Cost modelling and validation in network optimization

There are many costs associated with the building and running a telecommunications network, both capital (CAPEX) and operational (OPEX) costs. On the one hand, complex routing schemes activated in multi-layer networks, like MPLS, logically imply assignment of costs to a collection of modules, from more than one layer, so defining the cost model is a non-trivial task. On the other hand, optimization methods based on prices for sending packets via different paths, which include methods using shortest path as a design principle, require a cost model to be precise, in order to achieve a correct solution. A well-defined cost model enables us to analyze and check costs for each layer and for each subsystem within a layer, and provides a step-by-step procedure for validation of prices. To guarantee the correctness of the defined cost model, we produce a well-structured report which confirms the validity of prices and of the total cost. The criteria of the validation of the cost model is that the total cost of the network when calculated from the prices to be attributed to the traffic is equal to the total network cost (CAPEX + OPEX) less any discounts. This validation process is similar to double-entry bookkeeping which is common in the business world and which enables detection of errors in accounting systems. It is shown that the requirement for validation justifies pricing based optimization. We provide examples where the application of this concept in a multi-layer network design software tool ensures that optimization works correctly, and one example where prices are not valid and the optimization fails.