A design method for cost minimization of highly reliable communication networks

In designing a highly reliable nonhierarchical communication network, it will be more practical if not only the required number of disjoint routes is ensured, but also the cost is minimized. Under certain assumptions, the communication network cost can be evaluated by a simple matrix calculation. The branch cost and its associated switching node cost are evaluated by the sum of the fixed cost and the traffic proportional cost. It is assumed that the traffic is carried through a minimum cost route in a normal state and, based on this assumption, the traffic offered to each branch is calculated. Then the communication network cost for given nodes, branches, and traffic matrix, is obtained by matrix calculation. Moreover, the approximate formula, which has already been proposed, is used to calculate the number of disjoint routes. Therefore, the minimization of the network cost under the constraints of the number of disjoint routes can be done by matrix operation. Since the matrix calculation is especially suitable for a supercomputer, relatively large‐scale networks can be designed systematically by the proposed design method.