Simple, Reliable, Scalable and Energy Efficient Wireless Sensor Networks
Chenyi Guo · Research Repository (Delft University of Technology) · 2010
Wireless communication and networking technology has facilitated people to be connected with each other closely. Cellular network is evolving now from the third generation to the fourth generation. In the meanwhile we are experiencing the demand for wireless networks which can facilitate the communication between humans and environment, human and machines or even machine and machine. Such networks will help us know more about our surroundings which could lead us towards a better and greener life. Wireless Sensor Networks (WSN) is one candidate among such networks. It turns sensing tasks from small scale, centralized and expensive to large scale, distributed and low-cost by connecting small battery powered sensors with wireless links. We start the thesis by introducing WSN, its background and current status in Chapter 1. Although a lot of work has been reported in the literature on WSN, there are still many challenges. In this thesis, we focus on five of them, namely, (1)energy, (2)reliability, (3)scalability, (4)ease of use and (5)ease of set up. Energy is a challenge since WSNs are powered by batteries or even energy harvested from the ambience. The second, fourth and the fifth challenges are the hindrance in the way of high adaptation of WSNs while the third one will challenge when WSNs are largely deployed. Motivations and contributions of the thesis are also presented in the first chapter. Chapter 2 gives an overview of the literature in the several categories, such as physical layer, MAC layer, networking layer, synchronization and real deployment. Our work in the rest of the thesis is related to the work introduced in this chapter. We present the first result of our research in Chapter 3, which focuses on energy and reliability challenge on link layer. To improve the reliability of a link, we have to know the quality of the link. Thus we firstly analyze and try to improve link quality estimation methods in the chapter. We propose a new method for the estimation of packet delivery ratio which balances estimation accuracy and the overhead it causes. Then Minimum Energy Packet Forwarding (MEPF) protocol is proposed with the purpose of delivering a packet reliable with least amount of energy. MEPF tries to achieve the objective by tuning transmission power online for each packet with respect to the link quality. If a packet is lost, MEPF retransmits it smartly only when the link is considered to be good enough. Experimental results prove that MEPF uses almost the lowest possible transmission power without increasing the packet loss and retransmits a lost packet only once to eventually deliver it. We move a layer up from MAC to the network layer in Chapter 4. We organize a network into a better topology to improve energy efficiency and scalability. Two types of topologies are considered in this chapter, flat and clustered. In the former one each node has the same role while in the latter one nodes are organized into clusters where a node is either a Cluster Head (CH) or a Cluster Member (CM). We firstly analyze why a clustered topology may save energy then we quantify the saving. Since traffic is reduced in a clustered network, less contention or collision is expected and more nodes can communicate simultaneously. Thus a clustered network is highly scalable. To form a cluster topology from a flat one, we propose a cluster forming protocol which selects least amount of CHs which have the highest remaining energy. Thus they can live longer under higher traffic load compared to CMs. Simulation results show the feasibility and performance of the proposed protocol. Chapter 5 improves accuracy of localization, one of the most important WSN applications. One reason for the low accuracy is that the radio coverage of small and inexpensive antennas on sensor nodes, especially those in a Body Area Sensor Networks (BASN), is not omnidirectional. This problem leads to the failure of many localization protocols to achieve good accuracy since they are based on the assumption of omnidirectional antennas. In the chapter we proposed to use multiple receivers to locate a person in the context of a BASN. This method improves localization accuracy from a single receiver by mitigating the errors caused by varied and non-uniform beamwidth of antennas and combating fading with spatial diversity. We test this method in two classes of localization methods. The outcome of experimental results show that the method achieves a higher accuracy than a single receiver. Thus the reliability of localization is improved. Setting up a WSN especially for experiments is cumbersome and time-consuming process. It impacts the ease of use and set up. Thus we propose a framework for flexible and low-cost testbed in Chapter 6. Such a testbed only has sensor motes. Other than experiments, testbed management such as downloading the experimental code, reprogramming, testbed control, logging and collecting experimental results and synchronization are all carried out by the sensor motes wirelessly without extra devices. Thus a low-cost testbed can be set up quickly. A case study which realizes components in the framework is also presented. Finally the results of the thesis are summarized in Chapter 7. Future work is also presented there.