Comparing Optimal and Adaptive EV Charging in Smart Cities:  MILP vs. Reinforcement Learning

Alberto Bazán-Guillén; Pablo Andres Barbecho Bautista; Mónica Aguilar Igartua; Francesca Cuomo

doi:10.1109/MSWiM67937.2025.11308997

Comparing Optimal and Adaptive EV Charging in Smart Cities: MILP vs. Reinforcement Learning

Alberto Bazán-Guillén
, Pablo Andres Barbecho Bautista
, Mónica Aguilar Igartua
, Francesca Cuomo

Departamento de Ingeniería Eléctrica, Electrónica y Telecomunicaciones

Producción científica: Contribución a una conferencia › Artículo › revisión exhaustiva

Resumen

The coordinated scheduling of electric vehicle (EV)
charging is a critical challenge for smart cities, particularly in
high-density infrastructure such as Mobility Hubs (MHs). This
paper evaluates and compares two prominent approaches to the
EV Charging Scheduling Problem (CSP): Mixed-Integer Linear
Programming (MILP) and Reinforcement Learning (RL). We
formulate a shared problem framework and apply both strategies
under two structured scenarios: a small-scale deterministic
benchmark and a medium-scale, realistic deployment with higher
heterogeneity. Results show that MILP achieves optimal cost and
SoC compliance in tractable cases but struggles with scalability.
RL, based on Proximal Policy Optimization (PPO), achieves nearoptimal
performance while scaling to 100 EVs with minimal
computation time. Despite occasional SoC deviations, the RL
agent exhibits robust and adaptive behavior under dynamic
conditions. This study offers actionable insights for selecting
and deploying EV scheduling strategies in real-world urban
environments.

Idioma original	Español
DOI	https://doi.org/10.1109/MSWiM67937.2025.11308997
Estado	Publicada - 31 oct. 2025
Evento	21st Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN 2025). - Universidad Politécnica de Cataluña, Barcelona, Espana Duración: 27 oct. 2025 → 31 oct. 2025 Número de conferencia: 21 http://pewasun.upc.edu/PEWASUN2025/

Conferencia

Conferencia	21st Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN 2025).
Título abreviado	PE-WASUN 2025
País/Territorio	Espana
Ciudad	Barcelona
Período	27/10/25 → 31/10/25
Dirección de internet	http://pewasun.upc.edu/PEWASUN2025/

ODS de las Naciones Unidas

Este resultado contribuye a los siguientes Objetivos de Desarrollo Sostenible

ODS 7: Energía asequible y no contaminante
ODS 11: Ciudades y comunidades sostenibles

Acceder al documento

10.1109/MSWiM67937.2025.11308997

Otros archivos y enlaces

http://pewasun.upc.edu/PEWASUN2025/program.html

Citar esto

Bazán-Guillén, A., Barbecho Bautista, P. A., Aguilar Igartua, M., & Cuomo, F. (2025). Comparing Optimal and Adaptive EV Charging in Smart Cities: MILP vs. Reinforcement Learning. Papel presentado en 21st Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN 2025)., Barcelona, Espana. https://doi.org/10.1109/MSWiM67937.2025.11308997

@conference{8467c6d8bbb149b4884fdf8bbd3940f5,

title = "Comparing Optimal and Adaptive EV Charging in Smart Cities: MILP vs. Reinforcement Learning",

abstract = "The coordinated scheduling of electric vehicle (EV) charging is a critical challenge for smart cities, particularly in high-density infrastructure such as Mobility Hubs (MHs). This paper evaluates and compares two prominent approaches to the EV Charging Scheduling Problem (CSP): Mixed-Integer Linear Programming (MILP) and Reinforcement Learning (RL). We formulate a shared problem framework and apply both strategies under two structured scenarios: a small-scale deterministic benchmark and a medium-scale, realistic deployment with higher heterogeneity. Results show that MILP achieves optimal cost and SoC compliance in tractable cases but struggles with scalability. RL, based on Proximal Policy Optimization (PPO), achieves nearoptimal performance while scaling to 100 EVs with minimal computation time. Despite occasional SoC deviations, the RL agent exhibits robust and adaptive behavior under dynamic conditions. This study offers actionable insights for selecting and deploying EV scheduling strategies in real-world urban environments.",

