Traffic light synchronisation algorithms are the core computational logic that determines when each signal in a network changes phase, and in what sequence. Rather than treating each intersection as an isolated unit, synchronisation algorithms coordinate signals across a corridor or zone so that vehicles encounter a progression of green lights, queues are kept manageable, and pedestrian safety is preserved throughout. The design and tuning of these algorithms is one of the most consequential decisions in traffic signal engineering, directly affecting throughput, emissions, incident rates, and the experience of every road user on the network.
What synchronisation actually means
Synchronisation does not simply mean all lights change at the same time. It means the timing relationships between signals are deliberately set so that a platoon of vehicles departing one intersection arrives at the next one during its green phase. This concept, commonly called a "green wave" or bandwidth optimisation, underpins the oldest and most widely deployed class of synchronisation algorithm. The mathematics behind it relates signal offset (the delay between phase starts at adjacent intersections) to the distance between those intersections and the design speed of the corridor.
In a simple two-way arterial, the optimal offset for one direction of travel is rarely optimal for the opposing direction. Algorithms must therefore find a compromise that maximises the combined bandwidth available to both directions, weighted by traffic volumes. This trade-off becomes significantly more complex at network scale, where multiple crossing arterials, turning movements, and pedestrian phases all compete for green time.
Fixed-time vs adaptive approaches
The most fundamental distinction in synchronisation algorithm design is between fixed-time plans and adaptive control. Fixed-time plans pre-calculate signal timing based on historical traffic data and store a library of timing plans, each assigned to a time-of-day period. The SCATS (Sydney Coordinated Adaptive Traffic System) and SCOOT (Split Cycle Offset Optimisation Technique) platforms, both used extensively across Australian road networks, represent different points on the spectrum between these two approaches.
SCATS operates by selecting from pre-optimised timing plan libraries and making real-time adjustments to cycle length, split, and offset within defined bounds. SCOOT, developed in the UK and deployed in a number of Australian corridors, uses a rolling online optimisation model that continuously adjusts timing in response to detector data. Both systems rely on the same underlying mathematical framework: minimising a cost function that typically combines measures of delay, stops, and queue length across the network. Engineers choosing between platforms must weigh the complexity of detector infrastructure, the variability of local demand patterns, and the ongoing data requirements for keeping timing plans current.
Modern adaptive signal control has pushed this further. As explored in detail in the context of how AI-driven traffic signal control works in practice, machine learning models can now forecast demand a few cycles ahead and pre-position phase splits to absorb predicted peaks, rather than reacting to congestion after it has formed.
Key algorithmic components
Regardless of the specific platform, most synchronisation algorithms share a common set of computational components:
- Cycle length selection: The total time allocated to one complete sequence of phases at a reference intersection. Longer cycles reduce the proportion of time lost to phase transitions but increase maximum waiting time for any given movement. Typical values in Australian urban networks range from 60 to 120 seconds.
- Split calculation: The allocation of green time within the cycle to each signal phase. Splits are usually calculated using Webster's formula or a derivative, which minimises total intersection delay as a function of saturation flow and arrival rate for each approach.
- Offset optimisation: The phase relationship between adjacent intersections. Offset is the variable that creates the green wave. Most network-level optimisation tools iterate over a large solution space, evaluating thousands of offset combinations to find the configuration that maximises aggregate bandwidth or minimises network-wide delay.
- Detector integration: Inductive loop detectors or, increasingly, video and radar sensors provide real-time counts and occupancy data. The algorithm uses this feed to verify that conditions match the timing plan in use and to trigger plan switches or adaptive adjustments.
Handling special demands: pedestrians, cyclists, and emergency vehicles
A synchronisation algorithm operating purely on vehicle throughput objectives will systematically under-serve pedestrians and cyclists. Australian standards and Austroads guidelines require that pedestrian crossing times are included as minimum constraints in any timing plan, which means the algorithm must satisfy pedestrian walk and clearance minimums before allocating remaining green time to vehicle phases. In high-pedestrian environments such as CBDs, entertainment precincts, and school zones, these constraints can become binding, effectively setting the floor on cycle length regardless of vehicle demand.
Emergency vehicle preemption (EVP) is a separate but related control layer. When an emergency vehicle approaches an intersection, the preemption system overrides the synchronisation plan, clears a path for the vehicle, and then executes a recovery routine to restore coordinated operation as quickly as possible. The design of that recovery routine matters: a poorly implemented recovery can disrupt the green wave on an arterial for several cycles after a single preemption event. Well-engineered systems return to the synchronised plan via a transition phase that minimises the downstream impact on platoon progression.
The interaction between pedestrian signal demands, emergency vehicle preemption systems, and the base synchronisation plan is one of the more technically demanding aspects of modern signal controller design, requiring careful priority logic to avoid conflicts between competing override requests.
Network-level optimisation and smart city integration
At the network scale, synchronisation shifts from corridor-level offset calculations to area-wide traffic assignment. Tools such as TRANSYT and various proprietary network optimisers model the entire signal network as a system of interconnected queues, seeking a timing configuration that minimises a network cost function across all links simultaneously. The solution space for even a moderate-sized urban network is astronomically large, so practical optimisers use heuristic search methods including genetic algorithms, simulated annealing, and gradient descent variants to find high-quality solutions within acceptable computation times.
Integration with broader smart city infrastructure adds further capability and complexity. When signal controllers receive data feeds from connected vehicles, public transport priority systems, parking sensors, and pedestrian counters, the synchronisation algorithm can incorporate demand information that would otherwise be invisible to roadside detectors alone. This is the operational context behind strategies for AI traffic light optimisation, where real-time data from multiple sources feeds a predictive model that continuously refines signal timing across the network.
For infrastructure professionals working on new signal deployments or retiming projects, the algorithm selection and parameterisation process should begin at concept design stage. The choice of detection technology, controller firmware capabilities, and communications architecture all constrain which algorithms can be practically implemented and maintained over the asset's operating life. Getting that alignment right at the outset is what separates a signal network that performs as designed from one that drifts into suboptimal timing plans within a few years of commissioning.
Commissioning, validation, and ongoing maintenance
An algorithm that performs well in simulation must be validated against real-world behaviour before it is accepted as the operational timing plan. Commissioning typically involves field observation of queue lengths, travel time runs along the corridor, and comparison against the modelled performance predictions. Discrepancies often point to incorrect saturation flow assumptions, detector faults, or geometric constraints that were not fully captured in the model.
Ongoing maintenance of timing plans is frequently under-resourced relative to its impact. Traffic demand patterns shift as land use changes, new developments open, and travel behaviour evolves. A timing plan calibrated on data that is three or more years old will progressively diverge from optimal, and the degradation in network performance can be substantial before it is noticed. Best practice involves systematic retiming on a two-to-three-year cycle, supported by continuous detector data collection and periodic travel time audits to track performance between major retiming exercises.