Linear programming based heuristic algorithms for tunnel assignment and reconfiguration in MPLS networks

Multi-Protocol Label Switching (MPLS) sets up a connecting tunnel for each origin-destination pair of traffic demands to send data packets over Internet Protocol (IP) networks. We should assign each tunnel to one of the available candidate paths to achieve efficient use of the network and avoid congestion. Additionally, we need to update the bandwidths and routing of the tunnels efficiently in short periods of time to accommodate fluctuating traffic demands, which we refer to as the reconfiguration problem. In this paper, we propose formulations and linear programming-based heuristic algorithms for the tunnel assignment and reconfiguration problems arising in MPLS networks. First, we propose a formulation and algorithms for the tunnel assignment problem, considering realistic constraints such as port limitations and priority queue capacity restrictions. Our model incorporates unidirectional, bidirectional, protected, and unprotected tunnel cases altogether. Then, we present a heuristic algorithm to obtain a schedule for changing the bandwidths and routing of tunnels from the current configuration to a target configuration quickly while adhering to network capacity constraints. We report computational results using some random data from the real-life telecommunication network of the Republic of Korea.