On Improving Handover Delay in Software-Defined Mobile Networks

Abstract

The Software-Defined Networking (SDN) paradigm has become essential in tackling several issues and challenges in conventional networking, especially in mobile wireless networks. In order to realize the benefits brought by SDN to mobility management, we study the effects of SDN in conjunction with the OpenFlow protocol on the handover procedure. In this new setting, the completion of a handover procedure still requires the exchange of signaling messages between the control and data planes through OpenFlow channels, especially in the case of a hard handover, where the “breakage” in an on-going session is correlated to the exchange of management and reconfiguration messages. Consequently, the handover still suffers delay. One of the main causes of delay is the processing delay of those signaling messages. Three of the main factors contributing to the processing delay are the handling procedures, forwarding mechanisms of the network’s devices, and the load on controllers. Therefore, in this work, we target these three factors. First, we design an effective handling procedure of the exchange of signaling messages in the OpenFlow channels between the controllers and switches. Second, we provide a forwarding mechanism that is designed to reduce the number of switches involved in configurations of the data plane. Third, we propose a load balancing mechanism to mitigate the over-loading state that may cause longer delay periods. As for our first target, we provide an analytical model, which gave us the ability to model two handover-related OpenFlow messages. Following our analysis, we propose a novel solution to make handover more efficient and less interruptive. Furthermore, we study our solution in the Long Term Evolution (LTE) architecture. As for our second target, we take network heterogeneity into account, and accordingly, we provide an efficient approach using the MobileIP protocol, which shows improvement in decreasing total processing delay. Regarding the third target, we employ “vertical mobility” in our load balancing framework to minimize the maximum response time; thus, the users’ mobility-related procedures can be completed with less delay.