Traffic Signal Control with Connected Vehicles

Modern vehicles are equipped with many electronic sensors which monitor a vehicle's speed, position, heading, and lateral and longitudinal acceleration. Although the technology exists to do so, vehicles rarely communicate this information wirelessly to other vehicles or roadside infrastructure. Researchers are anticipating the deployment of wireless vehicle communication, and have begun developing applications that use this new technology to improve safety and reduce congestion. This system is known as connected vehicles.

Most traffic signal timing plans are designed to minimize vehicle delay based on the volumes seen in the past, not the present. In-pavement loop detectors and video detection are sometimes used to make small adjustments to timing plans, but are too inaccurate, expensive, unreliable, and limited in physical range too provide the level of detection needed to fully adapt to traffic in real time. However, in a connected vehicle environment, vehicles within 300 meters of an intersection could communicate directly and continuously with a traffic signal through a dedicated wireless channel. A new decentralized traffic control logic, the Predictive Microscopic Simulation Algorithm (PMSA), is presented which uses new data made available by connected vehicles to continuously adjust signal phasings to minimize anticipated vehicle delay over a 15-second horizon. Testing in simulation showed significant improvements in under-saturated conditions compared to a coordinated-actuated system, with as few as 25% of vehicles communicating. Performance worsens when the corridor is near capacity, suggesting the need for either a longer horizon period or alternative timing methods. The algorithm is unique among connected vehicle signal control algorithms in that it does not record vehicle movements, either individually or aggregated, nor does it re-identify vehicles along a corridor, thereby protecting driver privacy.

Other connected vehicle mobility applications have been proposed. Similar to the PMSA, most experience benefits when at least 20% of vehicles are able to participate, with benefits increasing with higher penetration rates. In an attempt to improve the performance of these applications at low penetration rates, two algorithms are presented to estimate the locations of non-communicating (unequipped) vehicles based on the behaviors of communicating (equipped) vehicles. The first algorithm, used on arterials, estimates unequipped vehicle locations based on observed gaps in a stopped queue, and simulates the forward movement of these vehicles using a commercial traffic simulation software package. Testing in simulation showed that the location estimation algorithm generated small improvements in the performance of the PMSA when compared to an equipped vehicle-only scenario at penetration rates of 25% or less.

The second algorithm, used on freeways, compares an equipped vehicle's actual acceleration with its expected acceleration as predicted by a car-following model. Unexpected behavior triggers the insertion of an unequipped vehicle estimate. This estimate is represented as a modeled vehicle, and moves forward as predicted by the car-following model until it is overlapped by a equipped vehicle, at which point it is deleted. Based on analysis from field data, the algorithm is able to predict the locations of 30% of vehicles with 9-meter accuracy in the same lane, with only a 10% of vehicles communicating. Similar improvements were found at other initial penetration rates of less than 80%. The algorithm was applied to an existing connected vehicle ramp metering algorithm, and was able to significantly improve its performance at low connected vehicle penetration rates, while maintaining performance at high penetration rates. Both the freeway and arterial algorithms are first attempts to estimate lane-level, individual vehicle locations in real-time based on the speeds and acceleration behavior of equipped vehicles from empirical data sets.