Cognitive Networks: Towards Self-Aware Networks

Introduction

Motivation

As the Internet moves well beyond the classical services of email, file transfer and remote login, and as the explosion of Internet portable devices continues, the applications and services needed to support mobility will have different network requirements. This has motivated research on active and programmable networks, on a large scale in the 1990s, which represents a novel approach to network architecture in which the switches perform customized computation on the messages flowing through them ^[]. The objective was to make the network adaptive and programmable, but despite the many research projects, networks still have to be managed by human administrators. This is simply because the network is still not aware of its state or needs, and doesn't have knowledge of its goals and how to achieve them, and it is not able to reason for its actions ^[]. Such properties would render the network adaptive and self-governed. Current data networking technology perform reactive adaptation by responding to changes in the environment after a problem has occurred ^[]. Instead, adaptation should be proactive by actively affecting the network when the environment changes.

Mobile computing is becoming pervasive but in order to reach its full potential, significant improvements are needed in usability of end users and manageability of administrators ^[], which can be realized by incorporating intelligence into the network environment, and autonomic networking is the key ...