Danger Is Ubiquitous: Detecting Malicious Activities in Sensor Networks Using the Dendritic Cell Algorithm

Abstract

Abstract. There is a list of unique immune features that are currently absent from the existing artificial immune systems and other intelligent paradigms. We argue that some of AIS features can be inherent in an application itself, and thus this type of application would be a more appropriate substrate in which to develop and integrate the benefits brought by AIS. We claim here that sensor networks are such an application area, in which the ideas from AIS can be readily applied. The objective of this paper is to illustrate how closely a Danger Theory based AIS -in particular the Dendritic Cell Algorithm matches the structure and functional requirements of sensor networks. This paper also introduces a new sensor network attack called an Interest Cache Poisoning Attack and discusses how the DCA can be applied to detect this attack.

Towards Understanding the Requirements and Limitations of Reputation-Based Systems

“…In general, such work aims to augment traditional security mechanisms with support for reactivity by using behavioural analysis techniques and ranges from intrusion detection systems [19], through to reputation-management. In the latter, the aim has been to provide two basic functions: (i) the capacity to learn (on-line) from and adapt to the behaviour of other nodes and (ii) to leverage the experiences of others, so that learning can be bootstrapped and a more accurate estimate of the behaviour of nodes under observation can be built.…”

Frequency analysis for dendritic cell population tuning

“…Other application areas include fault detection within sensor networks [12] and threat classification within the field of robotic security [15]. All of these areas feature problem environments where there is a great deal of uncertainty associated with the information relevant to making the decision.…”

The Application of a Dendritic Cell Algorithm to a Robotic Classifier

“…Thus far, the majority of problems presented to the DCA are related to computer security, specifically the detection of port scans [3] it is also applied to a static machine learning dataset [10] and the detection of intrusions in sensor networks [13]. Work performed by Greensmith et al [12] has indicated that the DCA performs well for time-dependent real-valued data, such as that seen in robotics applications.…”