An Architecture for Software Defined Drone Networks Mohannad Alharthi School of Computing Queen’s University Kingston, Ontario harthi@cs.queensu.ca Abd-Elhamid M. Taha Electrical Engineering Department Alfaisal University Riyadh, Saudi Arabia ataha@alfaisal.edu Hossam S. Hassanein School of Computing Queen’s University Kingston, Ontario hossam@cs.queensu.ca Abstract—Drones or Unmanned Aerial Vehicles (UAVs) are utilized in a wide range of applications, as they are considered flexible and cost-effective. Novel applications have been recently explored, such as providing communications and Internet cov- erage where ground infrastructure is lacking or in temporary situations. In this paper, we propose a drone-based network architecture enabled by Software Defined Networking (SDN) to provide dynamic and flexible networking capabilities, suitable for different types of drone applications and deployments, while we discuss associated challenges related to SDN in done networks. I. I NTRODUCTION Unmanned Aerial Vehicles (UAVs), aka drones, are consid- ered key instruments in emergency situations such as search and rescue, surveillance, and various scientific and civilian applications. The goal of using drones is reducing the cost of missions and eliminating associated risks of sending human personnel to conduct risky or costly tasks, especially in the case of natural disasters and tasks in difficult to reach areas. As drones became more cost-effective and capable of carrying communication technologies, they became an attractive solu- tion to deploy as a fleet of cooperating drones to cover larger geographical areas and relay information to remote targets. This also enabled more advanced applications as currently explored in the literature, e.g., as flying base stations (BSs) in 5G networks, and as flying networking and computing infrastructure to support drone missions and communication networks. Drone-based networks can offer great flexibility to satisfy dynamic and unpredictable service demands, while reducing the cost as an alternative to deploying fixed ground infrastructure. The use of multiple drones instead of a single one can be motivated by cost reduction and service enhancements [1]. Generally, using multiple small connected drones can cost less than using a single large drone [2], in addition to extending the coverage area and possibly complete tasks faster. While equipped with communication capabilities, a key advantage of such systems is that drone missions can still operate even if some drones fail. Benefits also include the ability to scale out the network by adding more drones as needed. Additionally, connectivity between drones enables creating a network in the sky that relay information over a large area, while drones’ computing resources can provide some network functions and application-specific processing, so the network can be creatively applied in a wide range of scenarios. Our vision for the drone network is that it can be reusable for a variety of applications. Thus, it is practical to develop an architecture that is flexible and cost-effective for doing so. Software Defined Networking (SDN) enables fast adoption of new innovations in networking technologies by moving the behavior of networking devices to a logically centralized controller. The decoupling of the control and the data planes allows for central network management that is unified and flexible regardless of the underlying hardware. As well, it enables implementing adaptive network solutions on the go for the network according to current state and environment. Addi- tionally, SDN can virtualize the network for multiple tenants simultaneously, and isolate unrelated application traffic. In this paper, we propose an architecture for an SDN-based drone network that we believe is suitable for implementing a wide range of drone deployment scenarios. It is useful for use- cases such as network coverage, and sensing and scanning, and is able to operate where no access to networking infrastructure is available. It can be utilized by law enforcement agencies, mobile network providers, and scientific researchers. We also discuss challenges in designing SDN-based drone networks. In the following section, we offer some background ma- terial. In Section 3, we discuss some related work, and in Section 4 we present the challenges and requirements of our architecture and its design. Finally, in Section 5, we discuss use-cases of our architecture. II. BACKGROUND A. Software Defined Networking (SDN) In SDN [3], networking devices (specifically switching devices) are turned into simple but flexible devices that expose their functionality through a programming interface. Instead of operating independently with hardcoded functionality, de- vices are controlled by logically centralized controllers that implement the control logic and dynamically direct how packet flows are routed via well-defined API. Internal flow tables in switches are manipulated by instructions received from the network controller. The API let the controller insert forwarding rules based on packet headers and associated forwarding and processing actions. The controller maintains a global view 978-1-5386-8088-9/19/$31.00 ©2019 IEEE