Mobile Intelligence

26.1 INTRODUCTION

In this chapter, we model distributed denial of service attacks in wireless sensor networks and define a centralized neural network-based scheme for detection of such attacks. The scheme involves the training and subsequent clustering of patterns of network traffic flow for attack classification purposes.

Wireless sensor networks (WSNs) consist of a collection of hundreds to thousands of tiny devices called sensors or sensor nodes. The limited on-board memory resources of such tiny devices restricts the size of applications, program codes, and actual data that can be stored in their memory. Therefore, most applications and programs designed for high-performance computing devices cannot be accommodated unaltered into the small memory space of sensor nodes.

The on-chip processing capability of a typical sensor node is several orders of magnitude less than that of a standard desktop processor. These networks generally have a rooted topology with a computing device called the base station at the root of the network. The base station has several orders of magnitude more power and a longer lifetime as compared to a standard sensor node [18]. Moreover, the base station has larger storage capacity and higher communication bandwidth links. Operations of the base station include information dissemination, network initialization, node activation and revocation tasks, and interfacing with other WSNs.

Sensor nodes are generally deployed in harsh and inaccessible environments ...