Comparative Metaheuristic Approaches to Tourist Itinerary Optimization in Low-Infrastructure Urban Contexts: A Case Study of Tirana

Abstract

Tourist itinerary planning is a central component of smart tourism, yet it remains challenging in developing cities where computational resources and digital infrastructure are limited. This study examines Tirana, Albania, as a representative case for urban pedestrian-based tourist itinerary optimization. The analysis is carried out using both exact and heuristic optimization techniques, including Brute Force, Genetic Algorithm (GA), Simulated Annealing (SA), and a Hybrid Greedy + SA approach that integrates deterministic initialization with stochastic refinement. Distances between attractions were derived from OpenStreetMap, enabling fully reproducible experiments conducted on multiple datasets representing different sets of attractions with increasing size and under varying conditions, including ideal, noisy, and incomplete information. The results show that while exact computation rapidly becomes impractical as the instance size grows, metaheuristic methods, particularly SA and the hybrid variant consistently deliver high-quality and stable solutions. To evaluate real-world applicability under digital and computational constraints, the hybrid algorithm was implemented as a mobile-ready Progressive Web App and executed entirely on a resource-constrained device, demonstrating near-instantaneous optimization and confirming its feasibility for fully on-device use without reliance on backend servers. Overall, the study shows that lightweight metaheuristics, especially the Hybrid Greedy + SA method, offer a robust, scalable, and mobile-ready approach to urban tourism itinerary planning, suitable for deployment in environments with limited computational and infrastructural resources