Software-defined architecture and routing solutions for mobile ad hoc networks
Date
2019
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Delaware
Abstract
A mobile ad hoc network (MANET) is a self-organizing infrastructure-less network of mobile nodes needing multi-hop communication. MANETs support a wide range of applications in vehicular, mesh, sensor, and IoT networks, as well as in military operations, emergency search and rescue operations, disaster relief efforts, and providing Internet connectivity to remote regions. MANET characteristics make routing packets challenging because node mobility results in dynamic changes to the network topology. Unsynchronized transmissions result in increased interference, packet losses, and link instability, making communication unreliable, especially over multi-hop routes. ☐ Traditional MANET solutions have followed the decentralized paradigm, in which the nodes select routes for forwarding data packets either independently or as a group. In the past few years, Software-Defined Networking (SDN) has introduced a paradigm shift, in which centralized architectures have been sought-after for both wired and wireless networks. These architectures are designed to have an SDN Controller (SDNC), responsible for selecting network routes and dynamically controlling the network behavior in a centralized manner. Most existing SDN-based architectures for ad hoc networks propose using one or more of the following: infrastructure for hosting fixed SDNC, out-of-band single-hop links for control communication, location-tracking for learning network topology, or pre-existing IP connectivity for control communication. Such architectures are inadequate for infrastructure-less networks having intermittent links and susceptibility to high interference, packet losses, and collisions. ☐ In this dissertation, we present an architecture for Software-Defined Mobile Ad Hoc Network (SD-MANET). We presume none of the constraints imposed by the previous architectures. We design our SD-MANET architecture to have the following three functions: (1) learning route to SDNC, (2) learning network topology, and (3) sending network routes. We cater to the needs for proactive, reactive, and hierarchical routing protocols for MANETs by designing the following SD-MANET routing protocols: • 1. PCC, a proactive routing protocol • 2. CORR, a reactive routing protocol • 3. CPR, an improved proactive routing protocol • 4. HCPR, a hierarchical routing protocol ☐ MANET applications using low data rates for providing long-range communication endure low network capacity. In several contexts, the capacity is so low that the use of control packets completely overwhelms the network, rendering all existing solutions inadequate even for moderate-size networks. We address this issue by designing two novel zero-control-packet protocols: • 1. ECHO, a protocol designed for efficient network-wide broadcast • 2. VINE, a protocol for delivering a message to the specified destination ☐ These protocols do not use any control packets whatsoever. Instead, they include some additional information in the data packet header for building states in the nodes. Subsequent data packets get forwarded based on these states. The additional information in the data packet header is constant in size and does not scale with network size or density. We show it to be insignificant compared to the control packet sizes used by traditional routing solutions. We use some of the features of the ECHO and VINE protocols for designing the CORR, CPR, and HCPR protocols. ☐ We evaluate all our routing protocols for a wide range of scenarios, addressing scalability, load, density, and mobility. We present their simulation results and theoretical analysis. The results indicate that our SD-MANET protocols perform as good as, and in most scenarios outperform, the state-of-the-art protocols. In low-capacity networks, where all existing solutions perform unsatisfactorily, our protocols not only meet the requirements but also scale with network size and show superior reliability.