EC-RLPA: a dynamic pricing law framework for smart connected vehicles integrating edge computing and reinforcement learning

Zhou, Bei · 2026 · Crossref

DOI: 10.1504/ijict.2026.152070

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the conflict between the efficiency demands of dynamic pricing in smart connected vehicles (SCVs) and strict regulatory compliance requirements, such as those imposed by the EU Digital Services Act and China’s data security regulations. Traditional pricing models often fail to account for real-time traffic fluctuations or embed legal constraints, leading to high violation rates and significant fines for platforms. To resolve this, the authors propose the Edge-Cloud Reinforcement Learning Pricing Architecture (EC-RLPA), a framework that integrates edge computing for low-latency data processing with reinforcement learning for compliant decision-making. The study aims to transform regulatory rules into executable mathematical constraints, ensuring that pricing algorithms are compliant by design rather than subject to post-hoc auditing. The EC-RLPA framework utilizes a three-layer architecture: edge, fog, and cloud. At the edge layer, vehicle trajectory data from the HighD dataset is processed using lightweight convolutional networks, with Laplace noise added for differential privacy to protect sensitive information. The fog layer aggregates these features and fuses them with OpenStreetMap data to generate dynamic geo-fencing boundaries. The cloud layer employs a multi-agent proximal policy optimization (MAPPO) algorithm, trained on New York City taxi demand data, to generate pricing strategies. Crucially, the reward function incorporates penalty terms for exceeding price caps and geographic discrimination, effectively encoding legal constraints into the learning process. The system was validated using a simulation environment combining HighD trajectory data and New York taxi records, benchmarked against traditional Q-learning, non-compliant PPO, and edge-greedy algorithms. Experimental results demonstrate that EC-RLPA significantly outperforms baseline models in both compliance and efficiency. The framework achieved a regulatory compliance rate of 92.3%, a substantial improvement over the 65.2% rate of traditional Q-learning methods. Decision latency was reduced to 142 milliseconds, representing an 82.5% decrease compared to traditional solutions, while raw data transfer was reduced by 80% due to edge processing. Economically, the system generated an average revenue of $283 per hour, outperforming non-compliant models despite the constraints. Ablation studies confirmed that removing compliance penalties caused the compliance rate to drop to 63.5%, while static geo-fencing resulted in a 17.2% loss in peak revenue. The study also identified a Pareto optimal point for the compliance penalty coefficient, balancing regulatory adherence with economic gain. The significance of this research lies in its demonstration of a "regulation-technology-data" coupling framework, providing a technical pathway for embedding legal compliance directly into AI-driven pricing systems. By converting textual regulations into real-time price caps and fairness constraints, EC-RLPA offers policymakers a tool for dynamic traffic management and provides platforms with a mechanism to avoid fines while maintaining profitability. The findings suggest that intelligent transportation systems can achieve both high responsiveness and strict regulatory adherence, offering a model for other industries facing similar tensions between algorithmic efficiency and legal compliance. Future work is suggested to address cross-platform competition and external factors like extreme weather.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• dms regulation