day, as well as easy comparison between different time periods.

An important note on privacy. (We’ve highlighted it in red because this is important.)

MAC addresses don’t contain personally identifiable information, but they could be used to “track” a specific vehicle in ways that violate citizen privacy. As soon as Spectrum detects a Wi-Fi MAC address, it is hashed or scrambled, using an algorithm that does not allow the original MAC address to be reverse engineered. The hash key changes every 24 hours to ensure that a single driver’s commutes cannot be correlated over time.

What do APM Dashboards and Reports Look Like?

Here are the two most popular.

Travel Time: This graph shows the selected single day’s travel time (orange) vs. the 12-week historical trend (blue). The 24-hour data is presented as individual vehicle captures (dots) and the associated median travel time (orange line). The trend is presented as median (blue dotted line) and variability bands of 80th, 90th and 95th percentile travel times are shown for the selected comparison window.

Travel Time

Travel Time

Congestion Scan: This index graph shows the travel time for different segments of the corridor and for various times during the day. This can highlight not only when, but also where congestion is building.

Congestion Scan

Congestion Scan

Part four of this blog series will appear in a few weeks. It will cover the final category of Miovision Traffic Insights:  Maintenance and Performance.

2 replies
  1. Sing Wong
    Sing Wong says:

    Quote: “Spectrum is able to calculate travel times for 5% to 10% of all vehicles travelling along a route, which produces a statistically accurate model of the true traffic conditions.”

    Is there a rule of thumb or guideline for the number of Spectrum units needed for an arterial? How would you size Spectrum for a grid system?

    • miovision
      miovision says:

      Hello Sing. Thank you for your question and our apologies for the delayed response.

      Rule of thumb for Number of Spectrum Units for Travel Time

      • The number of units required really depends on the data granularity an agency is seeking. For example, you could have a 5 mile corridor with five intersections and only one sensor at each end of the corridor. This would give you travel times for the corridor as a whole, but with no data granularity between intersections (ie. one travel time value for the entire 5 mile distance). The more sensors you have, one sensor at each of the five intersections for example, the better the data granularity between intersections (ie. one travel time for each “link” of two intersections).
      • Due to the relatively wide detection range of Wi-Fi sensors, the best deployment scenarios that we have experienced are sensors that are 1300-3300ft (400-1000m) apart.
      • The same guidelines apply for grid systems. It would be ideal to have sensors 1300-3300ft (400-1000m) apart in all directions.

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *