Lagrangean Relaxation Algorithms for Fixed-Charge Capacitated Relay Network Design

Kewcharoenwong, P., Li, Q., and Uster, H., Lagrangean Relaxation Algorithms for Fixed-Charge Capacitated Relay Network Design, Omega, Vol. 121, December 2023.

The links to the relay network data instances are given below.

Please right click on the link and click “Save Link As” to save it on your machine.

As presented in Section 5.1 of the paper, in the generation of the random test, N nodes are uniformly distributed over a rectangular region with dimension 150100 (width height), which are used to present the potential RP locations and origins/destinations of commodities. After nodes are set, we determine D percent of the node pairs to have demand between them, thus, the number of commodities |Q| is given by |N|(|N|-1)/100. The combinations of |N|and D are used to define instance classes with different size. Each class consists of 10 randomly generated instances. The demand w_ij for a commodity [i,j] is randomly generated using a uniform distribution U[10,20].

In this example data, we have 560 txt files. Each file represents an instance of randomly generated demand matrix. The file name is N_20_D_20_demand_Iteration_1.txt, where the first 20 is the value of |N| (the number of nodes), the second 20 is the value of D (the percent of the node pairs that have demand). By combining the values of |N| and D (|N|=20, 30, 40, 50, 60, 70, 80, 100, 120, 140; D=20,40,60), we obtain 56 instance classes with different size.

The fixed RP location costs and link capacities with different problem sizes are presented in Table 1 of the paper.