Refreshment strategies for the shortest path caching problem with changing edge weight

Abstract

The problem of caching shortest paths which aims at reducing the computational time of servers has been widely studied. All the existing methods addressing this problem assume that the graph status does not change with time. Based on this assumption, they analyze shortest paths query logs and prefer to load paths with the most query frequency into the cache. However, the graph status is actually affected by many factors and undoubtedly changes with time in the real work. As the existing approaches ignore the change of graph status, they cannot guarantee the efficient use of caches. In this paper, we first exploit properties related with changing graphs. Then we develop an algorithm to detect shortest paths affected by weight change of edges. After detection affected paths in a cache, several heuristic based refreshment strategies are proposed to update the cache. In the experimental section, performances of proposed refreshment strategies are compared. © 2014 Springer International Publishing Switzerland.