A number of traffic control strategies can be implemented in Road Network Operations in order to improve traffic flow, prevent congestion and enhance throughput. ITS software – supported by a wealth of real-time data enabling accurate estimates to be made of the status of the road network – is used to develop optimal management and control strategies that support network policy objectives. These will vary from one location to another, but commonly include maximising traffic throughput, minimising delays and congestion, maintaining road safety for all road-users – including safe crossings for pedestrians and cyclists – environmental targets (to reduce noise levels and/or air pollution) and bus/tram signal priority for some locations.
Control methods include:
Urban Traffic Control (UTC), with:
Motorway control systems, including:
Field controllers are needed to implement these strategies. They are the “brains” of the local system, and provide the means for accessing, monitoring and controlling field equipment (such as a ramp meter, a traffic signal, or a vehicle detector).
Computer software is needed to provide these functions. Some of the functions that an ITS system software may be required to provide include urban traffic control, traffic control on arterial roads and motorway control systems.
Urban Traffic Control refers to a package of technologies aimed at managing and controlling traffic flowing over urban networks – to minimise delay, maximise efficiency, improve safety and reduce emissions and fuel consumption. A large part of urban traffic control involves software to optimising signal plans at intersections to achieve these objectives. This requires extensive sensor networks to collect real-time traffic information – for example, loop detectors, closed-circuit TV cameras and video image processing, or non-intrusive traffic detectors. Based on the collected information, intelligent algorithms aim to optimise the signal plans. Different levels of control and sophistication are seen in urban traffic control systems (See Urban Traffic Control).
Several types of field controllers are available which respond to traffic demands to facilitate turning movements and allow time for cross-traffic. In the USA the Type 170 Controller was developed in the early 1970s by the California Department of Transportation. Its successor – Type 2070 – was introduced in 1992. More recent examples are the NEMA signal controllers and the Advanced Traffic Controller (ATC – 2005) – the most advanced controller in the USA.
Traffic signal controllers work on the basis of a timing cycle that is broken into “phases” – the order in which each traffic stream is given green time, whilst other traffic is held at red. A simple cross-road intersection may have just two phases: North/South, and East/West. A busy four-way intersection, with large volumes of turning traffic, might have up to eight phases – one phase for each of the four traffic directions and a phase for each of the turning movements.
In the United Kingdom, a relatively new controller called Microprocessor Optimised Vehicle Actuation (MOVA) was developed to overcome some of the limitations associated with traditional Vehicle Actuation (VA) control. A unique feature of MOVA is that it has two modes of operation – one for congested traffic conditions and one for uncongested or free-flow conditions. For free-flow conditions, the aim of MOVA is to deal with any queues that have accumulated during the red phase. An algorithm assesses the traffic loads from different intersection approaches and determines whether extending the green time is beneficial. If it is, the green phase is extended to let traffic through. This continues until the controller moves to a different phase. During congestion, MOVA’s operational objective changes to maximise the capacity or throughput for the intersection as a whole.
Motorway control systems focus on better management of motorway segments to enhance capacity and increase throughput. Over the years, several Decision Support Systems (DSS) have been proposed and developed to help this process. These DSS can provide recommendations to traffic operators on possible traffic control strategies – such as dynamic route guidance, ramp metering, changeable speed limits and optimal signal timing.
Automated motorway control systems (sometimes referred to as Active Traffic Management or ATM systems), use different concepts to achieve their goal – such as speed harmonisation, temporary shoulder usage, dynamic routing and signing, junction control and ramp metering (See Highway Traffic).
Active Traffic Management has been widely implemented in Europe, and is becoming a tool for managing congestion (both recurring and non-recurring), in the USA as well. The main technological components of ATM are similar to the UTC systems and include extensive sensing and monitoring systems, communications, controllers, and intelligent algorithms.
The benefits of Active Traffic Management systems include:
Urban Traffic Control (UTC) systems require traffic signals, signal controllers, ramp meters and dynamic message signs (Variable Message Signs – VMS) to control traffic. They also require:
Different approaches and measures are used for real-time traffic management and control in urban areas.
Computer signal control systems first appeared in the 1960s when computers were first used to coordinate the traffic signal controllers for a group of intersections – but without the benefit of today’s “feedback” of information from the field detectors to the computers. In this type of system (known as open-loop control) the traffic plans that are implemented are not responsive to actual traffic demand. Instead, plans are developed “off-line” using data from historical traffic counts – and implemented according to time-of-the-day and the day of the week.
