The user-controlled K shortest paths with diversity

Abstract

The user-controlled K shortest path problem with diversity (UKSPD) is a general form of the K shortest path (KSP) problem in graphs. Instead of finding the K shortest from point to point, the UKSPD determines the level of similarity of the K shortest paths through user input parameters. In this paper, we formally describe the UKSPD problem, which acts as a multi-objective optimization problem. Considering the application of genetic algorithm in multi-objective optimization, we propose an improved genetic algorithm to solve the UKSPD problem. The basic mechanism of the whole algorithm is as follows: chromosomes are directly represented as paths, crossover and mutation operations are performed to ensure the connectivity of the paths, and the user input parameter in each iteration determines the similarity of the selected paths. The proposed algorithm is tested on the New York City Map and compared with the improved Dijkstra algorithm, and the experimental results illustrate the effectiveness of the proposed genetic algorithm.