| Summary: | 碩士 === 國立成功大學 === 土木工程學系 === 89 === Finding shortest paths for a traveler from a given location to a desired destination is a important process in Intelligent Transport Systems. The most formidable challenge to researchers is to fuel the speed of determining shortest paths up because the extreme bound is reached by using advanced data structures and searching algorithms such as many researchers utilize a priority queue (binary heap) implementation which reduces the calculating time of Dijkstra’s algorithm to O(mlogn). Although altering the data structures and search routines used in label-setting algorithms gains some degree of computational improvement, the best strongly polynomial-time bound runs in O(m+nlogn) and still be inefficient to calculate shortest paths by using poor equipment in traveler’s vehicle.
Different from these approaches, focus on using advanced data structures and searching algorithms, we proposed “2-phase reduced cost strategy”, which leads existing algorithms scan smaller scale arcs during determining shortest paths by utilizing special propertis of highway network structure and introducing reduced cost into second phase calculating.
Some substantial test was done, and test results obviously show that the algorithm we proposed is highly efficient, which can reach 32 times as traditional way in determining shortest path of network in Taiwan with turn constraints.
|
|---|
