traffic signal assignment

Combining Traffic Assignment and Traffic Signal Control for Online Traffic Flow Optimization

• Conference paper
• First Online: 14 April 2023
• Cite this conference paper

• Xiao-Cheng Liao 10 ,
• Wen-Jin Qiu 10 ,
• Feng-Feng Wei 10 &
• Wei-Neng Chen 10  

Part of the book series: Communications in Computer and Information Science ((CCIS,volume 1793))

Included in the following conference series:

• International Conference on Neural Information Processing

1237 Accesses

6 Citations

With the continuous development of urbanization, traffic congestion has become a key problem that plagues many large cities around the world. As new information technologies like the Internet of Things and the mobile Internet develop, the interconnection between vehicles and road facilities provides a new mechanism to improve transportation efficiency. In this paper, we adopt the mechanism of vehicle-road coordination, and propose a new dynamic traffic flow optimization approach that combines the traffic assignment method and traffic signal control method together. For traffic assignment, a gene expression programming (GEP) based online navigation algorithm is proposed to generate a generalized navigation rule for the vehicles on the road network. Each vehicle can dynamically select an appropriate route for itself through the navigation rule based on its own states and information about the nearby road network. For traffic signal control, the Maximum Throughput Control (MTC) method is adopted. MTC checks the states of the intersections periodically and greedily takes the action that maximum the throughput of the intersections. By combining these two methods, the vehicle-road coordination mechanism can significantly improve the efficiency of city traffic flow optimization. The experimental results yielded based on the CityFlow simulator verify the effectiveness of the proposed approach.

This work was supported in part by the National Key Research and Development Project, Ministry of Science and Technology, China (Grant No. 2018AAA0101300), and in part by the National Natural Science Foundation of China under Grants 61976093. The research team was supported by the Guangdong Natural Science Foundation Research Team No. 2018B030312003.

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Subscribe and save.

• Get 10 units per month
• Download Article/Chapter or eBook
• 1 Unit = 1 Article or 1 Chapter
• Cancel anytime
• Available as PDF
• Read on any device
• Instant download
• Own it forever
• Available as EPUB and PDF
• Compact, lightweight edition
• Dispatched in 3 to 5 business days
• Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Similar content being viewed by others

Tri-level programming model for combined urban traffic signal control and traffic flow guidance

A graph-theoretic approach for optimizing signalized intersections under connected vehicle environment

Interchange flow control with dynamic obstacles optimized using genetic algorithms—a concept of virtual walls

Wardrop, J.G.: Road paper: some theoretical aspects of road traffic research. Proc. Inst. Civil Eng. 1 (3), 325–362 (1952)

Google Scholar  

Beckmann, M., Mcguire, C.B., Winsten, C.B.: Studies in the Economics of Transportation. Yale University Press, New Haven (1956)

Smith, M.J.: The stability of a dynamic model of traffic assignment-an application of a method of Lyapunov. Transp. Sci. 18 (3), 245–252 (1984)

Article   MathSciNet   Google Scholar  

Roughgarden, T., Tardos, É.: How bad is selfish routing? J. ACM (JACM) 49 (2), 236–259 (2002)

Article   MathSciNet   MATH   Google Scholar  

Akamatsu, T., Wada, K., Iryo, T., Hayashi, S.: A new look at departure time choice equilibrium models with heterogeneous users. Transp. Res. Part B: Methodol. 148 , 152–182 (2021)

Article   Google Scholar  

Osawa, M., Fu, H., Akamatsu, T.: First-best dynamic assignment of commuters with endogenous heterogeneities in a corridor network. Transp. Res. Part B: Methodol. 117 , 811–831 (2018)

Shou, Z., Chen, X., Fu, Y., Di, X.: Multi-agent reinforcement learning for Markov routing games: a new modeling paradigm for dynamic traffic assignment. Transp. Res. Part C: Emerg. Technol. 137 , 103560 (2022)

Han, S., Fang, S., Wang, X., Chen, X., Cai, Y.: A simulation-based dynamic traffic assignment model for emergency management on the hangzhou bay bridge. In ICCTP 2010: Integrated Transportation Systems: Green, Intelligent, Reliable, pp. 883–895 (2010)

Tian, Y., Chiu, Y.C., Gao, Y.: Variable time discretization for a time-dependent shortest path algorithm. In 2011 14th International IEEE Conference on Intelligent Transportation Systems (ITSC), pp. 588–593. IEEE (2011)

Peeta, S., Ziliaskopoulos, A.K.: Foundations of dynamic traffic assignment: the past, the present and the future. Netw. Spat. Econ. 1 (3), 233–265 (2001)

Larsson, T., Patriksson, M.: An augmented Lagrangean dual algorithm for link capacity side constrained traffic assignment problems. Transp. Res. Part B: Methodol. 29 (6), 433–455 (1995)

Zhao, D., Dai, Y., Zhang, Z.: Computational intelligence in urban traffic signal control: a survey. IEEE Trans. Syst. Man Cybern. Part C (Appl. Rev.) 42 (4), 485–494 (2011)

Hunt, P.B., Robertson, D.I., Bretherton, R.D., Royle, M.C.: The SCOOT on-line traffic signal optimisation technique. Traff. Eng. Control 23 (4) (1982)

Koonce, P., Rodegerdts, L.: Traffic signal timing manual (No. FHWA-HOP-08-024). United States. Federal Highway Administration (2008)

Lowrie, P.R.: SCATS: a traffic responsive method of controlling urban traffic control. Roads Traff. Author. (1992)

Diakaki, C., Papageorgiou, M., Aboudolas, K.: A multivariable regulator approach to traffic-responsive network-wide signal control. Control Eng. Pract. 10 (2), 183–195 (2002)

