Research On False Data Detection And Filtering In Wireless Sensor Networks

With the widespread use of wireless sensor networks in various fields, the security threat increases significantly. However, sensor nodes have limited computing power, restricted memory space and communication bandwidth and wireless sensor networks are usually deployed in unattended environments or hazardous areas. In addition, the wireless channel is open and shared. All of these factors exert a great challenge for the research of security mechanisms for WSN.Attacks aimed at WSN can be divided into two branches:outside attacks and inside attacks. Outside attacks are usually simple to be detected and prevented and there have been a number of security mechanisms that can resist most of the outside attacks effectively. While inside attacks are more insidious and can cause more damage since attackers have obtained keys or authentication information stored in compromised nodes and can masquerade as trusted node. False data injection attack launched by compromised nodes is a typical inside attack. Bogus reports that have been injected into the network can cause false alarm, wrong decision, as well as energy waste in forwarding nodes.In this dissertation we focus on how to detect and filter the false reports injected by compromised nodes within fewer hops effectively. We analyze the advantages and drawbacks of existing schemes of domestic and foreign and propose node legitimacy based false data filtering scheme (NLFS).To address the problem that existing schemes cannot resist false data injection attack launched by nodes from different geographical areas, NLFS not only verifies the correctness of authentication information in false packets like other schemes but also verify the legitimacy of nodes that endorse the report, and attackers have to obtain at least t keys in the same cluster to construct a packet that cannot be detected by the forwarding nodes and Sink. In addition, NLFS only utilizes the relationships between node IDs to verify the legitimacy of nodes without other software or hardware overhead, such as complex localization algorithms and expensive positioning equipments. Theoretical analyses and simulation results show that NLFS has obvious advantages in false data filtering capability, memory overhead and tolerant of compromised nodes.