Abstract

In this paper, we consider the problem of assigning users to switches and controllers at minimum connectivity costs. We simultaneously minimize the existing latency between controllers and switch controllers for software-defined networks (SDN). In particular, we propose two mixed-integer quadratic programming formulations and their standard linearization counterparts. All the proposed models are solved with the CPLEX solver. More precisely, the quadratic models are solved with the branch and cut (B &C) algorithm of the solver. Whilst the linear ones are solved with both, the B &C and with the automatic Bender’s decomposition algorithmic option of the CPLEX solver. To the best of our knowledge, this work constitutes a first attempt to propose mathematical formulations while including user connectivity to a backbone software-defined network. Notice that we consider which is the best strategy for connecting users to the backbone, either connecting them to switches or to the controller nodes of the network. Our preliminary numerical results indicate that for most of the benchmark instances considered the second strategy which connects users to switches has a better performance in terms of objective function values. On the other hand, we observe that the linear models offer better performance in terms of CPU times to get the optimal or best objective values. Finally, we see that the linear models can be solved faster with the branch and cut algorithm than using Bender’s decomposition approach.