Varaiya, P.: The max-pressure controller for arbitrary networks of signalized intersections. In: Advances in Dynamic Network Modeling in Complex Transportation Systems, pp. 27–66. Springer, New York (2013). https://doi.org/10.1007/978-1-4614-6243-9_2

Wei, H., et al.: Colight: learning network-level cooperation for traffic signal control. In Proceedings of the 28th ACM International Conference on Information and Knowledge Management, pp. 1913–1922 (2019)

Chu, K.F., Lam, A.Y., Li, V.O.: Traffic signal control using end-to-end off-policy deep reinforcement learning. IEEE Trans. Intell. Transp. Syst. 23 , 7184–7195 (2021)

Ying, Z., Cao, S., Liu, X., Ma, Z., Ma, J., Deng, R.H.: PrivacySignal: privacy-preserving traffic signal control for intelligent transportation system. IEEE Trans. Intell. Transp. Syst. 23 , 1629–16303 (2022)

Noaeen, M., et al.: Reinforcement learning in urban network traffic signal control: a systematic literature review. Expert Syst. Appl. 199 , 116830 (2022)

Ferreira, C.: Gene expression programming: a new adaptive algorithm for solving problems (2001). arXiv preprint cs/0102027

Tang, Z., et al.: Cityflow: a city-scale benchmark for multi-target multi-camera vehicle tracking and re-identification. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 8797–8806 (2019)

Gudang Sub-distict. https://www.openstreetmap.org/#map=14/30.2813/120.1034 . Accessed 1 June 2022

Dijkstra, E.W.: A note on two problems in connexion with graphs. Numerische mathematik 1 (1), 269–271 (1959)

Hart, P.E., Nilsson, N.J., Raphael, B.: A formal basis for the heuristic determination of minimum cost paths. IEEE Trans. Syst. Sci. Cybern. 4 (2), 100–107 (1968)

Branke, J., Hildebrandt, T., Scholz-Reiter, B.: Hyper-heuristic evolution of dispatching rules: a comparison of rule representations. Evol. Comput. 23 (2), 249–277 (2015)

Download references

Author information

Authors and affiliations.

South China University of Technology, Guangzhou, China

Xiao-Cheng Liao, Wen-Jin Qiu, Feng-Feng Wei & Wei-Neng Chen

You can also search for this author in PubMed   Google Scholar

Corresponding author

Correspondence to Wei-Neng Chen .

Editor information

Editors and affiliations.

Indian Institute of Technology Indore, Indore, India

Mohammad Tanveer

Indian Institute of Information Technology - Allahabad, Prayagraj, India

Sonali Agarwal

Kobe University, Kobe, Japan

Seiichi Ozawa

Indian Institute of Technology Patna, Patna, India

University of Innsbruck, Innsbruck, Austria

Adam Jatowt

Rights and permissions

Reprints and permissions

Copyright information

© 2023 The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

About this paper

Cite this paper.

Liao, XC., Qiu, WJ., Wei, FF., Chen, WN. (2023). Combining Traffic Assignment and Traffic Signal Control for Online Traffic Flow Optimization. In: Tanveer, M., Agarwal, S., Ozawa, S., Ekbal, A., Jatowt, A. (eds) Neural Information Processing. ICONIP 2022. Communications in Computer and Information Science, vol 1793. Springer, Singapore. https://doi.org/10.1007/978-981-99-1645-0_13

Download citation

DOI : https://doi.org/10.1007/978-981-99-1645-0_13

Published : 14 April 2023

Publisher Name : Springer, Singapore

Print ISBN : 978-981-99-1644-3

Online ISBN : 978-981-99-1645-0

eBook Packages : Computer Science Computer Science (R0)

Share this paper

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

• Publish with us

Policies and ethics

• Find a journal
• Track your research

This ensures a consistent approach is taken throughout the system for working between individual signals and intersecting corridors. This approach also suggests that the absolute value of the cycle length is less important than maintaining a relationship between adjacent systems of traffic signals.

Critical Intersection Methods (Webster, HCM)

Much of early research regarding cycle length selection recommended evaluating the intersections identifying the critical intersection which was typically the intersections with the highest demand. A cycle length is established for this location and is selected so that it will be sufficient to maintain undersaturated conditions. This fundamental assumption is currently being investigated to determine if there are further strategies for dealing with oversaturated conditions.

The critical intersection approach considers a signalized intersection in isolation to other intersections and for this reason may not always yield the optimum cycle length. Most of the analytical tools developed for cycle length selection focus on undersaturated flow (12). The tools also do not consider the constraints of the intersections beyond the lost time and saturation flow rate. These critical intersection approaches to cycle length selection are primarily for isolated intersections and are all based on the assumption that vehicular delay is most important. This approach analyzes the intersection with the heaviest traffic to determine a minimum cycle length and used that to set the remaining intersections. The first step is to consider each intersection as though it is isolated to determine the minimum (optimum) cycle length needed at each intersection, as though it were isolated(13). The traditional models use Webster’s model to determine optimal cycle length. Webster used computer simulation and field observation to develop a cycle-optimization equation intended to minimize delays when arrivals are random. (14) The formula is as follows:

Equation 6-1

Figure 6-19 Webster's Optimum Cycle Length

In practice, much of the assessment of signalized intersections is completed using the Highway Capacity Manual (HCM) procedure on “Signalized Intersections.” The HCM provides few pieces of guidance on cycle lengths, but also notes limitations to the methodology. The current methodology does not take into account the potential impact of downstream congestion can have on intersection operation. Nor does the methodology detect or adjust for turn-pocket overflows and the impacts they have on through traffic and intersection operation.

The HCM offers a quick estimation method for the selection of a cycle length. The formula for cycle length estimation is as follows:

Equation 6-2

Primarily, this calculation is intended for planning-level analyses. This equation suggests that as the intersection approaches capacity, the cycle length should increase up to a maximum value, which the HCM suggests is set by the local jurisdiction (such as 150 seconds). The minimum cycle length suggested for use is 60 seconds. The equation does not explicitly address the pedestrian crossing requirements, left turn type and minimum green times necessary to meet driver expectancies.

