A Hybrid Approach for Routing Path Computation in Software Defined Ultra Dense Network

Abstract

In an OpenFlow based Software Defined Network (SDN), computation and installation of flow rules can be done proactively or reactively. Even though the proactive approach of SDN provides better performance in terms of transmission delay compared to its reactive counterpart, cannot be implemented in an extensive, unpredictable, and delay-sensitive network like Software Defined Ultra-dense Network (SD-UDN) of distributed Mobile Edge Computing (MEC) servers due to the memory limitations of SDN forwarding devices. Furthermore, the reactive approach fails to provide service to delay-sensitive applications. In this paper we have developed a hybrid approach that retains the property of installing flow rules only along the computed path and reduces the initial path computation delay during packet transmission. The proposed approach also reduces reactive computation burdens by resolving significant computation loads proactively and minimizes the total number of node expansions during path computation with A* algorithm. Emulation results in mininet demonstrate an improvement in the proposed approach's performance in terms of initial path computation delay to about 90% compared to the state-of-the-art reactive approaches.