Heuristic Algorithm for Solving a Multimodal Location-Based Concierge Service Problem

Document Type

Journal Article

Publication Date

2006

Subject Area

planning - methods, planning - route design, planning - signage/information, ridership - commuting, technology - geographic information systems, mode - bus, mode - rail, mode - subway/metro

Keywords

Wireless communication systems, Underground railways, Travel costs, Subways, Shortest path algorithms, Seoul (Korea), Routing, Points of interest, Peak periods, Off peak periods, Nodes (Networks), Multimodal transportation, Multimodal systems, Links (Networks), Heuristic methods, GIS, Geographic information systems, Geocoding, Concierge service, Computational time, Bus routes, Algorithms

Abstract

As a location-based services problem, the concierge service problem is to find a minimum total cost for purchasing predetermined types and quantities of items with the shortest paths connecting the locations where the items are available. These locations are defined as points of interest (POIs). The total cost includes purchasing and stopping costs at POIs as well as the travel cost from origin to destination. To respond to a request for such a service, a search algorithm should be reasonably fast in computational time and high in accuracy. A heuristic search algorithm was developed by a Euclidean distance approach using a geographic information system. The algorithm was implemented with 1,248 POIs in the Seoul, South Korea, metropolitan area as a case study site. The road network is composed of 52,915 nodes and 77,339 links, and the existing subway network, and all existing bus routes were used for the implementation. Four scenarios were examined: a request for a service in peak and off-peak periods with and without turning restrictions. The results indicate that the computational time in each case is less than 4 s with the transit networks only and less than 30 s if the road network is used. The paper also evaluates the solutions generated from the algorithm.

Share

COinS