Line-Leasing and Planning for Reconfigurable Private Data Networks
Posted: 5 May 1998
Date Written: Undated
Our paper studies the line-leasing decision and the circuit- routing problem in the design of reconfigurable private networks. These networks can automatically modify their logical configurations when necessary by employing digital connections. By exploiting the economies of scale available when using higher-capacity circuits, we develop an efficient method for determining a line-leasing policy that minimizes the packet delay and line-leasing costs. In addition, we also show how to derive efficient cross connection schemes and traffic-routing decisions for various short-run input traffic patterns. The various decision problems are formulated as non-linear mixed integer programs. We use Lagrangean relaxation and subgradient optimization to get near-optimal solutions with duality gaps ranging between 3 percent and 14 percent. We develop two distinct approaches for solving the line-leasing problem as a function of the potential input traffic patterns. One approach uses the average daily traffic and the other uses the more detailed intra-day traffic flow patterns. Numerical experiments indicate that the average traffic level can serve as a simple basis for deriving a near-optimal topology for reconfigurable networks. This finding allows network managers to reduce the costs involved in designing networks that exploit the flexibility of Digital Crossconnect Systems. We also show that it is economically beneficial for managers of large-scale private data networks to incorporate the logical design and traffic routing problems into their long-run line-leasing problems.
JEL Classification: L96, L86, C6
Suggested Citation: Suggested Citation