| Abstract |
Transport is an essential part of our lives. Optimizing transport provides significant economic and life quality improvements. Real-time traffic optimization is possible with the help of a fast communication network and decentralized sensing in smart cities. There are several analytical and simulation-based methods for traffic optimization. Analytical solutions usually look at more straightforward cases, while simulations can also consider the behavior of individual drivers. This article focuses on optimization methods and provides efficient traffic control based on simulations. The optimization goal is to find the proper sequence and timings of traffic light signals to ensure maximum throughput. In the article only the waiting time is selected as optimization criterion, but with knowledge of the vehicle stock (fuel type, fuel consumption, start-stop settings, number of passengers, etc.) it can be easily expanded to multi-objective optimization. In the literature, there are many optimization solutions, but all have some disadvantages mainly the scalability and the connectivity. Bacterial evolutionary algorithm and hill climbing algorithm are proposed in this paper with special area operators for the traffic optimization task. The developed memetic optimization algorithm can be efficiently scaled to optimize the traffic of even large cities. The method is efficient and well parallelized for real-time optimization use. For this study, a part of the city is examined in a SUMO simulation environment. The simulation result shows that our scalable memetic algorithm outperforms the currently applied methods by 35–45%. © 2023, The Author(s), under exclusive license to Springer Nature Switzerland AG. |
