Intelligent Transportation Systems Traveling Salesman Problem (ITS-TSP) - a specialized tsp with dynamic edge weights and intermediate cities

In this paper we present the Intelligent Transportation Systems Traveling Salesman Problem (ITS-TSP), which is a heuristic algorithm loosely based on the traditional TSP with three variations: the edge weights can change constantly, not every node in the graph must be visited, and simple cycles can exist. This problem has direct application to the transportation sector where vehicles leave from a source and need to visit a certain set of locations before returning back to the source. The ITS-TSP algorithm is analyzed to show a worst case running time of O(V³), assuming that all V nodes in a graph must be visited. There is a pre-processing cost of O(V²E!) that must be incurred, though this must only be performed one time for a graph. The algorithm is a heuristic that provides routes that are optimal based on a snapshot of the graph, though as the edge weights change over time, the solution may not be optimal. On a graph of the transportation network in Anchorage, Alaska, we tested the ITS-TSP with live data gathered through vehicle-tracking devices installed in 65 vehicles. With the weight on each edge representing the amount of time to traverse a roadway, the ITS-TSP algorithm always computed the route with minimum cost based on the snapshot of the network.