|
Schemes | Advantages | Disadvantages |
|
Detection scheme based on watermarking technology [10] | Effectively detect whether the data is discarded or tampered | The time delay of watermark extraction is large |
Detection scheme of sensor network malicious nodes based on double threshold [11] | Reduce the false alarm rate and improve the accuracy of event area detection | The problem of threshold selection is the difficulty of this method |
Adaptive security mechanism for on-demand access control [12] | Fully consider the security requirements of each node | May cause malicious nodes to launch malicious attacks using cuts in network security measures |
A dynamic trust evaluation method for Internet of Things nodes [13] | Effectively reduce the influence of malicious recommendation behavior on trust evaluation | Complex implementation process and large amount of computation |
Trust evaluation scheme based on data mining model in SIoT [16] | Comprehensively measure the value contribution of trust value assessment | The definition of social boundary is vague and cannot accurately calculate social familiarity |
Perceived network security connection model [17] | Describe the changes of trust in various states from a sociological perspective | Without combining subjective and objective, the accuracy of node trust assessment is deficient |
Dynamic trust management system [18] | To defend against external attacks but also to defend against internal compromise attacks | High memory and energy cost, high computational complexity |
Trust management model of Internet of Things based on social relations [19, 20] | Improve the reliability of recommendation trust evaluation and enhance the ability of model against malicious recommendation behavior | Cannot accurately reflect node behavior |
Trust-aware network security connection model [21] | Extending the definition of trust and realizing the function of data collection and analysis from multiple data sources | Lack of dynamic adaptability |
Sensory network security connection model based on distributed adaptive filtering [22] | Good environmental adaptability, fully considering the computing ability of sensing nodes | Lack of feedback control to nodes, unable to resist malicious attacks perfectly |
Intelligent Beta reputation and dynamic trust evaluation model [23, 24] | Reduce the energy consumption of trust computation | The convergence rate of the model is reduced, and the malicious nodes cannot be identified quickly |
|