Figure 6-20 HCM Cycle Length Estimation

In both Webster’s and the HCM’s estimation, the sum of the critical lane flows is a representation f the demand at the intersection. The critical lane is defined as the intersection approach with the reatest demand of all the approaches that are serviced during a given signal phase. For example, during the main street phase, on a street with two-way traffic, the critical lane would be the one lane n either direction that has the greatest demand. Y is the sum of the critical lane flows divided by 900, which is the percent of available intersection capacity that is in demand. If Y = 1, the intersection is saturated and the equation is no longer applicable. The situations where longer cycle lengths degrade intersection performance are a result of specific elements that lead to poor performance.

Longer cycle lengths will increase congestion in cases such as when:

• Upstream throughput exceeds downstream link capacity. Long cycles may move more vehicles through an intersection than can be handled downstream
• Turning bay storage is exceeded. Long cycle lengths may cause vehicles in left-turn bays to back up into through lanes. In a similar manner, long cycles may cause through traffic to back up beyond turn bays, restricting their access
• Increased variability in actuated green times. Long cycles result in high variability in the side street green time used, which may result in poor arrival types at the downstream intersection. This is particularly noteworthy when split times exceed 50 seconds (15).

The delay experienced by a motorist depends on cycle length and volume. Higher volumes always lead to longer delays. Shorter cycle lengths reduce delay, provided they do not result in inadequate intersection capacity. Oversaturated conditions require special considerations, and these models are not valid during that range of conditions.

Network Approaches

The network approach to cycle length selection considers multiple intersections to determine an optimal cycle length. Most applications of network approaches use signal timing optimization models.

There are a number of computer programs that can be used to assist in selection of a cycle length. The Federal Highway Administration's (FHWA's) Traffic Analysis Toolbox describes additional resources (https://ops.fhwa.dot.gov/trafficanalysistools/toolbox.htm). Three of the more popular programs of this type are Synchro, PASSER™ II, and TRANSYT-7F (16).

These signal timing optimization models consider the network being analyzed and determine an optimal solution based on a given set of inputs. The range of cycle lengths is based on the users input. The optimization models use the individual intersection characteristics, the volume to capacity ratios of each intersection, the link speed, and the distance between the intersections to estimate the performance for each individual cycle length and resulting plan. The models make assumptions based on the inputs related to the splits and offsets to determine performance measures that can be compared to timing policies. Because the models are imperfect a significant amount of effort is necessary to take an initially screened plan to a point that can be field implemented. The timing policies described previously, the optimization policies, and the criteria for determining which criteria to use to select a signal timing plan must be considered prior to and as a part of the optimization process.

A recent FHWA publication devotes a significant number of pages describing various optimization software packages available (17), so only the pertinent elements will be described here. The optimization models change with new versions of the software and for that reason, the documentation is best handled by the individual software producer. Their guidance related to the development of timing plans is most important.

The PASSER program uses the concepts described in Webster’s equation to determine the appropriate cycle length. The program uses a hill climbing algorithm to estimate delay at each intersection with the cycle length input to further quantify the performance of the system and maximize bandwidth using the pre-calculated splits as input to that model. At the optimization stage, it can find the cycle length, offsets, and phase sequences that produce maximum two-way progression. Essentially, PASSER uses the concepts described in Webster’s equation for selection of cycle length and, “after calculating the minimum delay cycle length for all intersections in the arterial street, the largest minimum delay cycle length is selected as the shortest cycle length for the system and that the longest allowable cycle length should be no more than 10 to 15 seconds longer than the shortest allowable cycle length to minimize the excess delay at the non-critical intersections.” (18)

TRANSYT-7F allows the user to define the performance function used for optimization. TRANSYT-7F was initially designed to select signal timings that produce minimum network delay and stops. Subsequent modifications added the capability to select several other objectives, including minimization of fuel consumption and maximization of progression opportunities. During its optimization process, TRANSYT-7F generates second-by-second flow profiles of vehicles on all links in the network and analyzes these profiles to determine performance measures. This model considers the formation and dissipation of queues in space. In addition, it accounts for flow interactions on adjacent links through a step-by-step analysis of all links in the system. TRANSYT-7F assesses cycle length by calculating equal saturation splits and applies a hill-climbing method to optimize signal offsets and splits.

In similar fashion, Synchro uses its algorithm to estimate arrivals at each intersection in the network and to calculate percentile signal delay, stops, and a queue penalty, which addresses the impact of queuing on arterial performance(19). The performance index is calculated for each cycle length based on the splits and offsets assumed within the model as constrained by the user. There are various steps to developing an “optimal” timing plan, but one of the limitations of this model is the inability to define the parameters within the Performance Index calculation.

Equation 6-3

6.6.3 Split Distribution

The splits operate as a part of the coordinated timing plan, essentially acting as another set of maximum green times for the non-coordinated phases. Once a cycle length is determined, split distribution is the process of determining how much of the cycle should be provided to each of the phases. These are maximum durations a phase may be served before it must terminate and yield to the next phase. Splits are typically allocated to provide a design level of capacity to all of the minor movements, with the remaining residual time allocated to the coordinated movement (21).

The coordination split for a phase i, expressed in seconds, is calculated by the sum of the green, yellow and red times, gi + yi + ri. A split for phase i, expressed as a percentage, is calculated as 100(gi + yi +ri)/C. This gi value is independent of the maximum green time for phase i and the walk and flashing don’t walk (if applicable). The actuated logic described in Chapter 5 applies to the phase and thus the phase may not use the entire split percentage allocated within the cycle. Figure 6-21 shows these three timers, the basic timing parameters for vehicles and pedestrians and the corresponding split time associated with the coordination plan. The coordinated phase is slightly different in this regard, in that it receives the remaining time available in the cycle length.

