Towards Cognitive Autonomous Networks

8Cognitive Autonomy for Network‐Optimization

Stephen S. Mwanje¹, Mohammad Naseer Ul‐Islam¹, and Qi Liao²

¹Nokia Bell Labs, Munich, Germany

²Nokia Bell Labs, Stuttgart, Germany

Optimization is the act, process, or methodology of making something (such as a design, system, or decision) as fully perfect, functional, or as effective as possible [1]. In Mathematics, it refers to the techniques for finding a local or global maximum or minimum value of a function of several variables subject to a set of – constraints. Both these definitions apply to networks where, after the network (or an entity thereof) is configured and commissioned into operation, further actions may be needed to ensure the most effective use of that network's resources. Reasons for this may include:

1. – the availability of real operating conditions which may be different from the models used in planning and configuration
2. – changes in the operating points requiring reconfiguration of network parameters for the new conditions (including changes in traffic behaviour, propagation conditions due to new buildings, weather changes, or new deployments).

Self‐optimization (SO) implies that the entity under optimization undertakes the optimization process or methodology by itself without being triggered or controlled by another external entity. In networks, this implies that the network triggers, runs and controls the optimization algorithms without any intervention from the human operator(s). Note that there is no strict ...