Cognitive Networks: Towards Self-Aware Networks

1.1. Introduction

The development of computer networks has seen a paradigm shift from static, hierarchical network structures to highly distributed, autonomous systems without any form of centralized control. For networking nodes, the ability to self-adapt and self-organize in a changing environment has become a key issue. In conventional network structures, e.g., the Internet, there is usually a hierarchical order with centralized and static control. For example, hosts are aggregated to local area networks (LANs), which are connected via gateways to wide area networks (WANs) and network domains, etc., all using static connections and addressing. Recently, however, the trend leads more and more to networks that dynamically set up connections in an ad-hoc manner. Mobile ad-hoc networks (MANETs) are a prominent example, but also overlay structures such as peer-to-peer (P2P) networks require a scalable, robust and fully distributed operation with self-adaptive and self-organizing control mechanisms. The main control functions are no longer performed at intermediate nodes like routers, but shifted to the end-user nodes. Additionally, the location of these nodes may now be no longer static but can be mobile, imposing new challenges on the search for shared information in P2P networks or the location of a node in an ad hoc network.

For these types of new dynamic networks, the following three requirements for network control are considered mandatory:

• Expandability (or scalability): facing ...