Determining adequate split times can be challenging. If a split time is too long, other approaches may experience increased delays, while if a split time is too short, the demand may not be served. There are often opportunities to vary controller parameters to allow for the fluctuations in daily traffic flow. As shown in Figure 6-21, there are many factors that should be considered in developing signal timing for both a single and a series of intersections.

Figure 6-21 How a phase times

Maximum green values may be ignored during coordination using the INHIBIT MAX feature in many controllers. This allows the phase to extend beyond its normal maximum green value.

There are various policies for determining the necessary split time for a movement. The intent with split times is to provide sufficient time to avoid oversaturated conditions for consecutive cycles, but over the course of an analysis period (15 minutes or one hour) split distributions seek to provide a volume to capacity ratio that is consistent with the operating agency’s design standards. In many cases, this will provide an opportunity for fluctuations to be met with the slack time or variable green time, and the actuated operation will reduce phases as necessary to maintain efficient operations. Slack time is defined as the additional time in a cycle that is more than the minimum split times for the phases at the intersection. One common policy allocates the green time such that the volume to capacity ratios for the intersection critical movements are equal in the coordinated cycle length. Another policy is to allocate a minimum amount of time to the minor streets and the remainder to the major or coordinated phases to enhance progression opportunities and maximize bandwidth. The latter methodology is used in traditional coordination, assuming the non-coordinated phases gap out. With many controller parameters and features, this allocation of green time can depend on pedestrians, transit phases, and gap settings.

Coordinated Phase

The length of the coordinated phase split is defined by the demand on the other movements. The coordinated phase receives the time within the cycle that is unused by the other phases of each ring.

There is a close relationship between the rings with the barrier on the intersecting street and therefore the various movements must be considered carefully. In periods of low demand on the non-coordinated phases, the coordinated phase may receive the entire cycle in the absence of an opposing call. The opposing call must be received before the permissive window expires.

Non-Coordinated Phase(s)

For a non-coordinated phase to time during a cycle, a call must be active, or the phase could be activated if a corresponding phase in the other ring is active and dual entry is enabled. For instance, if phase 4 has an active call, the last movement (phase 8 is most likely) will be green as long as phase 4 is active and dual entry is enabled for phase 8. Once the phase is active the basic signal timing settings (described in Chapter 5) and the split defined in the coordination plan determine the length of the phase.

6.6.4 Offset Optimization

Offsets should consider the actual or desired travel speed between intersections, distance between signalized intersections, and traffic volumes. In an ideal coordinated system, platoons leaving an upstream intersection at the start of green should arrive at the downstream intersection near the start of the green indication. For the users, this is a relative offset, where the time-distance relationship is observable and promotes progression. The actual offset is not always observable because of the actuated logic within the controller that can provide an early return to green.

The HCM suggests that an analyst should review the time-space diagrams to analyze arterial progression and the effectiveness of offsets for a set of signal timing plans. The actuated coordination logic of each signal controller causes the green time allocated to the side street to vary on a cycle by cycle basis. Thus, the time-space diagram is dynamic because of the phenomenon of “early return to green” that results from variable demand on the non-coordinated phases. The HCM translates this assessment of offsets into an arrival type that is used to modify the second delay term of the delay equation. Determining the quality of progression factor (PF) term of the HCM average intersection delay equation is a difficult task, even if observed in the field. In fact, a study where traffic engineers were asked to observe several identical video clips of vehicles arriving at a traffic signal indicating those subjective assessments had wide ranges in estimated arrival types.

Traditional methods for field optimization have included an engineer or technician observing in the field to determine whether the timing plan is operating and whether the offsets are effectively progressing traffic between intersections. These observations provide the engineer or technician with a limited ability to review conditions. With a 100-second cycle, there are 36 cycles during an hour, and the effect of an early return to green on one cycle may not be indicative of the next cycle’s performance. Assuming an observation period of at least three cycles leads the field performance assessment with a limited review of the conditions. Engineers are supplementing the field review with improved data.

6.7 COORDINATION COMPLEXITIES

This section discusses the various complexities of signal coordination. There are many variables that that must be considered to achieve an acceptable coordination plan. The guidelines address the following topics:

• Hardware limitations
• Pedestrians
• Phase sequence
• Early return to green
• Heavy side street volumes
• Turn bay interactions
• Oversaturated conditions 6-38

Each of these topics is addressed separately in the remainder of this section.

6.7.1 Hardware Limitations

The microprocessor-based traffic signal controllers used today allow vendors to add new features by changing firmware or software. Advancements in processing power have led to many developments (e.g., alternative phase sequences, number of available timing plans, etc) as well as a variety of signal timing practices. The incompatibility of the equipment and various functionalities has led to some frustration for maintenance personnel. Various government agencies and the National Electrical Manufacturers Association have been actively developing standards to address some of the inconsistencies in hardware.

Although the actual traffic signal system technology and standards have evolved significantly in recent years, many issues such as funding, political support, management, training, inter-jurisdictional coordination, and common regional visions for system operation have significant opportunities for improvement (22). Limitations of funding to upgrade older traffic signal controllers have led to systems that meet today’s needs, but “do not provide the building blocks for cost-effectively implementing integrated and interoperable systems.” (23).

A particular example of this, as it relates to coordination, is the use of various cycle lengths throughout a day and during different times of the year. It is conceivable that a different set of timing plans is needed for summer and winter months (associated with tourist or other traffic trends), and five different timing plans are desired during a typical weekday and five additional plans warranted on the weekend. This may require the traffic signal controller to operate 20 different plans, which may not be possible due to memory storage in older versions of the hardware.

