The tour construction framework for the dynamic Travelling Salesman Problem

Abstract

The dynamic Travelling Salesman Problem (TSP) is a variation of the TSP with important real world applications. In the static TSP, the dataset must be complete and never change throughout the processing of the solution. The static TSP provides a rich theoretical framework in which to study the full optimization of the shortest route for a given set of points. The points in a static dataset are fixed forever, and there is no concept of time in the traversal of the dataset. On the other hand, the dynamic TSP considers the actual traversal of the dataset with respect to time. The dynamic TSP is more suited to a wide range of real-world problems, in which the dataset is incomplete and changeable. Most of the existing high-performing TSP solvers are constrained to static TSP datasets only, and these solvers are not readily transformed to handle dynamic TSP datasets. A recently introduced TSP solver is the Tour Construction Framework (TCF), which integrates both global and local heuristics in a complementary framework in order to efficiently solve the Travelling Salesman Problem (TSP). A potential advantage of the TCF is the ability to robustly solve dynamic TSP problems. In this research, standard TSP datasets are used to formulate dynamic TSP datasets and the TCF and mainstream TSP solvers are applied to solve these large dynamic TSP datasets. The performance of the TCF is evaluated for speed and accuracy.