Multi-objective Link-Separation Multipath Selection Using k Max-Min for Software-Defined MANETs
Pu Zhao, Wentao Zhao, Qiang Liu · 2019
Mobile Ad-Hoc Networks (MANETs) have been extensively studied and been developed for the last decades. In the fifth generation (5G) era, Software Defined Networking (SDN) has been emerging as a key technology enabling enormous killing applications. Hence, it is intuitive to interact MANETs with the SDN paradigm to construct more adaptive, flexible and scalable networks, which are called software-defined MANETs (SDMANETs). However, SDMANETs have two main limitations: (1) The bandwidth of SDMANETs is limited due to their wireless underlying infrastructure; (2) The connectivity of SDMANETs is notstablebecauseofnodes'highmobilityatanytime.Toaddress these limitations, we propose a multi-objective link-separation multipath selection (LSMS) algorithm using k Max-Min for SDMANETs. Specifically, the proposed method first transforms a LSMS problem to a multi-objective optimization one by jointly considering path bandwidth, length and load balance. After solving the problem, the proposed method can obtain k linkseparation paths with the maximal bottleneck bandwidth but the minimal bandwidth difference among these paths. Comparative experiments among the Shortest Path First Algorithm (SPF), the First k-Max Link-Separation path selection method and the proposed method over seven Waxman random topologies and two real networks (Cernet and DFN) demonstrate the superior performanceoftheproposedmethodintermsoftheaveragepath throughput and the bandwidth difference of bottleneck links in different routing paths.