6.7.2 Pedestrians

Pedestrian operations can have a direct impact on the ability to maintain coordination along an arterial. For some agencies, pedestrian crossing time is provided for all coordination plans within the split time for the phase, while other agencies (and controllers) allow traffic signals to suspend coordination when there is a pedestrian call, requiring the controller to resynchronize after a pedestrian call. Providing for pedestrian crossing time every cycle may result in a larger cycle and a reduction in green time available for main street movements. This may result in a less than optimal timing plan.

The provision of pedestrian timing and the effects of that pedestrian timing on coordination are two distinct concepts. Pedestrian timing is required for all phases that serve pedestrians. However, when pedestrian activity is relatively low, it may be desirable to allow a pedestrian call to have an impact on coordination because the network system is more efficient without accommodating pedestrians within the coordinated cycle length.

Pedestrian timing for non-coordinated phases

The effect of pedestrian timing on coordination is most commonly seen as it affects minor street timing. Figure 6-22 illustrates the basic principle of pedestrian timing for the minor street where the vehicle split is sufficient to accommodate the required pedestrian time.

Figure 6-22 Non-coordinated phase operation with pedestrian timing completed before the force-off for that phase.

When the split for the phase in question is not sufficient to cover the pedestrian timing, the controller times the phase beyond its force-off point, as illustrated in Figure 6-23. The response of the controller depends on a two factors: (1) demand for subsequent non-coordinated phases and (2) non-actuated versus actuated operation for the coordinated phases.

Figure 6-23 Non-coordinated phase operation with pedestrian timing exceeding phase split

When the coordinated phases are non-actuated, the coordinated phases must begin timing sufficiently in advance of the controller’s yield point to enable full vehicle timing (minimum green) and pedestrian timing (walk plus flashing don’t walk). Should the amount of time be insufficient to cover these timing requirements, the controller will time the coordinated phase past the yield point and fall out of coordination, as shown in Figure 6-24. It is at the yield point that the controller logic determines the method by which the controller will transition back into coordination.

Figure 6-24 Loss of coordination due to pedestrian call

As a general rule, it is desirable to accommodate pedestrian timing entirely within the split for a given phase. By doing so, any pedestrian calls that may occur can be accommodated without causing the controller to time the phase beyond its force-off point. In these circumstances, the controller loses coordination and must transition back into coordination.

In practice, it is possible to use smaller splits than are needed to cover pedestrian timing without adversely affecting coordination. The ability to do this depends on the capability of the controller. For example, using one particular brand of controller, coding vehicle split times of 85 to 90 percent of the pedestrian timing (walk plus flashing don’t walk plus vehicle clearance interval timing) results in an immediate loss of coordination. In other cases, when these force-offs are combined with cycle lengths that are long enough to allow a controller to temporarily shorten its cycle length during transition without violating the controller minimum (typically at least 10 percent greater than the controller minimum cycle length), the controller will typically resynchronize within a cycle or two, thus having minimal adverse effect on coordination.

A questionnaire survey from the NCHRP 172 project (24) indicated that as a general rule, pedestrian minimum time should be used for the side street when a pedestrian call occurs more than 20 percent of the cycles.

Pedestrian timing for coordinated phases

The amount of time needed to serve vehicle volume or provide bandwidth along the major street usually results in coordinated phase splits that are sufficient to accommodate pedestrian timing. Many controllers require that pedestrian timing be accommodated within the coordinated split timing to allow any type of coordinated operation.

For controllers operating with non-actuated coordinated phases, the major street splits must be large enough to accommodate all vehicle and pedestrian minimum timing requirements. For actuated coordinated phases, however, it is sometimes possible to provide a split for the coordinated phases that is less than that required to serve pedestrians. In practice, this works acceptably only if (1) pedestrian demand along the major street results in relatively few pedestrian calls, and (2) demand for the non-coordinated phases is frequently less than the split. In these cases, the controller can take advantage of the unused time from the non-coordinated phases to serve the coordinated pedestrian timing without passing the yield point and falling out of coordination.

6.7.3 Phase Sequence

The sequence of phases, particularly those of left turns, can provide measurable benefit to arterial operation. The most common phase sequencing decision, whether to lead or lag left turns, can have a particularly strong impact on the ability to provide bandwidth in both directions of an arterial. Other phase sequence decisions, such as the sequence of left turns on the minor street or the sequence of split phasing on the minor street, do not directly impact arterial bandwidth but can affect arterial delay. These two concepts are discussed in the following sections.

Major street left-turn phase sequence

Modern controllers allow left turn phase sequences to be varied by time of day. This has traditionally been done only for protected left-turn operations, but the use of specific display techniques allows this to be extended to protected-permissive operations (see Chapter 4).

The basic concept of lagging a major street left turn is to time the left turn after the opposing through movement (assumed to be one of the coordinated phases) terminates. Figure 6-25 illustrates a typical time-space diagram showing an arterial with only leading lefts and the same arterial with both leading and lagging lefts. The arterial demonstrated in the figure has a major intersection on each end and a minor intersection in the middle. As can be seen in the figure, a lagging left turn at the middle intersection facilitates better progression in both directions because it allows the two platoons to arrive at different times in the cycle. In addition, the two major intersections benefit to some degree from selective lagging left turns.

Figure 6-25 Vehicle trajectory diagrams showing the effect of changes in phase sequence

Figure 6-25a is a vehicle trajectory diagram for an arterial with only leading left turns on the major street.

Figure 6-25b is a vehicle trajectory diagram for the same arterial but using selective lagging left turns on the major street.

One of the potential consequences of lagging left turns that are actuated is that the end of the adjacent coordinated phase becomes less predictable. In terms of dual-ring operation, the lagging left turn is typically served after the deterministic (yield) point is reached. The lagging left turn extends the concurrent (adjacent) through movement time indirectly, not as a result of any particular timing within the coordinated phase itself. As a result, only the detection for the lagging left turn is used to determine when to gap out the lagging left turn phase and the adjacent coordinated phase. Therefore, it is possible that the adjacent coordinated phase may gap out earlier than expected from cycle to cycle.

