C-EQ-ALINEA: Distributed, Coordinated, and Equitable Ramp Metering Strategy for Sustainable Freeway Operations

AI summary

Overview

Research area: Traffic management and control — specifically equitable (fairness-aware) ramp metering on freeway networks, with implications for AI ethics/safety in public-infrastructure automation (arXiv category cs.CY). Technical level: Intermediate. The control equations are simple, but interpreting the fairness metrics and the microsimulation setup requires some familiarity with traffic engineering and control theory. Scope: The paper proposes C-EQ-ALINEA, a decentralized, coordinated, equity-aware extension of the classical ALINEA ramp-metering controller, and benchmarks it against ALINEA and METALINE in a calibrated 24-hour SUMO microsimulation of Amsterdam's A10 ring road.

What This Paper Is About

Ramp metering (traffic signals on freeway on-ramps) improves overall freeway efficiency, but conventional controllers such as ALINEA treat each ramp in isolation, which can produce highly uneven waiting times across ramps and provoke non-compliance, public opposition, and even political revocation of systems. Existing fairness-aware ramp-metering methods address this but typically depend on centralized optimization, detailed traffic models, or data-intensive learning, making them hard to deploy in networks that already run legacy ALINEA systems. The goal of C-EQ-ALINEA is to achieve meaningful fairness gains using only minimal, lightweight extensions to the widely deployed ALINEA controller.

Key Contributions

1. A decentralized, coordinated, equity-aware controller. C-EQ-ALINEA adds a coordination term to the classical ALINEA P-control law, based on flow information exchanged with neighbouring ramps, with no central optimizer and no additional infrastructure.
2. Distance-based neighbour weighting with two normalization variants. Neighbour influence is weighted by distance using u_j = max(0, 1 − d_nj / L_max) and then normalized to w_j = u_j / Σ(u_j); the paper explores both global distance normalization (L_max = maximum distance across all consecutive ramp pairs in the network) and local distance normalization (L_max = maximum distance within the neighbourhood), for neighbourhood sizes m = 1, 2, and 3.
3. Evaluation against four distinct fairness notions. The method is assessed under Harsanyian (average delay), Egalitarian (Gini coefficient of delays), Rawlsian (maximum delay), and Aristotelian (demand-weighted average delay per on-ramp) definitions of fairness, rather than a single equity metric.
4. A reproducible, open-source implementation released at https://github.com/DerKevinRiehl/c_eq_alinea.

Main Findings

• Efficiency is preserved or improved. Against the uncontrolled scenario (total delay 3591.2 h; average speed 47.0 km/h; average delay 282 s/veh), ALINEA reduced total delay by approximately 37.9% to 2230.1 h and raised average speed to 58.7 km/h. METALINE reduced total delay to 2078.7 h (−42.1%) with average speed 59.6 km/h.
• Best configuration: global normalization with m = 3. C-EQ-ALINEA reduced total delay to 1478.8 h (approximately 58.8% below the uncontrolled case), raised average speed to 66.7 km/h, and reduced average delay to 126 s/veh (2.1 min/veh), with total travel time falling from 6717.1 h to 4357.5 h. This configuration exceeded METALINE on several efficiency indicators.
• Local normalization is more sensitive to neighbourhood size. With local distance normalization at m = 3, total delay fell to 1610.4 h (−55.2%) with average speed 65.3 km/h, but the m = 1 case was notably weaker, with a higher total travel time of 5677.5 h and smaller delay reductions.
• Throughput is not sacrificed. Arrival rates stayed in the 99.7–99.9% range across all controllers, and the best C-EQ-ALINEA configuration kept departed and arrived vehicle counts above 40,000.
• Harsanyian fairness (average delay). The average delay across all ramps was 282.8 s in the uncontrolled scenario; ALINEA reduced it by 103.7 s (36.7%) and METALINE by 97.2 s (34.4%); C-EQ-ALINEA with global normalization and m = 3 reduced it by 157.4 s (55.7%).
• Egalitarian fairness (Gini). All controllers lowered the Gini coefficient below the uncontrolled 0.2635, and C-EQ-ALINEA achieved the most equitable distribution, a 28.2% improvement over the uncontrolled baseline.
• Rawlsian fairness (maximum delay). The worst-case delay of 550.4 s under no control was consistently reduced by ramp metering, with C-EQ-ALINEA yielding the largest improvement (52.99%).
• Aristotelian fairness (demand-weighted average). Delay patterns aligned more closely with each on-ramp's demand, with C-EQ-ALINEA achieving the strongest alignment (weighted average delay of 118.2 s versus 289.4 s uncontrolled).
• Trip-distance stratification. In the uncontrolled scenario, relative delays (delay per kilometre travelled) were less evenly distributed, with longer trips typically facing lower relative delays; controlled scenarios produced more equitable relative delays, with METALINE most uniform and C-EQ-ALINEA (global normalization, m = 3) ranking second.
• Additional indicators matched. Queuing metrics (maximum, average, and cumulative queue length, merging rates), metering statistics (phase change frequency and average metering rate), and user metrics (maximum delay, waiting time) showed similar fairness trends to the per-ramp delay results.

