You are on page 1of 4

The paper is about the fault tolerance of the wireless sensor networks.

Wireless sensor networks are communication networks that consists of sensor nodes that collect data process it and send it to the end stations via other sensor nodes and gateways. Wireless sensors are in form of clusters networks. Where, one or more sensors form a cluster and mulitple clusters form a super cluster. Each super cluster has a gateway that acts way to other super clusters and transit networks.The sensor nodes collect data from the environment and send them to a node called sink node for processing which involves a great deal of traffic overhead in WSN which already have bandwidth lesser than that of wired communcation network. (diagram of WSNs) Recently to reduce traffic overhead of the WSNs a concept of Mobile Agents (MA) are introduced. MA are software codes that travel accross the network performing computations for the sensor nodes rather than sensor nodes themselves sending data to sink nodes for processing. So, MAs can be used to greatly reduce the communication cost, especially over low bandwidth links, by moving the processing function to the data rather than bringing the data to a central processor. The faults in the sensor nodes of the clusters must be detected and reported in correct point of time to make communication system work properly. In this paper, fault detection and correction is carried out using Mobile Agents based on intersting architecture of Honey Bee Dancing. Honey bees communicate by dancing and other bees understand this dancing language. e.g. If a honey bee returns after collecting food it dances to communicate the source of that food to other bees in the hive. Honey bees perform three types of dances round dance waggle dance tremble dance Each of the three dances have different meaning that bees communicate to their fellows. By round dance bees mean that the source of the food is within 50 mts. of hive. Waggle dance means that the source is beyond the distance of 50 mts. of the hive. If a bee performs tremble dance it means that it want to change its current role. Mobile Agents in proposed in this paper perform two basic roles sniffers correctors Sinffers MAs sniff the error in the sensor nodes and corrector MAs correct the error in sensor node. Mobile Agents have three types Local Agents (LAs) Metropolitan Agents (MAs) Global Agents (GAs) In local agents, Local Sniffers (LSs) and Local Correctors (LCs) are present on some sensor nodes of the cluster. LS detect error and report to LC which correct the fault. Metropolitan agents consist of Metropolitan Sniffers (MSs) and Metropolitan Correctors (MCs) that monitor super cluster at the gateways. Whereas, Global Agents having GSs (global sniffers) and GCs (global correctors) look after the whole of the

network and are present at the sink. This paper also classifies the faults into following three basic types: endemic fault epidemic fault pandemic fault Faults inside the cluster are reffered to as endmic faults. When same type of fault occurs at the different nodes of the different clusters within a super cluster it is known as epidemic fault. Whereas, pandemic fault refers to the same type of fault in multiple super clusters of the network. As stated previously, faults are detected by sniffers. Sniffers move about the network and collect attributes. These attributes are saved in databases, and are compared at different points in time. If there is no difference between the previous values and new values then there is no error and previous value is replaced by new value in the database. Error is detected when there is unexpected difference between the two values. Faults detected by the sniffers are communicated to correctors, so that they can repair the fault. This communication is based on honey bee dancing. Honey bee dancing is employed by the sniffers to signal the correctors to locate the faulty areas of the cluster and to correct the faults.If the fault is detected by LS in the cluster at some node then LC of that particular area will mend it. But if many nodes in the cluster are faulty then LS will perform round dance to indicate endmic fault. The LCs of that cluster after seeking permission from their LSs move to the faulty region and correct errors. After correcting the error the LC reprot to the LS which in turn tell to MS that maintains the log. When the LS of the cluster perfroms waggle dance it means that fault in the cluster has turned into epidemic. Waggle dance performed has two dimensions: indication that the LCs are lesser in numbers as compared to the faults informing LSs and LCs of other clusters about the fault

the LCs of epidemic cluster cannot handle the faults by themselves so LCs from other cluster take permission from their respective LSs and move to the faulty cluster to correct the faults. after correcting the faults they return back to their local clusters. If one or more LSs in different clusters of the super cluster start waggle dance then MS considers the fault as super cluster epidemic. And MS keep the log of it. MS itself performs waggle dance when the epidemic fault is spread to all clusters of the super cluster and reports to GS.

If the fault spreads to several super clusters of the network i.e. the outbreak of pandemic fault, then GS starts tremble dance which is an indication to MSs and LSs to start multiplying the number of LCs and MCs for the correction of pandemic faults. After correction the GS keeps a check whether waggle dance is performed again or not. If it is again performed it means that the problem is not resolved and GS again performs tremble dance but if it is not performed then GS issues a kill message to bring the number of correctors back to its normal position. Hence, the authors have proposed a solution of faults in sensor clusters using Mobile Agents and honey bee dancing based communcation system.

You might also like