💻 computer science

FairSCOSCA: Fairness At Arterial Signals -- Just Around The Corner

This paper introduces FairSCOSCA, a fairness-enhancing extension to existing traffic signal systems like SCOOTS and SCATS that utilizes green phase optimization and early termination to significantly reduce waiting time inequality and discrimination between road users without compromising overall traffic efficiency, as validated through a microsimulation case study in Esslingen am Neckar.

Original authors: Kevin Riehl, Justin Weiss, Anastasios Kouvelas, Michail A. Makridis

Published 2026-04-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Kevin Riehl, Justin Weiss, Anastasios Kouvelas, Michail A. Makridis

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a busy city street as a giant, chaotic dinner party. The "arterial road" is the main dining table where the most important guests (heavy traffic) sit, while the "feeder roads" are the smaller side tables where fewer guests arrive.

For decades, the traffic lights controlling this party have been run by a very strict, efficiency-obsessed host named SCOOTS/SCATS. This host's only rule is: "Keep the main table moving as fast as possible."

The Problem:
Because the host only cares about the main table, the guests at the side tables often sit there for hours, waiting for a crumb of attention. They get so frustrated that they might start breaking the rules (running red lights) or just give up. While the main road flows smoothly, the side roads are a nightmare of endless waiting. This is unfair.

The Solution: FairSCOSCA
The authors of this paper, Kevin and his team, invented a new set of rules for the host called FairSCOSCA. They didn't fire the host or build a new kitchen; they just gave the host two new "fairness tools" to tweak how they serve the guests.

Here is how it works, using simple analogies:

1. The "Wait-Time Watch" (FairSCOSCA 1)

The Old Way: The host looks at the main table and sees it's full, so they keep the green light on for the main road. They ignore the side table, even if a single person has been waiting there for 10 minutes.
The New Way: The host now keeps a mental tally of how long everyone has been waiting.

If the main road is busy, the host gives them the green light.
BUT, if someone on the side road has been waiting too long, the host says, "Okay, main road, you've had your turn. Side road, you've waited long enough; it's your turn now."
The Result: The host balances the load. The main road still moves fast, but the side road guests don't get stuck in an endless purgatory.

2. The "Early Exit" Button (FairSCOSCA 2)

The Old Way: Once the host decides to give the green light to the main road, they stick to it until the timer runs out, even if the main road is suddenly empty and a car on the side road is screaming, "I'm here! Let me through!"
The New Way: The host watches the green light closely. If they see a car on the side road arrive just as the main road's light turns green (and the main road is actually empty), the host hits an "Early Exit" button.

They cut the main road's green light short by a few seconds.
They immediately give those few seconds to the waiting side road.
The Safety Net: To make sure the main road doesn't lose out in the long run, the host promises to give the main road those few seconds back in the next round.
The Result: No one gets stuck waiting for a light that isn't actually being used. It's like a waiter who realizes a table is empty and immediately serves the hungry guest at the next table, then fixes the order later.

The Big Test: The "Esslingen" Simulation

The team tested these new rules in a computer simulation of a real street in Germany (Esslingen). They compared the new "Fair" host against:

The Old Host (SCOOTS/SCATS): Good at speed, bad at fairness.
The "Max-Pressure" Host: A robot that only looks at how many cars are in a line right now. It tends to ignore side roads completely.
The "Fixed-Cycle" Host: A robot that changes lights on a strict timer, regardless of traffic. (The worst performer).

The Results: A Win-Win

The study found something surprising: You don't have to choose between speed and fairness.

Efficiency: The new "Fair" system was actually faster or just as fast as the old system. The main road didn't slow down.
Fairness: The side roads saw massive improvements. The "unfair" long waits disappeared.
The "Gini" Score: In economics, the Gini coefficient measures inequality (like wealth gap). The new system lowered this score for traffic, meaning the waiting time was distributed much more equally among all drivers, whether they were on the main road or the side road.

Why This Matters

Traffic lights are usually programmed by expensive, proprietary software that companies guard like secrets. This paper is like handing the public a free, open-source manual that says, "Hey, you can tweak these existing systems to be fairer without buying new hardware."

In a nutshell:
Think of traffic lights as a turnstile at a subway station. The old system only let the biggest crowds through, leaving the lone traveler waiting forever. FairSCOSCA is the new rule that says, "We'll let the big crowd through, but if that lone traveler has been waiting too long, we'll let them squeeze in, and then we'll make sure the big crowd gets their extra turn later."

It turns a system that was "efficient but cruel" into one that is "efficient and kind."

1. Problem Statement

Traffic signal control in arterial networks is critical for managing urban congestion. While systems like SCOOTS (Split, Cycle, and Offset Optimization Technique) and SCATS (Sydney Coordinated Adaptive Traffic System) are globally deployed and highly efficient, they prioritize traffic efficiency (minimizing total delay) over fairness.

The Issue: Efficiency-focused logic often leads to "unfair" outcomes, such as excessive waiting times for vehicles on feeder roads (side streets) while prioritizing busy arterial roads. This creates horizontal discrimination and can lead to public dissatisfaction, red-light running, and safety issues.
The Gap: Existing research on fair traffic control often relies on complex machine learning models with limited real-world applicability or lacks systematic benchmarking against established industrial controllers. There is a need for practical, transparent modifications to existing systems that improve equity without sacrificing efficiency.

