Wireless sensor networks (WSNs) are often subject to failures caused by energy depletion, software or hardware fault of nodes, environmental events, hostile attacks, and other reasons. It is critical to ensure a WSN application system is available during some presence of fault or interruption. Recent work in topology control has shown that a reasonable topology can improve the robustness of WSN. However, due to the limited resource of sensor nodes, topology control cannot easily tradeoff between fault tolerance and energy saving. To address this issue, we present a regular hexagonal-based clustering scheme (RHCS) and a scale-free topology evolution mechanism (SFTEM) for WSNs, that increases network survivability as well as maintains energy balance. RHCS uses a regular hexagonal structure for clustering sensor nodes, which satisfies at least 1-coverage fault-tolerance. SFTEM combines the reliability of RHCS with scale-free properties to connect clusters to form a robust WSN, which exploits the synergy between reliable clustering scheme and topology evolution, and can tolerate comprehensive faults including random failure and energy failure. In addition, to evaluate the performance of SFTEM, the simulation experiments were carried out to compare three factors including fault-tolerance, intrusion-tolerance and energy balance with other methods in literature. The simulation results show that, the performance of SFTEM is superior to those of the referenced topology evolution mechanisms of WSNs.
Node redundancy would be most effective to enhance the FT capability of sensor nodes. Hence, we refer the duplex sensor node as an FT sensor node.
In FT sensor node model, we assume that the redundant node is in a cold standby mode. The inactive node becomes active only when the active node is diagnosed faulty.
The sensing and transmission range of a sensor node are modeled as a disk of radius rs and rc, respectively. Zhang and Hou have proved that if the ratio between the transmission range and the sensing range, denoted as rcs, is not smaller than 2, then coverage implies connectivity. They have also shown that a regular triangular lattice pattern is optimal when the ratio 3csr.
If the node clustering scheme removes k nodes and still maintains the coverage of the scheme, the scheme is said to have k-coverage fault-tolerance.
When the strong FT node keeps operate properly, regardless of whether or not common FT nodes fail, RHCS is regarded as effective. When the strong FT node crashed but no common FT node fails, RHCS is considered effective;
When the strong FT node crashed, once any common FT node fails, RHCS is considered breakdown. Based on above assumptions, we exploit different initial state of RHCS and obtain the RFP.
Wireless sensor networks (WSNs) are usually composed of a large number of distributed sensor nodes organized in an ad-hoc pattern to monitor environments. In many applications, it requires high coverage and reliability to accomplish rigorous monitoring tasks, such as military mission volcanic monitoring and forest fire prevention. It further exacerbates the design challenge of meeting application requirements. WSNs always operate in unattended or hostile environments. The sensor nodes in WSNs are easy to breakdown caused by energy depletion or natural disaster and deliberate attack. In addition, the failed sensor nodes would reduce the coverage of the network, would split originally connected network, and even lead to an entire global network paralysis.
For example, if the several sensor nodes are breakdown and miss detecting the activity of the volcano malfunctions and gives fault readings, it might result in unneeded panic or loss of lives due to the absence of warning.In order to ensure high quality of service, it is essential for a WSN to be able to detect its faulty sensor nodes before carrying out necessary recovery actions. Fault detection in WSN is a technique which identifies a fault when it occurred and pinpoints the type of fault and its location. Fault detection techniques can be classified into centralized, distributed and hybrid
in scale-free WSNs, a few key nodes possess most connections of network. The energy of these nodes will be depleted much faster than other nodes, thus threatening the normal operation of the entire network. To tolerate comprehensive faults and keep energy balance, we exploit the synergy between reliable clustering scheme and topology evolution. In this paper, we first construct a reliable clustering scheme of nodes and analyze its reliability based on the Markov model. And then, we present a scale-free topology evolution mechanism of WSNs.
A regular hexagonal-based clustering scheme (RHCS) with FT sensor nodes as the vertexes of the hexagon is constructed. We characterize the reliability and fault rate hierarchically at FT sensor node and RHCS using Markov model. Then we obtain the random failure probability (RFP) of RHCS. We discuss the energy failure probability (EFP) of RHCS. Then we combine the RFP and EFP to model the JFP of RHCS. The relationship between the JFP and its important parameters is analyzed by the mathematical method to prepare the theory for topology evolution mechanism.A scale-free topology evolution mechanism (SFTEM) based on RHCS is presented. We treat a RHCS as an FT cluster, and evolve the topology based on the FT clusters. The connection strategy combines joint failure probability (JFP) and other characteristics of FT cluster, including node degree, node saturation and the distance between the cluster heads
WSNs are susceptible to failure due to the vulnerability of sensor nodes and attacks from malicious intruders. Hence, the fault-tolerance is an important issue in WSN applications. In this paper, we construct a regular hexagonal-based clustering scheme (RHCS) of sensor networks and analyze the reliability of RHCS based on Markov model. Then, we present a scale-free topology evolution mechanism (SFTEM). We also analyze the dynamic characteristics of SFTEM using mean-field theory.
Simulation results show that the node degree distribution of SFTEM follows a power law distribution, and both the fault-tolerance and intrusion-tolerance of RHCS outperform other models. However, our study has not taken into account the transformation of backup nodes after node failures. In the future, we will focus on developing a scheduling technique for the backup nodes that will wake up one or more backup nodes when the failure occurs in the network.