Development of Survivability Protocols in Wireless Personal Area Networks for IoT Applications

Abstract

Internet of Things (IoT) is a renascence of the Internet that gathers rapid momentum propelled by the evolutions in mobile and sensing devices, wireless communication and networking technologies, and cloud computing. The proliferation of IoT applications like healthcare monitoring services and others has accelerated the demand for wireless networks of tiny devices. Wireless sensor networks, enhanced communication technologies, distributed intelligence for smart objects, wireless radio frequency identification systems, and several other Wireless Personal Area Network (WPAN) technologies and communication solutions together enable the promising next-generation Internet, IoT. Many IoT applications are developed for continuous health monitoring and fitness supervision. IoT healthcare application would gather physiological parameters by using non-invasive wearable devices like wristbands, belts. Elderly care, smart homes, wearable technologies, personal assistance for belongings tracking are some of the practical use cases of Consumer IoT (CIoT), which brings consumer devices to the network. A healthcare monitoring application of the CIoT network is considered in this research that consists of many numbers of small body area networks, which are composed of six to nine different sensors and a master coordinator. The coordinator node collects and aggregates these packets to compose data chunks and transmits to the local processing and storage station. Many numbers of persons with such a system constitute the entire application network. The base station would be connected to the cloud and make the whole system remotely accessible.
In such IoT applications, network survivability is an important attribute to be considered. Survivability for a network topology has to be achieved through many approaches, such as reliable communication techniques, efficient utilization of energy resources, adaptive techniques to reduce undesired topology changes and performance degradation, and efficiency to get along the security issues and failures. Efficient communication protocol design and topology structure evaluation for undesirable changes are of utmost importance, which is tried through this research. Protocols are developed at different stack layers, mainly routing and MAC layers, to maintain network survivability at different levels such as node, link, path, and topology. Simulation experiments are implemented to analyze the proposals at different network topologies that pretend real IoT application scenarios. Further, a centralized network evaluation tool is developed to analyze the network performance and scrutinize the structural changes in the topology for improving the survivability of the network. Network survivability is an important attribute to be considered in IoT vii applications. Some researchers observe the network survivability as the topology coverage and the time to get the network disconnected. Further, the survivability of a network has to be maintained at different levels. Path survivability is the stability in the routing paths. Link survivability ensures the efficient usage of the channel between two peer nodes of a hop. Node survivability can be achieved by effectively utilizing the node’s energy and making it less congested with packet bursts. Algorithms at different layers of the protocol stack should jointly operate to maintain the survivability of the network. The thesis comprises four contributory chapters along with the introduction, literature survey, and conclusion
In the first contributory chapter, a protocol at the routing layer is designed that selects the next-hop node in the data forwarding path which maximizes the network survivability. The proposed routing protocol is a data forwarding technique that maintains the network survivability by choosing the path which has a higher survivability factor. It also tries to minimize the congestion at the nodes by including the network traffic information (congestion level of a node) from the physical layer as the route choosing factor. The routing choice decision-making process also includes the signal strength information of the previously received packets.

An algorithm that efficiently allocates the channel dynamically among the contending nodes is proposed as the next contribution, in which the nodes get priority in the contention resolution process whose corresponding receivers are having stronger survivability metrics. Survivability of links between the contending nodes and their receivers are also considered during the process of channel allocation. The application data rate of the sender and the service rate of the receiver are considered for improving the receiver’s node survivability. Link survivability metrics like channel strength indicator, link quality indicator, and path loss distance between the hops are used to decide the spell for accessing the medium. Information from other layers and the assistance of proposed survivable path routing protocol are used with cross-layer design for the efficient assignment of the channel access. The protocols have been evaluated in an IoT remote healthcare application with a cross-layer design of developed MAC and routing techniques.
The proposed survivability aware protocols are implemented in the protocol stack of Contiki-OS. Real-time communications are addressed where there are many nodes simultaneously transmit their application data frames towards the base station. Since the network layer routing protocol in Contiki is by definition made suitable for modifications, the proposal tries to maximize the survivability of the links between hops, to reduce the energy disparity in the nodes, and to avoid congestion at the relay nodes. Contiki provides an IEEE 802.15.4 compatible CSMA driver at the MAC layer. This CSMA protocol is used to adapt to the proposed channel allocation technique. The proposed survivability aware protocols are adapted with the protocol stack of the Contiki, which is a multi-tasking OS for networked, resource-constrained embedded systems and wireless personal area networks. The protocols are tested with hardware implementation on FIT IoT-Lab, which is an online viii testbed infrastructure for IoT-enabled wireless sensor networks.
A python-based centralized topology evaluation module is developed to analyze the performance by using the data collected from the mesh network deployment with Contiki-OS using the Collect-View plug-in of the Cooja simulation environment. The developed tool evaluates the network health based on the collected information from the deployed topology. The collected data contains the details about each node in the network, such as the IP address of the node, hop count from the gateway node, next-hop node towards the gateway, uplink and downlink packet delivery ratio, physical layer channel number, Tx/Rx airtime, etc. The global network structure is reconstructed by the proposed tool from this local information by using python libraries for networking and graph theory. Once the network structure is constructed, it is decomposed to different DODAGs, i.e., subtrees. Each subtree is grounded to the gateway node that is connected with the outside Internet backbone. The structure and characteristics of these subtrees may change over time. The tool evaluates the operational health of the subtrees and investigates any possible changes in the structure based on historical data for better functionality. The final chapter of the thesis presents the concluding remarks inferred from the proposals, along with the emphasis on the achievements and limitations. The future extents for improvements are outlined at the end.