Research On Fault-Tolerant Mechanism For Wireless Sensor Networks

Wireless sensor networks(WSN) can be applied in many areas, such as:military, structure health surveillance, medical care and environmental monitoring. Wireless sensor networks aim to collect the data produced by sensor nodes and transmit sensing information to users for processing and making decision. For wireless sensor networks, guarantee of the integrity and accuracy of sensing data is the main demand of Qos.However, in harsh environments, the quality of data collection and transmission decreases, due to failure sensors, unreliable wireless channel and network congestion. The asymmetry characteristics between sensing and transmission capacities are mainly manifested in two aspects. Firstly, sensors in general wireless sensor networks collect easily transmitted information such as light intensity, temperature and other scalar information, while long-term exposure in harsh environments makes sensors prone to failure. Secondly, with the development of semiconductor technology, reliable transmission of a mass of multimedia data is still difficult to be achieved by the low rate wireless communication devices, even though multimedia sensor nodes are capable of collecting the data more accurately. It’s necessary to design fault tolerant mechanism for WSN. This dissertation focuses on the challenges of fault tolerant mechanism in wireless sensor networks. The main works are as follows:(1) A faulty sensor node tolerant algorithm based on cut point set is proposed in the presence of failure sensor issues, by introducing the concepts of spatial correlation model, strong correlation graph and cut-point set. The algorithm first finds out a cut-point set, which has strong spatial correlation with faulty sensor node. According to the observations of cut-point set, the faulty sensor node is able to predict its missing sensor readings by using orthogonal intersection estimation method. Analytic results show that, the algorithm not only can tolerate the faulty sensor node, but also accurately predicts miss-readings, keeps network connectivity and overload balance. The results of miss-readings estimation, obtained from simulations and a greenhouse monitoring experiments, show that the methodology presented in this paper can successfully predict the missing sensor readings.(2) We further propose a faulty sensor node tolerant algorithm based on irregular sensing model. At first, the failure sensor node classifies its neighbor nodes, by utilizing the geographic information and spatial correlation among neighbor nodes. Therafter, the boundary of irregular sensing model is sketched based on a distributed convex hull deciding algorithm. According to the boundary of irregular sensing model, some neighbor nodes’ measurements are selected to predict failure sensor’s miss-reading. In complex geographic environments, the algorithm presented in this paper can rstore the blind spot accurately.(3) Aiming at improving the bandwith utilization in visual sensor networks, an image compression algorithm based on divergence model is proposed. First, by using divergence model, a bi-level gray image is produced in each node. Because a gray scale pixel can be represented by1bit in bilevel image, the amount of image data is reduced. Thereafter, utilizing the feature of cooperation in visual sensor networks, an orthogonal extention mechanism is proposed. Due to the information represented by one byte is shared by several cluster nodes’bi-level pixels, the degradation of received images is controlled. Comparing with the traditional image compression algorithm, the quality of received image measured by PSNR is higher, as the average packet loss rate increases.(4) Considering the effect of transmission losses on the visual quality of images is always varying and depending on the burst loss length, an energy aware interleaving algorithm is presented in this paper. Among the existing transmission error control techniques, interleaving can improve the visual quality of images without redundant data incurred. Conventionally a larger interleaving data size will be more effective in converting long burst loss into isolated losses. This is at the cost of transmitting more pixels. But how to effectively reduce individual sensor’s data load in an energy-constrained distributed transmission network is still an unsolved issue. An energy-aware packet interleaving algorithm(EAPI) is proposed in this paper to regulate burst loss effects by spreading out packets according to each image region’s pre-calculated transmission income. Experimental results demonstrate that the proposed scheme can not only improve the end-to-end image transmission quality, but also prolong the lifetime of visual sensor network.(5) Based on the above achievements, we implement an image transmission prototype for visual sensor networks. The hardware platform model is iMote2, which is produced by Xbow enterprise. And we design our software based on Linux. Some functional modules are implemented, such as:image format convertion, image copies generation, interleaving transmission. Utilizing these functional modules, a centralized image search engine is completed.