Parallel savings heuristics for designing multi-center tree networks

Parallel savings heuristics are presented for the multicenter tree-network design problem, where the objective is to find the least cost connection between a given set of user and backbone nodes (centers). Each subtree has to satisfy the capacity restriction of the backbone nodes. This approach combines the savings method with the matching problem, and is compared to L.R. Esau and K.C. Williams' (1966) heuristic modified to handle the multicenter case. This parallel savings heuristic allows one to decide how many backbone nodes to open and how to connect the user nodes to these backbone nodes. The performance of the parallel savings algorithms is demonstrated on a very large set of test problems.<
>