One technique that has been used to eliminate this variability is to use a maximum recall on the lagging left-turn phase. In most controllers, this can be set by time of day and can often be paired with the specific timing plan containing the lagging left turn. The use of a maximum recall on the lagging left turn makes the end of the adjacent coordinated phase more predictable. On the other hand, if the demand for the lagging left turn is highly variable or is less than the split coded, the use of the recall on the lagging left turn may give the appearance of sluggish operation or defective detection.

In addition to the operational differences associated with lagging left turns, some have expressed concern over potential safety differences with having left turn sequencing changing by time of day. Even if lagging operation is used during coordinated operation during the majority of the day, the intersection is often configured to revert to leading-left operation when operating free (uncoordinated) during nighttime operations. There is no definitive research offering consensus on whether changing the left-turn sequence throughout the day has any negative safety consequences.

Minor-street phase sequence

It may be advantageous in some circumstances to adjust the left-turn phase sequence for the minor street. In doing this, it may be possible to reduce the delay and queuing for minor-street left turns as they enter the major street and arrive at downstream intersections. Although such adjustments may affect system-wide delays and stops, they will have no effect on the theoretical bandwidth for the coordinated phases.

6.7.4 Early Return to Green

One of the consequences of coordinated, actuated control is the potential for the coordinated phase to begin earlier than expected. This “early return to green” occurs when the sum total of the time required by the non-coordinated phases is less than the sum total of the vehicle splits coded for the phases. While this may reduce delay at the first intersection, it may increase system delay because of inefficient flow at downstream intersections or, most important, the critical intersection of the network. Figure 6-26 illustrates this within a time-space diagram.

Figure 6-26 Time-Space Diagram Example of Early Return to Green

Figure 6-26 shows that vehicles in coordinated phases that begin early may be forced to stop at one or more downstream intersections until they fall within the “band” for that direction of travel. This can result in multiple stops for vehicles and a perception of poor signal timing.

Early return to green can have a substantial negative effect on the performance of the coordinated phases. Research for offset transitioning has been completed to “smooth progression of a platoon through an intersection using the volume and occupancy profile of advance detectors” (25). Early return to green can be difficult to manage along a corridor, and it rarely can be completely prevented without eliminating most of the benefits of actuation. One technique that is sometimes used is to delay the start (offset shifted to the right) of the coordinated phase at a critical upstream intersection with sufficient non-coordinated demand (thus making its operation more predictable). Similarly, minor intersections downstream of this critical offset can be started earlier (offset is shifted left in a time-space diagram) to minimize the likelihood of a stop due to an early return to green. In either case, the engineer should use caution when shifting offsets to address early return to green in one direction may adversely affect operation in the opposite direction.

6.7.5 Heavy Side Street Volumes

Heavy side street volumes can affect the ability to progress through movements along an arterial. These volumes can come from either signalized intersections within the coordinated signal system or from unsignalized intersections, or from driveways between coordinated signals. Interchanges are a common source of heavy side street volumes.

In many cases, this additional demand proceeds along the remainder of the arterial and becomes part of the major street through demand at downstream intersections. However, this demand often enters the system outside the band established for through movements traveling end-to-end along the arterial. It is usually desirable to adjust downstream intersection timing to allow these heavy side street movements to proceed with a minimum of stops. In these cases, solutions that seek to optimize arterial bandwidth may be counterproductive to effective signal timing.

6.7.6 Turn Bay Interactions

Turn bay (or turn pocket) interactions can significantly reduce the effective capacity of an intersection. This is experienced when either demand for the turning movement exceeds the available storage space or when vehicle queues block the entrance of a turn bay. If left-turn demand cannot enter the left-turn bay due to impeding through vehicles, the left-turn phase will gap out early due to a perceived lack of demand. This results in some left turning traffic requiring more than one cycle to be served. In addition, the through movements lose capacity due to the impedance of left turns. In some cases, this may effectively remove an entire through lane from being able to effectively serve through demand.

Turn bay overflows also adversely impact progression. The impedance created by left-turn vehicles stored in the through lane prevents through traffic from proceeding to downstream intersections. As a result, any platoon of through vehicles passing through the affected intersection tends to be dispersed. This reduces the ability for downstream intersections to efficiently provide green time for the platoon.

Turn bay overflows can occur under both under- and oversaturated conditions. Even if enough green time is provided to serve a given turning movement, turn bay overflow can occur if the available storage is insufficient to store the queue for a given cycle. For cases where the turn bay is fed from a two-way left-turn lane, turn bay overflow rarely has significant adverse consequences for the adjacent through movements. For locations with raised medians, on the other hand, turn bay overflow can result in left-turning vehicles extending into the adjacent through lane.

Turn bay overflow can be managed in a number of ways:

• Turn bays can be extended to accommodate the necessary storage. This is typically the best solution, but it may be infeasible due to physical constraints, access needs, turn bay requirements for adjacent intersections, or other factors.
• Shorter cycle lengths can be used to keep overall queue lengths shorter, thus reducing the likelihood of overflow or turn bay blocking.
• If the left turn is protected, protected-permissive left-turn phasing may be considered to allow some of the left turn demand to be processed during the permissive portion of the phase. This reduces the overall queue length.
• If the left turn is permissive, protected-permissive left turn phasing may be considered to provide a period of higher saturation flow rates (the protected portion of the phase). This technique, however, may result in longer cycle lengths that partially offset the gain in capacity. Some agencies use a lagging protected period after the permissive period that is called only if there remains any unserved left-turn demand at the end of the permissive portion of the phase.
• Conditional service for the phase may be invoked, bringing up the movement twice during the cycle.
• If two receiving lanes are available, the adjacent through lane can be designated as a shared left-through lane and the phasing changed to split phasing. While this is rarely desirable for major street movements, it may be an appropriate solution for minor street movements.
• At an intersection with one heavy left-turn movement on the major street, it may be preferable to designate the left-turn phase as one of the coordinated movements paired with the adjacent through movement (e.g., phases 1 and 6 or phases 2 and 5). This allows any unused time to roll over to the left-turn phase of interest, thus reducing its effective red time and associated queue formation.

