Load Balancing in Network Voronoi Diagrams Under Overload Penalties
Input to the problem of Load Balanced Network Voronoi Diagram (LBNVD) consists of the following: (a) a road network represented as a directed graph; (b) locations of service centers (e.g., schools in a city) as vertices in the graph and; (c) locations of demand (e.g., school children) also as vertices in the graph. In addition, each service center is also associated with a notion of capacity and an overload penalty which is “charged” if the service center gets overloaded. Given the input, the goal of the LBNVD problem is to determine an assignment where each of the demand vertices is allotted to a service center. The objective here is to generate an assignment which minimizes the sum of the following two terms: (i) total distance between demand vertices and their allotted service centers and, (ii) total penalties incurred while overloading the service centers. The problem of LBNVD finds its application in the domain of urban planning. Research literature relevant to this problem either assume infinite capacity or do not consider the concept of “overload penalty.” These assumptions are relaxed in our LBNVD problem. We develop a novel algorithm for the LBNVD problem and provide a theoretical upper bound on its worst-case performance (in terms of solution quality). We also present the time complexity of our algorithm and compare against the related work experimentally using real datasets.
We would like to thank Prof Sarnath Ramnath, St. Cloud State University and the reviewers of DEXA 2018 for giving their valuable feedback towards improving this paper. This paper was in part supported by the IIT Ropar, Infosys Center for AI at IIIT Delhi and DST SERB (ECR/2016/001053).
- 1.Wikipedia: Catchment area. http://en.wikipedia.org/w/index.php?title=Catchment%20area
- 2.Okabe, A., et al.: Generalized network voronoi diagrams: concepts, computational methods, and applications. Int. J. GIS 22(9), 965–994 (2008)Google Scholar
- 5.U, L.H., et al.: Optimal matching between spatial datasets under capacity constraints. ACM Trans. Database Syst. 35(2), 9:1–9: 44 (2010)Google Scholar
- 6.U, L.H., et al.: Capacity constrained assignment in spatial databases. In: Proceeding of the International Conference on Management of Data (SIGMOD), pp. 15–28 (2008)Google Scholar
- 9.Xiao, X., Yao, B., Li, F.: Optimal location queries in road network databases. In: Proceedings of the 27th International Conference on Data Engineering (ICDE), pp. 804–815 (2011)Google Scholar
- 14.Jing, N., et al.: Hierarchical encoded path views for path query processing: an optimal model and its performance evaluation. IEEE Trans. KDE 10(3), 409–432 (1998)Google Scholar
- 16.Kleinberg, J., Tardos, E.: Algorithm Design. Pearson Education, London (2009)Google Scholar