author = "Alberto Baz{\'a}n-Guill{\'e}n and \{Barbecho Bautista\}, \{Pablo Andres\} and \{Aguilar Igartua\}, M{\'o}nica and Francesca Cuomo",

year = "2025",

month = oct,

day = "31",

doi = "10.1109/MSWiM67937.2025.11308997",

language = "Espa{\~n}ol",

note = " 21st Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN 2025). ; Conference date: 27-10-2025 Through 31-10-2025",

url = "http://pewasun.upc.edu/PEWASUN2025/",

}

Bazán-Guillén, A, Barbecho Bautista, PA, Aguilar Igartua, M & Cuomo, F 2025, 'Comparing Optimal and Adaptive EV Charging in Smart Cities: MILP vs. Reinforcement Learning', Papel presentado en 21st Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN 2025)., Barcelona, Espana, 27/10/25 - 31/10/25. https://doi.org/10.1109/MSWiM67937.2025.11308997

Comparing Optimal and Adaptive EV Charging in Smart Cities: MILP vs. Reinforcement Learning. / Bazán-Guillén, Alberto; Barbecho Bautista, Pablo Andres; Aguilar Igartua, Mónica et al.
2025. Papel presentado en 21st Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN 2025)., Barcelona, Espana.

Producción científica: Contribución a una conferencia › Artículo › revisión exhaustiva

TY - CONF

T1 - Comparing Optimal and Adaptive EV Charging in Smart Cities

T2 - 21st Symposium on Performance Evaluation of Wireless Ad Hoc, Sensor, and Ubiquitous Networks (PE-WASUN 2025).

AU - Bazán-Guillén, Alberto

AU - Barbecho Bautista, Pablo Andres

AU - Aguilar Igartua, Mónica

AU - Cuomo, Francesca

N1 - Conference code: 21

PY - 2025/10/31

Y1 - 2025/10/31

N2 - The coordinated scheduling of electric vehicle (EV) charging is a critical challenge for smart cities, particularly in high-density infrastructure such as Mobility Hubs (MHs). This paper evaluates and compares two prominent approaches to the EV Charging Scheduling Problem (CSP): Mixed-Integer Linear Programming (MILP) and Reinforcement Learning (RL). We formulate a shared problem framework and apply both strategies under two structured scenarios: a small-scale deterministic benchmark and a medium-scale, realistic deployment with higher heterogeneity. Results show that MILP achieves optimal cost and SoC compliance in tractable cases but struggles with scalability. RL, based on Proximal Policy Optimization (PPO), achieves nearoptimal performance while scaling to 100 EVs with minimal computation time. Despite occasional SoC deviations, the RL agent exhibits robust and adaptive behavior under dynamic conditions. This study offers actionable insights for selecting and deploying EV scheduling strategies in real-world urban environments.

AB - The coordinated scheduling of electric vehicle (EV) charging is a critical challenge for smart cities, particularly in high-density infrastructure such as Mobility Hubs (MHs). This paper evaluates and compares two prominent approaches to the EV Charging Scheduling Problem (CSP): Mixed-Integer Linear Programming (MILP) and Reinforcement Learning (RL). We formulate a shared problem framework and apply both strategies under two structured scenarios: a small-scale deterministic benchmark and a medium-scale, realistic deployment with higher heterogeneity. Results show that MILP achieves optimal cost and SoC compliance in tractable cases but struggles with scalability. RL, based on Proximal Policy Optimization (PPO), achieves nearoptimal performance while scaling to 100 EVs with minimal computation time. Despite occasional SoC deviations, the RL agent exhibits robust and adaptive behavior under dynamic conditions. This study offers actionable insights for selecting and deploying EV scheduling strategies in real-world urban environments.

UR - http://pewasun.upc.edu/PEWASUN2025/program.html

U2 - 10.1109/MSWiM67937.2025.11308997

DO - 10.1109/MSWiM67937.2025.11308997

M3 - Artículo

Y2 - 27 October 2025 through 31 October 2025

ER -