Explainable Federated Reinforcement Learning Framework for Intelligent and Ethical Smart City Systems

DOI

10.2991/978-94-6239-693-7_78How to use a DOI?

Keywords

Explainable Artificial Intelligence; Federated Learning; Reinforcement Learning; Smart Cities; Ethical AI; Edge Intelligence

Abstract

To support complex city systems, including transportation, energy distribution, citizen security, and environmental surveillance, smart city infrastructure is increasingly dependent on artificial intelligence. Although reinforcement learning can be used to make adaptive judgments in such settings, conventional centralized training designs raise privacy, regulatory, and ethical concerns. Also, the majority of reinforcement learning models are black boxes, hindering transparency and trust among the general public. This paper introduces an Explainable Federated Reinforcement Learning system that can be used to build intelligent, ethical, and privacy-conscious smart city systems through decentralized learning and interpretable decision-making mechanisms. The adopted model is a fusion of federated learning and reinforcement learning that enables a group of distributed agents in the city to train policies together without sharing raw data. Each agent has an explainability module that offers interpretable insights of decision behavior to allow accountability and governance oversight. This framework is tested through a massive simulation of a smart city that handles traffic and energy control across several distributed nodes and 1 million sensor entries. Cumulative reward, convergence speed, stability, fairness, and explainability fidelity are used to measure performance. The experimental evidence shows that the suggested solution yields the cumulative reward improvement of about 18% in comparison to the centralized reinforcement learning and does not involve exposure to raw data at all. The equity of resource distribution in urban areas increases by more than 14%, and the explainability module has a fidelity score of more than 0.9, indicating high consistency between the model's decisions and the explanations generated. The system is observed to be converging faster and to be robust in dynamic conditions. The findings indicate that explainable federated reinforcement learning provides a scalable, responsible solution for deploying artificial intelligence in the governance of a smart city. This framework incorporates balanced efficiency, transparency, and privacy, which enables the adoption of AI in socially sensitive urban settings in a manner that is trusted.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Roohee Khan
AU  - Anjali Goswami
PY  - 2026
DA  - 2026/06/16
TI  - Explainable Federated Reinforcement Learning Framework for Intelligent and Ethical Smart City Systems
BT  - Proceedings of the International Conference on Intelligent Systems for a Sustainable Future (ISSF 2026)
PB  - Atlantis Press
SP  - 793
EP  - 802
SN  - 2589-4919
UR  - https://doi.org/10.2991/978-94-6239-693-7_78
DO  - 10.2991/978-94-6239-693-7_78
ID  - Khan2026
ER  -