An efficient distributed control system for software-defined wireless sensor networks

Abstract

The advent of IoT as the next dominant player in cyber circles has ignited much research interest and a closer synergy between communication and computing models. Software-defined networking (SDN) and wireless sensor networks (WSNs) are some of the models envisaged to play a vital role in the IoT framework. SDN is an emerging network paradigm which has disrupted the status quo in networking and computing. This model is currently receiving much research attention and is being adopted rapidly by industry. It introduces flexibility, innovation, simplicity, and better management to networking. On the other hand, WSNs have always been used for monitoring physical and environmental factors such as temperature, humidity, vibrations, motions, seismic events, etc. The introduction and development of smart sensors have improved and advanced the WSNs. The emergence of the Internet of Things (IoT) paradigm has extended the scope of the demand of WSNs as they are considered to be the main building blocks of the IoT. However, WSNs continue to be plagued by challenges such as limited energy, computational capability, data storage, and communication bandwidth. The application of SDN to WSNs address most of the inherent WSNs challenges which have resulted in a new model of software-defined wireless sensor networks (SDWSNs). The SDWSN model is currently receiving much research attention as it has enormous potential for the future ICT. The SDN model advocates the separation of control logic and forwarding from the network elements. This decoupling leaves the element as a dump device and centralises the control logic in a controller. The controller in SDWSN is very vital and critical as it holds the intelligence and control of the whole network. The current major challenge is the centralisation of this controller. This makes the network vulnerable to malicious attacks as it becomes a simple target for adversaries. Another challenge is the fact that it stifles growth as it limits the scalability of the network and could potentially suffer performance degradation. Therefore, the reliability, performance, and efficiency of the network depends on the controller, despite operation. This study proposes an efficient distribution method for the SDWSN control system using the concept of fragmentation. This entails dedicating segments of the network to local controllers; these controllers are small and inexpensive but efficient. It also involves a global controller which has a global view of the network. This two-level architecture will leverage distribution, which will ensure availability and performance enabled by access. The purpose of this research study is to investigate if distributing an SDWSN control system is ideal, as well as to investigate the viability of the fragmentation model to achieve scalability, reliability and better performance. The evaluation shows that distributing the control system of the SDWSN is not only ideal but necessary. The fragmentation model also proved to bring a positive impact on the SDWSN control. The fragmentation model is envisaged to enhance the participation of SDWSNs in IoT. Therefore, the model is further optimised for ease of integration and deployment efficiency. This entails controller placement and controller re-election after controller failure mechanisms. The controller placement ensures a procedural and structured controller placement which aims at reducing the propagation latency between the sensor nodes and the local controllers as well as between the local controllers and the global controller. The controller re-election ensure that distance is taken into consideration when a controller is replaced by its peer after failure, thus ensures that the chosen replacement is closer to the failed controller. The two mechanisms were evaluated and proven to be efficient and improved performance.