| Recent advances in wireless communications, electronics and Micro Electro Mechanical System (MEMS) have enabled the development of low-cost, low-power, multifunctional sensor nodes, which are small in size and communicate in short distances. Wireless sensor networks are envisioned to consist of thousands of devices, operating in an unattended mode, and characterized by severe energy constraints. The unique features of wireless senor networks offer unprecedented opportunities for broad civilian and military applications, such as industrial automation, military tactical surveillance, national security, and emergency health care. One of the most important sensor network applications is monitoring inaccessible environrnents. The noisy environment and energy constraints, however, challenge the event detection problem.Most of recently proposed fault-tolerant event detection algorithms are only based on the spatial correlation, such as distributed Bayesian fault-tolerant algorithms, NED, and moving median based method. In these algorithms, the frequent exchanges of measurements among nearby sensor nodes give rise to much energy dissipation. Moreover, detection accuracy is poor at the boundary of event region and the edge of sensor networks.In order to improve the performance of detection accuracy and energy efficiency, we propose temporal correlation based and temporal-spatial correlation based fault-tolerant event region detection algorithms, leveraging spatial and temporal correlations. It is validated through simulations that, the proposed temporal based fault-tolerant can efficiently reduce the probability of detection fault for static environment where the characteristic does not change frequently, dissipating little energy. And the proposed temporal-spatial correlation based algorithm outperforms the spatial correlation based scheme in terms of detection accuracy and energy dissipation, thus making the proposed algorithm attractive in energy-efficient event region detection applications.In addition, many data aggregation and data-centric routing schemes are proposed for sensor networks, such as Directed diffusion, LEACH, PEGASIS, TAG, and CAG It's demonstrated through related work that, the transmissions in the network will be reduced employing data-centric routing that perform in-network data aggregation, and consequently the energy consumption is reduced. But none of the previous data aggregation schemes is intended for fault-tolerant event detection application.Based on the previous work of data aggregation and data-centric rooting, we propose a fault-tolerant data aggregation scheme for event detection application. The proposed data aggregation scheme coordinates query and rooting with fault-tolerant event detection application. Fault-tolerance can be obtained by checking the parent node's detection result during query distribution phase and the child node's detection result during collection phase. It's validated through theoretical analysis and simulation results that, given the number of neighbors N=4, the fault-tolerance of proposed data aggregation is equivalent to that of Bayesian optimal threshold decision scheme. The proposed fault-tolerant data aggregation is energy-efficient, as not all the sensor nodes need to exchange detection results. Time division makes nodes in radio idle state (low power state) in most time and forward single aggregation packet during each query epoch, saving more energy.The results in this paper are useful for future investigations on the fault-tolerant event region detection algorithms and fault-tolerant data aggregation in wireless sensor networks. |
PDF Full Text Request