QoS-Aware Flow Routing with Minimizing Active Links and Rule Capacity Constraints in SDN Networks

The growing demand for digitization necessitates modifications to current Internet technologies to effectively support Internet of Things (IoT) applications and services. Software-defined networking (SDN) supports network abstraction and dynamic management through a central controller. The ternary content addressable memory (TCAM) memory used in the OpenFlow switches is very costly and limits the flow table size. The limited size of TCAM memory motivates us to think about the flow rule replacement in SDN switches. This paper introduces a routing approach to meet traffic flow requirements in the SDN networks, consisting of two main phases - routing path selection and flow rule replacement. In the first phase, we formulate a minimum cost routing problem to minimize the network’s active links as an integer linear program (ILP) in SDN networks. We then introduce a greedy heuristic-based solution to solve the ILP in polynomial time. In the second phase, we propose a flow rule replacement method that utilizes idle timeout settings to prevent overflow in the flow tables of SDN switches. The simulation results indicate that our proposed approach reduces active links by $\mathbf{1 5 \%}, \mathbf{2 4 \%}, \mathbf{3 4 \%}$ in the Goodnet topology, and by $\mathbf{1 1 \%}$, $\mathbf{2 3 \%}$, 33% in the Sprint topology as compared to benchmark schemes ROSA, LARAC, and SPD, respectively. Additionally, we analyze the impact of idle timeout on flow table entries within both our proposed approach and the benchmark schemes.