| The resource-constrained sensors in mission-critical applications are subject to both
random failures and intentional compromise by the adversary, which poses severe
security threats in wireless sensor networks (WSNs). The different types of security
threats have been identified and addressed in an individual manner in the past. And
most solutions are cryptography based. In this paper, we argue that cryptography alone
is insufficient to fully address the insider attacks in the existence of both the
compromised and faulty sensor nodes. We further propose a proactive data security
framework (PDSF) to identify compromised and faulty nodes proactively and prohibits
them from participating network activities in a dynamic manner. The rationale behind
our approach is that a sensor's future behavior can be predicted (at least)
probabilistically by its past behavior. PDSF is divided into two key modules, that is,
misbehavior characterization \& monitoring, and trust management. The former
characterizes different types of misbehavior in WSNs and defines a set of monitoring
criteria. And the latter develops a trust management model, which adapts itself to the
resource constrained and application specific nature of the WSNs. |