This system is quite basic, but it still offers several advantages including:
The next level of sophistication is signal control systems where information from field traffic detectors is fed-back to the central computer. The computer uses the information on current traffic conditions to select the signal plan to be implemented (closed loop control). Plan selection is made according to one of the following methods.
Here, the system has access to a database (library) that stores a large number of different traffic patterns along with the “optimal” signal plans (developed off-line) for each traffic pattern. Based upon information received from the traffic detectors, the computer matches the observed traffic pattern to patterns stored in the library, to identify the most similar pattern. The “optimal” plan associated with the identified pattern is then implemented. This type of adaptive traffic control system is often referred to as a First Generation system. Its distinguishing feature is that the plans, while responsive to traffic conditions, are still developed off-line. First Generation systems work on the basis of current traffic data and do not generally have traffic prediction capabilities.
In this method, the “optimal” signal plan is computed and implemented in real-time. The optimal signal timings are computed in real-time using current data on traffic conditions obtained from detector information. This requires sufficient computational power to make the necessary computations on-line. It also needs enough data from the vehicle detectors to make the calculations. The systems that develop plans on-line are classified as either Second-Generation or Third-Generation systems. They typically have a much shorter plan update frequency compared to First-Generation systems, typically every 5 minutes for Second-Generation systems, and from 3 to 5 minutes for Third-Generation systems. In addition, some Third-Generation systems use forecasts of traffic conditions obtained from feeding the detector information into a short-term traffic-forecasting algorithm.
There are a number of examples of adaptive traffic control systems in use around the world. Amongst the most widely accepted algorithms are SCOOT, SCATS and RHODES.
SCOOT (Split, Cycle, Offset Optimisation Technique) is an adaptive traffic control system developed by the United Kingdom’s Transport Research Laboratory (TRL) in the early 1980s. SCOOT operates by attempting to minimise a performance index (PI) – typically, the sum of the average queue length and the number of stops across the controlled network. In order to do this, SCOOT modifies the length of the cycle, the amount of green time given to each signal phase (known as the time “splits”), and the offset time for each set of signals (the time difference between the cycle start times at adjacent signals). SCOOT computes these calculations in real-time in response to the information provided by vehicle detectors.
The operation of SCOOT is based upon Cyclic Flow Profiles (CFP). These are presented as histograms (graphical representations of user-specified ranges) that show the variation in traffic-flow over a cycle – which is measured by loops and detectors that are placed midblock on every significant link in the network. Using the CFPs, the offset optimiser calculates the queues at the stop-line. The optimal splits and cycle length are then computed.
Several additional features have been added to SCOOT to improve its effectiveness and flexibility, including the ability to:
The Sydney Co-ordinated Adaptive Traffic System (SCATS) was developed in the late 1970s by the Roads and Traffic Authority of New South Wales in Australia. For operation, SCATS only requires stop-line traffic detection, not the midblock traffic detection that is necessary for SCOOT. This simplifies installation since the majority of existing signal systems are equipped with sensors only at stop-lines. SCATS is a distributed intelligence, hierarchical system that optimises cycle length, phase intervals (splits), and offsets in response to detected volumes. For control, the network of signals is divided into a large number of smaller subsystems, each ranging from one to ten intersections. The subsystems run individually unless traffic conditions require the “marriage” – or the integration – of the individual subsystems.
Since 1991, the University of Arizona has been developing a real-time adaptive control system called RHODES, which stands for Real-time Hierarchical Distributed Effective System. RHODES is designed to take advantage of the natural stochastic (random) variations in traffic flow, to improve performance – a feature which is missing from systems such as SCOOT and SCATS.
The RHODES system consists of a three-level hierarchy that decomposes the traffic control problem into three components: network loading, network flow control and intersection control:
Level 3 is concerned with intersection control – determining the optimal time to change traffic signals for the next phase sequence and whether the current phase should be shortened or extended. The time frame for control level 3 is typically in the order of seconds and minutes.
Where there alternative routes available the problem of optimising traffic on the network can be tackled mathematically. For dynamic route guidance (DRG) the “objective function” (an equation expressing the operational function that needs to be maximised or minimised) – is the measure of the highway network’s performance that needs to be optimised. For example the objective might be to minimise the total travel time for all vehicles. The decision variables are the proportions of traffic that splits at each diversion point – to optimise network performance. The traffic-splits define how traffic should be distributed. The aim is to model traffic flow in the region and ensure that it is maintained at the nodes and along the links of the network, without congestion setting in. ITS software can then be used to solve the problem of optimising the objective function and recommending a routing strategy that will vary in real-time according to traffic conditions. Routing advice will be given in traffic broadcasts and on VMS, or with in-vehicle route guidance for those vehicles that are equipped.