6.7.7 Critical Intersection Control

A challenging aspect of timing an arterial street or a network of streets is the need to provide enough capacity for major intersections without creating excessive delay for minor intersections. Ideally, all of the intersections to be coordinated operate optimally with similar cycle lengths. However, most arterial streets do not have this optimal arrangement due to a mixture of simple signals (e.g., two phases) with more complex signals (e.g., eight phases), wide ranges in cross street volume (e.g., major arterials versus collectors), and variations in left-turning volumes.

The techniques to determine the ideal cycle length for each intersection in isolation were covered earlier. The critical system cycle length is the maximum optimal cycle length of any intersection in the system.

Several techniques can be used where there is a significant disparity in the ideal cycle length for each intersection:

• Each intersection is timed using the critical system cycle length. This ensures the ability to coordinate all of the intersections in the system. However, it may result in excessive delay at minor intersections.
• Each intersection is timed to either the critical system cycle length or to half that value. This technique is commonly referred to as “double cycling” (a minor intersection cycles twice as frequently as a major intersection) or “half cycling” (a minor intersection has half the cycle length of the major intersection). This method can often produce substantially lower delays at the minor intersections where double cycling is employed. However, it may become more difficult to achieve progression in both directions along the major arterial, which may result in more arterial stops than desired.
• The major intersections are operated free, and the minor intersections are coordinated using a shorter cycle length. Because the major intersections are operating free, it is impossible to provide coordination through the major intersections. Therefore, major street vehicles are likely to stop at both the major intersection and at a downstream intersection due to randomness in arrival at and departure from the major intersection. This technique can often result in lower overall system delay at the expense of additional stops along the major street.

6.7.8 Oversaturated Conditions

Timing for oversaturated conditions requires different strategies than those used for undersaturated conditions. An intersection that is operating at or over capacity requires all movements to operate at a saturation flow rate to serve demand. Beyond this, the timing plan may favor the movements with the most lanes to maximize the throughput of the intersection. Obviously, the timing plan must consider whether undesirable effects such as turn bay overflow or other conditions exacerbate the problem.

Under these conditions, arriving vehicles must join the back of a queue to ensure that they enter the intersection with a minimum amount of headway (maximum saturation flow rate); however, it is impossible for vehicles on the arterial to maintain a travel speed traditionally desired for coordination.

In addition, an oversaturated approach cannot serve all arriving demand, thus creating a residual queue at the end of a cycle that carries over to subsequent cycles. This residual queue depends on demand conditions and can grow from cycle to cycle. Even when demand drops, the residual queues create saturated conditions beyond the time period when the arriving demand would create saturated conditions by itself. These residual queues can extend to adjacent intersections and prevent traffic from exiting upstream intersections if the intersections are closely spaced.

The general technique for accommodating oversaturated conditions involves managing queues. The following sections present options available for accommodating oversaturated conditions, including benefits and trade-offs.

Queue storage on minor movements to favor major movements

It is sometimes possible to accommodate oversaturated conditions by favoring the coordinated movements at the expense of minor street movements. Under this strategy, the coordinated phases are timed for a volume-to-capacity ratio typically no higher than 0.95 to 0.98. The minor movements receive less green time, which results in demand-to-capacity ratios exceeding 1.0. This method has a few advantages. The major street receives priority, which helps maintain traffic flow along the major street. This typically benefits the heaviest movements through the intersection, as well as the transit and emergency vehicles who frequently use the major street. In addition, demand held on side streets cannot enter downstream intersections, thus improving the actual downstream flow rate.

However, this type of timing strategy can have significant disadvantages. While this method can theoretically keep the major street moving, it creates extensive delay and queuing on side streets. This can have negative safety repercussions and highly negative public feedback. More importantly, it may be impossible to reduce splits for side-street through movements due to pedestrian timing requirements. If pedestrian calls are frequent enough to cause the side street through movement to time to its full split, the only movements with time available for use are the major street left turns and the minor street left turns (if present). Queue spillback for the major street left turns can exceed available storage and spill into the adjacent through movement, creating operational and safety problems. As a result of these disadvantages, this technique is often not desirable.

Use of short cycle lengths

One of the most effective techniques in managing oversaturated conditions is the use of shorter cycle lengths. Shorter cycle lengths allow more frequent servicing of all movements at only a minor expense of additional loss time during the peak time period. This frequent cycling provides more equitable servicing of all movements and allows drivers to visibly experience progress, even if it takes multiple cycles to be served at a given intersection.

Queue management on major street movements

An alternative technique is to selectively store queues on major street movements. Candidates for this treatment frequently include major intersections that are spaced far enough from other signalized intersections to allow the queues to grow without creating upstream intersection impacts. In addition, drivers may be more accepting of congestion at major intersections than at minor intersections.

For network applications (e.g., downtown grids), it is often best to store queues outside the network using key signals to meter traffic into the grid network. While this creates congestion at some intersections, it allows a network of intersections to operate undersaturated, thus enabling traffic to progress through the network.

Use of actuated uncoordinated operation

Removing the cycle length constraint during the oversaturated period can result in efficient allocation of green time, provided gap timers are set appropriately. There are emerging strategies such as lane-by-lane detection and measurement of flow used as opposed to presence for control logic decision-making and operation.

