Research On Data Compression And Data Authentication In Wireless Sensor Networks

Wireless sensor networks (WSN), attracting plentiful research efforts due to their wide range of potential applications, have been a very active research area. WSN usually consist of a large number of inexpensive sensor nodes that have strictly limited sensing, computation, and communication capabilities. The main tasks for WSN are to collect information from areas under surveillance. It is an important issue to save communication energy, and meanwhile to ensure the sampled data secure. The benefits of in-network processing include minimized amount of data transmission, reduced energy consumption, improved overall bandwidth utilization, and prolonged lifetime. In hostile environments, such as battlefield monitoring and home security, we must take account of the data security, including confidentiality, authentication, integrity, and freshness during transmitting data to the Sink node.This dissertation focuses on the challenges of data gathering in wireless sensor networks, aiming at high energy efficiency, low network delay and secure aggregation. We make our great efforts to design data compression algorithms and data authentification schemes tailored for WSN. The main works are as follows:(1) When the spatial correlations among the sensory data don't exist or vary, it is better to design algorithms running on a sensor independently. By designing an error tree and solving the regression equations set, we propose a data compression scheme with infinite norm error bound for wireless sensor networks. The algorithms in the scheme can simultaneously explore the temporal and multiple-dimension-stream correlations among the sensory data. The temporal correlation in one stream is captured by the 1D Haar wavelet transform. We propose a single data stream wavelet compression algorithm with error bound, named SWCEB. For multivariate monitoring sensor networks, some streams from one sensor node are elected as the bases according to the correlation coefficient matrix, and the other streams from the same sensor node can be expressed with one of these bases using linear regression. Thus we propose a regression-based multiple data streams wavelet compression algorithm with error bound, named MWCEB. Theoretically and experimentally, it is concluded that the proposed algorithms can effectively exploits the temporal and multiple-dimension-stream correlations on the same sensor node and achieve a significant data reduction. (2) MWCEB needs manual intervention. By designing an incremental algorithm for computing regression coefficients, a self-adaptive regression-based multiple-streams wavelet compression algorithm with infinite norm error bound (AR-MWCEB) is proposed. Based on error bounds and compression gains, the self-adaptiveness means that our algorithms make decisions automatically to transmit raw data or regression coefficients and to select the number of data involved in regression. Theoretically and experimentally, it is concluded that the proposed algorithms can effectively exploit the temporal and multiple-dimension-stream correlations on the single sensor node and exploit the temporal and spatial correlations among multiple streams on the cluster head and achieve a significant data reduction. Furthermore, we observe that the algorithms are also pretty good when multiple-streams correlations are reduced or non-stationary.(3) A critical and practical demand is to online compress sensor data streams continuously with quality guarantee. We make the following contributes. First, using the built-in buffer of sensor node, we present a piecewise constant approximation based data compression algorithm with infinite norm error bound, which is named PCADC-Sensor and is a near online algorithm. Second, with infinite norm and square norm error bound respectively, we propose two online piecewise linear approximation based data compression algorithms in sensor node, named PLADC-Sensor. A necessary and sufficient condition of PLA uniform approximation is given. Third, we propose a piecewise linear representations based data compression algorithm in cluster head or sink, named PLRDC-Cluster. It need not raw sensory data and can be applied to calculate aggregate functions. Last, our experiments on real-world sensor dataset show that the proposed algorithms match the sensor data stream model and can achieve a significant data reduction.(4) Data aggregation techniques can greatly help conserve the scarce energy resources in sensor networks by minimizing the number of data transmissions. Conventional data aggregation methods are vulnerable, as cluster-heads receive all the data from sensor nodes and then eliminate the redundancy by checking the contents of the data. A secure energy-efficient data aggregation and authentication scheme called SEDAA is presented. Intermediate nodes, i.e. cluster-heads in each level, implement data aggregation based on pattern codes without leaking the contents of the raw data and only distinct data in encrypted form is transmitted from sensor nodes to the base station, so SEDAA is confidential. Data integrity and authentication exploit two main ideas:delayed aggregation and delayed authentication. Instead of aggregation messages at the immediate next hop, messages are forwarded unchanged over the first hop and then aggregated at the second hop. TheμTESLA is adopted for authentication of messages transmitted by the base station. Data freshness is gained by using session keys calculated by counters. Moreover, a scalable secure data aggregation and authentication scheme called SSDAA is also presented. TheμTESLA key chains are used to reveal and authenticate keys locally. At every round, authentication need not wait until aggregation has been completed, so it can be applied to large scale WSN with a little delayed time. Both SEDAA and SSDAA can defend against intruder node attacks and replay attacks, and can limit the effectiveness of compromised node attacks.(5) Based on the above achievements, we design and implement a data compression based monitor prototype system. The system architecture consists of three layers:sensor networks tier, data service tier, application tier. On the basis of TinyOS, the sensor networks tier implements a data sampling module, a clustering and routing protocol, and some data compression algorithms in nesC. The sensor networks tier runs in Micaz mote. All compressed sensing data are transmitted to the gateway mote. The gateway mote is connected to the on-site PC via USB port. The data service tier is a middleware for message/data exchange between application tier programs and the WSN gateway mote. The application tier is implemented on the on-site PC or remote client PC. Its functions include local or remote monitoring, data analysis and visualization.