A construction heuristic for the capacitated Steiner tree problem

Many real-life problems boil down to a variant of the Minimum Steiner Tree Problem (STP). In telecommunications, Fiber-To-The-Home (FTTH) houses are clustered so they can be connected with fiber as cost-efficiently as possible. The cost calculation of a fiber installment can be formulated as a capacitated STP. Often, STP variants are solved with integer linear programs, which provide excellent solutions, though the running time costs increase quickly with graph size. Some geographical areas require graphs of over 20000 nodes—typically unattainable for integer linear programs. This paper presents an alternative approach. It extends the shortest path heuristic for the STP to a new heuristic that can construct solutions for the capacitated STP: the Capacitated Shortest Path Heuristic (CSPH). It is straightforward to implement, allowing many extensions. In experiments on realistic telecommunications datasets, CSPH finds solutions on average in time O(|V|2), quadratic in the number of nodes, making it possible to solve 50000 node graphs in under a minute.

[1]  Haizhou Du,et al.  Vulcan: Solving the Steiner Tree Problem with Graph Neural Networks and Deep Reinforcement Learning , 2021, ArXiv.

[2]  Stephen G. Kobourov,et al.  Computing Steiner Trees using Graph Neural Networks , 2021, ArXiv.

[3]  Louis-Martin Rousseau,et al.  Learning TSP Requires Rethinking Generalization , 2020, ArXiv.

[4]  Max Welling,et al.  Attention, Learn to Solve Routing Problems! , 2018, ICLR.

[5]  Thorsten Koch,et al.  SCIP-Jack—a solver for STP and variants with parallelization extensions , 2017, Math. Program. Comput..

[6]  Alain Hertz,et al.  On the edge capacitated Steiner tree problem , 2016, Discret. Optim..

[7]  Renato F. Werneck,et al.  A robust and scalable algorithm for the Steiner problem in graphs , 2014, Mathematical Programming Computation.

[8]  A. Hertz,et al.  Cabling optimization of a windfarm and capacitated K-Steiner tree , 2014 .

[9]  Markus Leitner,et al.  A Partition-Based Heuristic for the Steiner Tree Problem in Large Graphs , 2014, Hybrid Metaheuristics.

[10]  Fabrizio Grandoni,et al.  An improved LP-based approximation for steiner tree , 2010, STOC '10.

[11]  Martin W. P. Savelsbergh,et al.  Combining Exact and Heuristic Approaches for the Capacitated Fixed-Charge Network Flow Problem , 2010, INFORMS J. Comput..

[12]  Juan-José Salazar-González,et al.  A local branching heuristic for the capacitated fixed-charge network design problem , 2010 .

[13]  Renato F. Werneck,et al.  On the Implementation of MST-Based Heuristics for the Steiner Problem in Graphs , 2002, ALENEX.

[14]  Varadharajan Sridhar,et al.  Benders-and-cut algorithm for fixed-charge capacitated network design problem , 2000, Eur. J. Oper. Res..

[15]  Philip N. Klein,et al.  A randomized linear-time algorithm to find minimum spanning trees , 1995, JACM.

[16]  Chak-Kuen Wong,et al.  New algorithms for the rectilinear Steiner tree problem , 1990, IEEE Trans. Comput. Aided Des. Integr. Circuits Syst..

[17]  Inmaculada Rodríguez Martín,et al.  A local branching heuristic for the capacitated fixed-charge network design problem , 2010, Comput. Oper. Res..

[18]  Stefan Voß,et al.  Capacitated Minimum Spanning Trees , 2009, Encyclopedia of Optimization.

[19]  Piet Demeester,et al.  Three heuristic techniques for topological access network design , 1997 .

[20]  Richard M. Karp,et al.  Reducibility Among Combinatorial Problems , 1972, 50 Years of Integer Programming.