Methodology in Plain English

The highway is modelled as a directed graph in which nodes are on-ramps and off-ramps, and each on-ramp has a signal controlling inflow. Detectors spaced roughly 100–200 m downstream supply occupancy measurements, which stand in for traffic density. Standard ALINEA sets each ramp's metering rate using a simple feedback rule: take the previous rate and adjust it by a gain factor K multiplied by the gap between a desired occupancy and the measured occupancy. C-EQ-ALINEA keeps that rule and adds one extra term: a coordination gain K_c multiplied by the difference between a distance-weighted average of neighbouring ramps' flow rates and the ramp's own flow rate. Closer ramps get more weight; weights decay linearly with distance and are normalized to sum to one. Two ways of setting the maximum distance scale were tested (global network-wide, and local to the neighbourhood), along with neighbourhood sizes of 1, 2, and 3 ramps on each side.

For evaluation, the authors used a calibrated 24-hour microsimulation of Amsterdam's A10 ring road — a 32 km beltway with 20 on-ramps and 22 off-ramps — in the open-source SUMO environment, with demand calibrated using data from Nationaal Dataportaal Wegverkeer and a mixed multimodal fleet composition derived from Centraal Bureau voor de Statistiek statistics. Each simulation included a 3000 s warm-up period, and results are means across 10 simulations with different random seeds, with standard deviations in brackets; the highly congested period from 10:00 AM to 06:00 PM was used for evaluation. All controllers and their parameters were tuned via an extensive grid-search with multiple random seeds per parameter combination to maximize network throughput. Fairness was then assessed by comparing per-ramp average delays across ramps (excluding ramps with negligible demand) under the four fairness notions.

Why This Matters

The paper argues that fairness is not merely a social objective but a prerequisite for sustainable deployment of ramp metering: systems perceived as unfair face signal violations, queue bypassing, and public and political opposition. The contribution is that such fairness can be improved through a minimal algorithmic extension to a controller that is already widely deployed, rather than through new centralized infrastructure or data-hungry learning systems.

Real-world applications:

• Transportation agencies operating legacy ALINEA systems, which would need only communication between ramp controllers and a firmware/software update — no heavy new hardware or centralized optimizers, according to the paper.
• Urban freeway networks with many closely spaced ramps, where distance-weighted coordination can smooth delay distribution across corridors.
• Public acceptance and compliance programmes, where documented fairness of delay distribution can counter the backlash that has led to deactivation or revocation of metering systems.
• Sustainability and equity reporting in transport planning, where Harsanyian, Egalitarian, Rawlsian, and Aristotelian metrics provide a multi-dimensional account of who bears delay.

Industry relevance: the approach targets agencies and operators seeking congestion relief and improved network equity without extensive computational infrastructure or complex optimization solvers, while preserving ALINEA's robustness to measurement noise and model uncertainty.

Future Directions

• Measure perceived fairness and user acceptance directly. The authors state these were not measured and remain an important direction, alongside extending evaluation beyond delay to emissions and safety.
• Test transferability. The evaluation covers a single, albeit complex, urban freeway network; further study is needed on different network structures, demand patterns, incident scenarios, ramp spacing, interchanges, managed lanes, and weaving sections.
• Handle physical and infrastructure constraints. Currently C-EQ-ALINEA does not handle limited ramp storage (risking queue spillback onto surface streets) and may suffer from incomplete detector coverage and aging infrastructure.
• Add adaptability and resilience. Suggested extensions include enhancing resilience under degraded conditions such as sensor failures and communication delays, incorporating short-term traffic prediction and deep learning for proactive control, applying online machine learning to auto-tune coordination gains and weight parameters from historical and real-time data, and conducting controlled field trials.

Target Audience

This paper is most useful to traffic engineers and freeway operations practitioners who manage existing ramp-metering infrastructure; transportation researchers working on coordinated and fairness-aware control, including those comparing ALINEA, METALINE, HERO, and SWARM-style approaches; and policy-oriented readers interested in the ethics and social acceptance of automated traffic management, given the paper's framing of fairness as a precondition for sustainable deployment. Readers with an interest in AI safety and ethics will find it relevant as a case study of embedding distributional fairness constraints into a lightweight, interpretable, already-deployed control algorithm rather than a centralized or learned system.

Authors’ abstract

Ramp metering is a widely deployed traffic management strategy for improving freeway efficiency, yet conventional approaches often lead to highly uneven delay distributions across on-ramps, undermining user acceptance and long-term sustainability. While existing fairness-aware ramp metering methods can mitigate such disparities, they typically rely on centralized optimization, detailed traffic models, or data-intensive learning frameworks, limiting their real-world applicability, particularly in networks operating legacy ALINEA-based systems. This paper proposes C-EQ-ALINEA, a decentralized, coordinated, and equity-aware extension of the classical ALINEA feedback controller. The approach introduces lightweight information exchange among neighbouring ramps, enabling local coordination that balances congestion impacts without centralized control, additional infrastructure, or complex optimization. C-EQ-ALINEA preserves the simplicity and robustness of ALINEA while explicitly addressing multiple notions of fairness, including Harsanyian, Egalitarian, Rawlsian, and Aristotelian perspectives. The method is evaluated in a calibrated 24-hour microsimulation of Amsterdam's A10 ring road using SUMO. Results demonstrate that C-EQ-ALINEA substantially improves the equity of delay distributions across ramps and users, while maintaining (in several configurations surpassing) the efficiency of established coordinated strategies such as METALINE. These findings indicate that meaningful fairness gains can be achieved through minimal algorithmic extensions to widely deployed controllers, offering a practical and scalable pathway toward sustainable and socially acceptable freeway operations. Open source implementation available on GitHub.