Red Light, Green Light: A Discussion of Traffic Flow Optimization
Traffic is often thought of as a kind of fluid flow network, much like a complex thermodynamic system or electrical circuit. However, what is unusual about this fluid network is that the flow is discontinuous, halting at red lights and resuming at green. It is the goal of the traffic engineer to optimize the traffic flow within a given road network. Within many major cities, such as New York, are engineers who monitor and often manipulate the timing of many traffic lights in response to current traffic conditions.
However, it is difficult, if not nearly impossible, to have each traffic light manually controlled—and it may not even be effective. It is hard to tell how the timing of one light might affect the overall traffic flow through the complex network. In New York, therefore, only certain parts of the network are manually controlled, while the lights on the major routes are autonomously controlled. Traffic engineers try to integrate automated timing strategies into the network that will optimize the traffic flow.
Optimization can involve multiple functions. In New York, during normal traffic conditions, engineers often try to minimize the time that a commuter might experience. From 7 to 10 AM, for example, the lights favor the in-flux of commuters on the Queensboro Bridge, and from 3 to 7 PM, the lights favor the great volume of traffic leaving Manhattan for night. At all other times, the lights are set to maximize overall efficiency within the city network. Often, however, the engineers need to directly control the flow of traffic to avoid a specific section of the road network, for example, if there has been an accident. As a result, engineers look for efficiency within a strategy as well as controllability. Multiple papers propose strategies, an important one of which was written by Brockfeld, Barlovic, Schadschneider, and Schreckenberg of University of Cologne, which discusses and analyzes three strategies: the synchronized switching, “green wave,” and the random switching models.
In examining the effectiveness of a network, individual links can be picked out from the network and they can be individually examined. Using this technique, it becomes immediately apparent that synchronized switching is ineffective. A network employing this system synchronizes all the lights along a stretch of road such that they switch simultaneously, but the periodic starting and stopping slows traffic considerably and makes it harder to control.
A “green wave” strategy is much more effective. The idea of a “green wave” of lights is to minimize the number of times a car needs to stop along a stretch of road. The timing of the switching is staggered based on the time it would take for a car to get from one intersection to another. In New York, the “green wave” strategy is employed on major roads. Light switching is staggered by 10 sec so that cars travelling at 30 mph (the NY speed limit) can smoothly pass through multiple intersections without having to stop, resulting in smoother traffic flow and happier drivers. From the driver’s perspective, the “green wave” model essentially eliminates many of the nodes within a portion of the network, whereas synchronized switching does not. However, the oscillation of traffic movement does not disappear; the period is just longer.
A third model, namely random switching, promises to eliminate the oscillatory behavior of the synchronized and “green wave” model. The random switching brakes up the oscillatory traffic pattern and instead provides a more steady-state flow trough a network. As a result, it allows a traffic engineer to have more control over the traffic on that road. Nugrahani, Alamsyah, and Ramdhani of Bogor Agricultural University claim that the random switching model in their analysis even yields the optimal traffic flow overall.
The question is therefore, why doesn’t New York employ the random switching model? Perhaps it would be more efficient, but it may have its set of issues. The unpredictability of the traffic lights could be aggravating to drivers and perhaps even dangerous. More research has to be done, and though traffic handling is becoming more efficient through the use of adaptive traffic lights (traffic lights with sensors that can detect oncoming traffic and adjust their timing), the shear volume of traffic is also increasing. And traffic is often unpredictable and disjointed, so it may not be possible to produce a strategy that will perfectly optimize traffic flow for all situations.
–another driver
Sources:
http://www.nytimes.com/1998/09/17/technology/choreographing-the-dance-of-traffic-lights.html?src=pm
http://pre.aps.org/pdf/PRE/v64/i5/e056132
http://math.usm.my/research/OnlineProc/AM27.pdf