Public transport priority (known as Transit Signal Priority or TSP in the USA) is a measure aimed at reducing delay for public transport vehicles (buses, trams, taxis) at signalised intersections by giving their movements preferential treatment. The methods for doing this can be divided into passive and active strategies. The basic difference is whether specialised sensors and detectors are used to detect approaching public transport vehicles. Without supporting technologies to specifically identify these type of vehicles, passive TSP technology simply improves conditions for all vehicles along a public transport corridor.
Active technologies detect an approaching bus or tram (this is typically accomplished by having a transmitter on the vehicle that communicates with a receiver or detector on the roadside signal controller). Different algorithms or strategies are available for active bus priority. Amongst the most common are:
Green Extension: this extends the green time if a bus or tram is detected, to allow the priority vehicle to pass – up to a certain pre-determined limit. This strategy only benefits a small portion of vehicles, but the reduction in delay for beneficiaries is significant (equal to the length of the whole red interval)
Early Green: shortens the green time for conflicting phases, by a pre-defined amount of time – for example when the bus arrives whilst the traffic light is in its red phase. Early green benefits a large percentage of buses, but the saving per vehicle is not as large as for Green Extension
Phase Rotation: under this strategy, the sequence of green time for different manoeuvres at the intersection is changed so that the priority vehicle is not held up. One common modification – that allows the vehicle to cross the opposing traffic stream – involves swapping a dedicated turn signal at the start of the cycle (the “leading” phase) to the end of the cycle (the “lagging” phase).
Actuated Transit Phase(s): involves establishing transit phases, which are only active when a bus/tram is present. In this case, a special transit signal face would display, for example, a letter “B” for Bus or “T” for Tram.
Phase Insertion: this allows the same phase to appear more than once during the same cycle in order to serve the transit vehicle.
An important function that ITS software can provide for Road Network Operations is the ability to detect incidents and abnormal conditions from automatic analysis of the real-time traffic surveillance data. The development of Automatic Incident Detection (AID) algorithms began in the 1970s, and since then many algorithms have been developed. They have had mixed success, primarily because of their relatively high false-alarm rates (measured as the ratio of the number of false detections and the total number of observations).
AID algorithms can be broadly divided into four groups based on the principle behind the algorithm’s operation. These groups are:
These are among the most commonly-used AID algorithms. They are based on the premise that the occurrence of an incident results in an increase in the density of traffic upstream and a decrease in traffic density downstream. The California Algorithm is one of the earliest comparative-type AID algorithms to be developed – and is often used for comparisons and bench-marking. Since the original California algorithm was first developed refinements have been made to its performance. At least 10 new algorithms have been produced, of which algorithms 7 and 8 are the most successful. The TSC 7 algorithm represents an attempt to reduce the false-alarm rate of the original algorithm. The TSC 8 algorithm test repeatedly for congestion effects upstream of an likely incident and monitors other traffic characteristics.
Catastrophe Theory derives its name from sudden changes that take place in one variable that is being monitored – whilst related variables, also being monitored, show smooth and continuous changes. For incident detection, catastrophe theory algorithms monitor the three fundamental variables of traffic flow – namely speed, flow and lane occupancy (density). When the algorithm detects a drastic drop in speed, without an immediate corresponding change in occupancy and flow, this is an indication that an incident has probably occurred. The McMaster algorithm developed at McMaster University in Canada is a good example of an algorithm based on this concept.
Statistical and time series methods are used to forecast future traffic states or conditions. By comparing real-time observed traffic data with data forecasts, unexpected changes are classified as incidents. An example of these algorithms is the Auto-Regressive Integrated Moving-Average (ARIMA) time series algorithm. ARIMA is used to provide short-term forecasts of traffic occupancies based upon observed data from three previous time intervals. The algorithm also computes the 95% confidence interval. If observations fall outside the 95% range as predicted by the model, an incident is assumed to have occurred.
Several Artificial Intelligence (AI) concepts have been applied to problems in transport engineering and planning. Automatic incident detection is one application. Detecting incidents is a good example of a group of problems known as pattern recognition or classification problems – for which AI theories are quite effective in solving. Amongst the AI concepts most often applied to the problem of incident detection are Artificial Neural Networks (ANNs). These use complex algorithms and multiple computer processors to recognise patterns and connections in the input data.