The application of optimisation methods to the design of large scale telecommunication networks

The design of a large scale telecommunication network is a complex process, involving the satisfaction of several tens of thousands of traffic demands in a network of several thousand nodes in the most efficient manner. The situation is further complicated by the hierarchical nature of the network, where digital signals of a particular bit rate are progressively multiplexed up to produce signals of higher bit rate which can then be more economically transmitted via a physical medium. The evolution of the network to meet changes in forecast demands over a period of years must also be considered. The problem may be formulated as a multilevel, multiperiod network optimisation problem where the levels of the network represent the different bit rates to which digital signals are multiplexed and the timesteps represent the successive periods for which forecast demands are available. The objective is to build a minimum cost network, subject to the satisfaction of the various constraints. The author has taken a heuristic optimisation approach involving the decomposition of the total problem into a series of manageable subproblems which are solved sequentially. At the heart of the overall optimisation procedure is the routing of individual demands, where Dijkstra's shortest path algorithm is used to determine the cheapest path.< >