On April 27th the City of Toronto issued an RFP for an Adaptive Traffic Signal Solution, commonly referred to as Smart Signals. The RFP is intended to provide a 30-intersection pilot project in late 2017 and early 2018, with the option to expand the program to up to 500 signals over the next five years if the pilot is successful.

The project is part of the City's larger initiative to modernize its traffic signal infrastructure. While the City has been using earlier versions of Adaptive Traffic Signal Control at about 350 intersections since 1992, the technology is outdated, and far more advanced options are now available.

Adaptive Signal System in Lancaster, PAAdaptive Signal System in Lancaster, PA, image courtesy of LancasterOnline

This is not the first time that the City has issued an RFP for this. An RFP was issued in January 2016 that contains a nearly identical summary page, with the only changes being to the number of intersections (20 in 2016 vs 30 in 2017), and the completion dates. It is unclear why this RFP is in essence being re-tendered.

The purpose of installing updated Smart Signals is to reduce traffic congestion and delays at intersections by using adaptive algorithms to optimize traffic flow in real-time. This is accomplished by detecting traffic demand on all approaches through the use of specialized cameras, and making dynamic changes to offsets, splits, and cycle lengths at each intersection.

Most current signal systems either use fixed signal timings (which can be programmed to vary by the time of day), or use in-pavement detection loops to perform minor modifications to the signal timings based on demand. While these technologies are useful, their limitations (both in terms of detection and computing power) limit the degree to which traffic patterns through an intersection can truly be adapted to. The latest generation of Smart Signals have the ability to not only detect whether or not there is a car in the cue, but how many cars. This distinction allows the system to perform more detailed calculations on the amount of time that would be required in that phase in order to clear the queue, amongst other things. The latest generation also allows for better communication between signals, as opposed to the current system in which most signals are programmed and operate relatively independently, with only limited communication between them.

Vehicle detection using an Adaptive Signal SystemVehicle detection using an Adaptive Signal System, image courtesy of Smart CCTV Ltd

There are two scales of potential application of this technology. The first is Smart Signals at individual intersections, which (in theory) improves the throughput by adapting to traffic conditions in real-time. While this type of application is useful for select "problem" intersections, from a network perspective it won't have a substantial impact on travel times across a corridor.

The second, which is what the City's RFP is trying to do, is implementation of this technology on a network scale. The corridors identified in the RFP are:

  • Sheppard Ave East between Neilson Ave and Meadowvale Rd
  • Victoria Park Ave between Highway 401 and Altair Ave (between Sheppard and Finch)
  • Sheppard Ave East between Victoria Park Ave and Heron's Hill Way (just east of Highway 404)
  • Yonge St between Castlefield Ave (550 m north of Eglinton) and Yonge Blvd
  • Lawrence Ave West between Yonge St and Duplex Ave

By implementing this technology across a large section of a corridor, the signals will be able to communicate traffic data amongst themselves, allowing for real-time synchronization to minimize wait times at intersections. For example, if Signal A knows how many cars it released heading eastbound, the Signal B immediately east of there can determine when its eastbound phase should turn green (based on the travel time between intersections), and how long it should stay green for in order to ensure that all vehicles that were released at the earlier intersection make it through that subsequent one.

This also has applications for transit. If a signal detects that a bus has passed through the intersection, it can relay that information to the next signal downstream to extend its green phase a few seconds longer in order to allow that bus to pass through the intersection, minimizing the delay for the bus and its passengers. By contrast, if a signal notices that a bus is waiting in the queue, it can adapt the cycle timing to reduce the amount of green time on the opposite phase, in order to give that bus a green faster.

Intersections included in the Meadowlands Adaptive Signal SystemIntersections included in the Meadowlands Adaptive Signal System for Traffic Reduction, image courtesy of MASSTR

An example of this type of system has been implemented in the Meadowlands area of New Jersey, across the Hudson River from New York City. Called the Meadowlands Adaptive Signal System for Traffic Reduction, the project incorporates 144 traffic signals into a self-adaptive network, which is monitored and controlled in real-time at the NJMC Traffic Management Center. The system handles in excess of 3 million vehicles each day, and it is estimated that it reduces vehicle delays by more than 1.2 million hours per year, and gasoline consumption by more than 1.2 million gallons per year.

The RFP closes on June 8th, and implementation is scheduled to begin in September 2017, to be concluded by March 2018. You can join the discussion on Smart Signals by visiting our forum thread, or by leaving a comment below.