2. Methodology

The authors propose FairSCOSCA, a fairness-enhancing extension to the standard SCOOTS/SCATS logic (referred to as SCOSCA). The methodology involves two novel design adaptations integrated into the existing three-optimizer framework (Green Phase, Cycle Length, and Offset Optimization).

A. Baseline: SCOSCA

The standard system uses:

Green Phase Optimization: Allocates green time based on the Degree of Saturation (DS).
Cycle Length Optimization: Adjusts cycle length to keep DS near 0.9.
Offset Optimization: Creates "green waves" based on congestion density.

B. Proposed FairSCOSCA Adaptations

The authors introduce two specific features to inject fairness into the logic:

1. FairSCOSCA 1: Cumulative Waiting Time Integration

Mechanism: Modifies the Green Phase Optimization. Instead of allocating green time solely based on current traffic load (DS), it incorporates the cumulative waiting time of vehicles on opposing phases.
Logic: If vehicles on a side road have been waiting for a long time, a penalty term ( $P(k)$ ) is applied. This term, defined exponentially based on the maximum cumulative wait, reduces the green time allocation for the dominant phase (arterial) to compensate.
Fairness Goal: Addresses Egalitarian (equal distribution) and Rawlsian (improving the worst-off) fairness by preventing extreme delays for feeder road users.

2. FairSCOSCA 2: Early Phase Termination

Mechanism: A secondary, real-time intervention. If a vehicle arrives at a red signal during the early portion of an active green phase, and the remaining green time exceeds a threshold ($TTG$), the controller triggers a pre-emption.
Logic: The current green phase is shortened by a fixed duration ($TEG$), and that time is immediately transferred to the waiting phase. To prevent instability, a compensation mechanism adds the lost time back to the shortened phase in the next cycle.
Fairness Goal: Reduces extreme individual delays for vehicles that just missed a green light, aligning with Rawlsian and Egalitarian principles.

3. Key Contributions

Practical Implementation: Unlike many theoretical fair-control papers, this work proposes simple, transparent modifications to the world's most widely used traffic control systems (SCOOTS/SCATS), making real-world deployment feasible.
Multi-Dimensional Fairness Evaluation: The study evaluates performance across four distinct normative fairness definitions:
- Egalitarian: Measured by Gini coefficient of delays.
- Rawlsian: Measured by Maximum Delay (worst-case scenario).
- Utilitarian: Measured by Total Travel Time.
- Harsanyian: Measured by Average Delay.
Comprehensive Benchmarking: The proposed method is rigorously benchmarked against:
- Fixed-Cycle (pre-timed).
- Max-Pressure (queue-length based).
- Standard SCOSCA (SCOOTS/SCATS).
Horizontal Equity Analysis: Specifically analyzes discrimination between Arterial (main road) and Feeder (side road) users, a critical issue in arterial networks.

4. Results

The methods were evaluated using a calibrated microsimulation of the Schorndorfer Strasse arterial network in Esslingen am Neckar, Germany (5 intersections, 22 traffic lights, mixed traffic including buses).

Efficiency

FairSCOSCA 1 outperformed the standard SCOSCA in efficiency metrics, achieving a 2.39% increase in flow, 6.35% increase in speed, and 2.77% increase in throughput.
FairSCOSCA 2 maintained efficiency levels comparable to standard SCOSCA, with no significant degradation.
Both proposed methods significantly outperformed Fixed-Cycle and Max-Pressure controllers.

Fairness & Equity

Overall Fairness: Both FairSCOSCA variants significantly reduced the Gini coefficient (inequality) and Maximum Delay compared to SCOSCA, Max-Pressure, and Fixed-Cycle.
Horizontal Equity (Arterial vs. Feeder):
- Standard SCOSCA showed discrimination against feeder roads (higher delays for side streets).
- FairSCOSCA 1 successfully reduced this discrimination, lowering delays for feeder road users while slightly improving arterial performance.
- FairSCOSCA 2 improved feeder delays but resulted in slightly higher delays for arterial users compared to the baseline, though still better than Max-Pressure.
Statistical Significance: Improvements in fairness metrics (Gini, Max Delay) were statistically significant ( $p < 0.01$ ) compared to the baseline.

5. Significance and Conclusion

Bridging Theory and Practice: The paper demonstrates that fairness can be integrated into legacy, industrial-grade traffic systems without requiring a complete overhaul or complex AI models.
No Efficiency Trade-off: Contrary to the common assumption that fairness comes at the cost of efficiency, FairSCOSCA 1 actually improved overall network efficiency while enhancing equity.
Public Acceptance: By addressing the "unfairness" of long waits on side streets, the system is more likely to gain public trust and compliance, which is crucial for safety and the success of future smart city initiatives.
Open Source: The authors provide an open-source implementation on GitHub, facilitating reproducibility and further research.

Limitations & Future Work:
The study relies on simulation. Future work should validate these findings with real-world floating car data, assess robustness during incidents/sensor failures, and explore multi-objective optimization that explicitly balances efficiency and fairness. Additionally, the model currently focuses on delay-based metrics and does not yet account for socio-demographic factors or specific mode prioritization (e.g., public transit) beyond the simulation parameters.