AIS for Misbehavior Detection in Wireless Sensor Networks: Performance and Design Principles

TL;DR

This paper evaluates the performance and design principles of Artificial Immune Systems (AIS) for detecting misbehavior in wireless sensor networks, emphasizing gene selection, interaction, and protocol considerations.

Contribution

It provides insights into effective AIS design for sensor networks, highlighting the importance of gene choice, interaction, and protocol constraints for improved misbehavior detection.

Findings

Careful application of AIS mechanisms is necessary to avoid security weaknesses.

Gene selection and interaction significantly influence AIS performance.

Data traffic patterns do not greatly affect overall detection performance.

Abstract

A sensor network is a collection of wireless devices that are able to monitor physical or environmental conditions. These devices (nodes) are expected to operate autonomously, be battery powered and have very limited computational capabilities. This makes the task of protecting a sensor network against misbehavior or possible malfunction a challenging problem. In this document we discuss performance of Artificial immune systems (AIS) when used as the mechanism for detecting misbehavior. We show that (i) mechanism of the AIS have to be carefully applied in order to avoid security weaknesses, (ii) the choice of genes and their interaction have a profound influence on the performance of the AIS, (iii) randomly created detectors do not comply with limitations imposed by communications protocols and (iv) the data traffic pattern seems not to impact significantly the overall performance. We identified a specific MAC layer based gene that showed to be especially useful for detection; genes measure a network's performance from a node's viewpoint. Furthermore, we identified an interesting complementarity property of genes; this property exploits the local nature of sensor networks and moves the burden of excessive communication from normally behaving nodes to misbehaving nodes. These results have a direct impact on the design of AIS for sensor networks and on engineering of sensor networks.