Data Mining Approach For IDS in WSN

computer science engineering

Abstract— This paper proposed a mechanism of intrusion detection created in CWSN. According to varied capabilities and probabilities of suffer attack among sink, CH, and SN, three individual IDSs are designed. An IHIDS for the sink, a HIDS for CH, and a misuse IDS for SN are proposed. Feedback mechanism is used between the sink and CH; HIDS will be retrained for the new type of attacks, which have been detected and classified by IHIDS. For monitoring the status of packets in a Cluster Based Wireless Sensor Network, it is necessary for the packets to establish normal patterns of behavior. Therefore, in this thesis, the rule-based analysis system is used to construct anomaly detection modules and the corresponding rules are defined by experts. Three individual IDSs for the sink, Cluster Head and Sensor Node are planned according to varied capabilities and probabilities attacks that they suffer from. For the sink, an IHIDS is proposed which has the learning ability; it not only decreases the risk of attack, but also learns and adds new classes by learning mechanism in real time when the sink suffers unfamiliar attacks. For Cluster Heads, a HIDS is proposed which has the same detection models as IHIDS, but there is no learning ability in HIDS. Its goals are to detect attacks competently and avoid resource wasting. However, HIDS updates the classes of attacks using the feedback mechanism between Cluster Head and the sink. For Source Nodes, a misappropriation IDS is proposed. A simple and fast method for SN is designed, to avoid SN overwork, and to save resources for the purpose of safety.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.