6.8 REFERENCES

• National Traffic Signal Report Card, National Transportation Operations Coalition, 2005.
• Fundamentals of Traffic Engineering (15th ed.), UCB-ITS-CN-01-1, January 2001
• Srinivasa R. Sunkari, Roelof J. Engelbrecht, and Kevin N. Balke, Evaluation Of Advance Coordination Features In Traffic Signal Controllers
• Albert Grover & Associates, “Minimize Delay, Maximize Progression with Protected Permissive Lead/Lag Phasing”, Technical Workshop. Inland Empire Section of ITE, December 14, 1995.
• Frederickson, J. Letter to the Arizona Insurance information association. October 2000.
• Kittelson & Associates, Evaluation of Traffic Signal Displays for Protected/Permissive Left-Turn Control. NCHRP 3-54(2), Final Report, FHWA, U.S. Department of Transportation, Washington, DC, 2003.
• Shelby, S. G., Bullock, D. M., and Gettman, D. Transition Methods in Traffic Signal Control. Transportation Research Record: Journal of the Transportation Research Board, No. 1978, Transportation Research Board of the National Academies, Washington, D.C., 2006, pp. 130-140.
• Zong Z. Tian, Tom Urbanik, Kent Kacir, Mark Vandehey, and Howard Long, “Pedestrian Timing Treatment for Coordinated Signal Systems”, viewed at http://wolfweb.unr.edu/homepage/zongt/Publications_files/PedTimingICTTS.pdf January 17, 2007.
• Signalized Intersections Guide
• Los Angeles Department of Transportation, “Early Methods for Traffic Signal Network Timing” http://www.lacity.org/ladot/TopicsAndTales/TrafficSigTiming6.pdf March 5, 2007.
• Nelson, Eric “Timing Strategies for Managing Oversaturation in Harris County”, Presentation at the TRB Summer Meeting, Woods Hole, MA, July 11, 2006
• Lieberman, Ed, NCHRP 3-38 (4), Transportation Research Board, 1992.
• Parsonson, “Signal Timing Improvement Practices”NCHRP Synthesis 172 National Research Council, Transportation Research Board, Washington, D.C., February 1992
• Webster, F.V., “Traffic Signal Settings,” Road Research Technical Paper No. 30, HMSO, London (1958)
• Teply, Stan “Saturation Flow at Signalized Intersections through a Magnifying Glass,”Proceedings of the Eight International Symposium on Transportation and Traffic Theory University of Toronto Press, 1983.
• Alexiadis, V. K. Jeannotte, A. Chandra, Traffic Analysis Toolbox Volume I: Traffic Analysis Tools Primer, FHWA -HRT-04-038, June 2004.
• Signal Timing Process, Federal Highway Administration, Sabra Wang, December 30, 2003
• Fambro, Sunkari, Sangineni, “Guidelines for Retiming Arterial Networks”, Texas Transportation Institute, February 1993
• Trafficware. Synchro 6 Users Manual, 2004.
• MnDOT Traffic Signal Timing and Coordination Manual, March 2005
• Bullock, D. and A. Sharma, MOST PROJECT Laboratory 6. Actuated Traffic Signal Control, DRAFT, January 2007.
• Transportation Infrastructure, Benefits of Traffic Control Signal Systems Are Not Being Fully Realized. GAO/RCED-94-105, Report to the Chairman, Committee on Energy and Commerce, House of Representatives, United States General Accounting Office, Washington, D.C., March 1994
• 23 Bullo, Darcy and Tom Urbanik, “Addressing Diverse Technologies and Complex User Needs”, http://onlinepubs.trb.org/onlinepubs/millennium/00116.pdf January 17, 2007
• Abbas, Montasir, Darcy Bullock, and Larry Head, “Real-time offset transitioning algorithm for coordinating traffic signals”, Transportation Research Record 1748, Washington D.C. 2001

TRID the TRIS and ITRD database

TRAFFIC SIGNALS IN ASSIGNMENT

A sound theory combining traffic control and route choice would be of use in the design and operation of traffic control systems, and in the design and evaluation of traffic management schemes and major road proposals. Section 2 of this paper briefly considers the difficulties which arise when traffic control and route choice are combined, and describes three attempts to circumvent these difficulties. (a particular policy po arises naturally from part of this work). In section 3 the existing theory is extended to yield a convex optimisation method of calculating (suboptimal) signal settings and link flows which are consistent with the control policy po. Throughout the paper it is supposed that the trip matrix is fixed and that there is a strict limit to the traffic flow along any link; the travel cost tends to + infinity as this limit is approached. (TRRL)

• Find a library where document is available. Order URL: http://worldcat.org/issn/01912615

Pergamon Press, Incorporated

• Publication Date: 1983-4
• Features: References;
• Pagination: p. 155-160
• Transportation Research Part B: Methodological
• Volume: 19B
• Issue Number: 2
• Publisher: Elsevier
• ISSN: 0191-2615
• Serial URL: http://www.sciencedirect.com/science/journal/01912615

Subject/Index Terms

• TRT Terms: Costs ; Decision making ; Itinerary ; Mathematical models ; Matrices (Mathematics) ; Network links ; Optimization ; Policy ; Route choice ; Traffic assignment ; Traffic control ; Traffic flow ; Traffic signal cycle ; Traffic signal timing ; Traffic signals ; Travel
• Uncontrolled Terms: Links ; Optimum ; Selection ; Trip
• ITRD Terms: 224 : Cost ; 576 : Cycle (traffic signals) ; 2248 : Decision process ; 699 : Itinerary ; 698 : Journey ; 6473 : Mathematical model ; 173 : Policy ; 9072 : Selection ; 679 : Traffic assignment ; 654 : Traffic control ; 671 : Traffic flow ; 565 : Traffic signal
• Subject Areas: Finance; Highways; Operations and Traffic Management; Policy;

Filing Info

• Accession Number: 00452141
• Record Type: Publication
• Source Agency: Transport Research Laboratory
• Files: ITRD, TRIS
• Created Date: May 31 1986 12:00AM