A framework for understanding network intelligence

The term network intelligence is widely used in the field of communication networks. A number of new and potentially limiting concepts and products based on the concept of network intelligence have been introduced, including smart flows, intelligent routing, and intelligent web switching. Many intelligent systems focus on a specific service, function, or device, and do not provide true end-to-end network intelligence. True network intelligence requires more than a set of disconnected elements; it requires an interconnecting and functionally coupled architecture that enables the various functional levels to interact and communicate with each other. We propose a uniform framework for understanding end-to-end communication network intelligence (CNI), which is defined as the ability of a network to act appropriately in a changing environment. We consider an appropriate action to be one that increases the optimal and efficient use of network resources in delivering services, and we define success as the achievement of behavior subgoals that support the service provider's ultimate goals, which are defined external to the network system. The framework presented incorporates the functional elements of intelligence into computational modules and interconnects the modules into networks and hierarchies that have spatial, logical, and temporal properties. Based on the framework proposed, we describe an end-to-end multiservice network application spanning the network management layer, optical layer, switching/routing layer, services layer, and other layers.