Performance Evaluation of Software Defined Network (SDN) Based Surveillance System

Abstract

There are several challenges in traditional video surveillance systems such as storage
size and power consumption. In this study, the software-defined network is used to find
solutions to these issues. SDN is a new technology that simplifies the control of the network by
data plane is separated from the control plane. Also, it enables the multi-camera networks in
video monitoring systems to be reconfigured dynamically. The main idea of the proposed system
is to find a way for activate the cameras recording when necessary only by taking advantage of
the characteristics of SDN and without using motion detection sensors. As far as we know, we
are the first to take advantage of SDN features to implement activity and sleep modes on cameras
without the need for sensors. A use case is installing cameras on a roadside to monitor vehicle
movement on the campus. We created an application that interacts with the controller through
RESTful API web services. The application sends the time of active mode and the time of sleep
mode for each camera to the controller. Also, the controller developed to activate the cameras or
return them to standby mode according to the specified time by the network application. Finally,
videos sent from the cameras are merged into a single video on the server, which reduces the
time and effort spent when analysing received video streams. The system was simulated using
the Mininet emulator and the POX controller. The results showed that the proposed system
overcomes the traditional system in terms of reducing power consumption and storage size, thus
reducing the cost of multi-camera systems.