Page 72 - Proceedings of the 2018 ITU Kaleidoscope

TOWARDS COGNITIVE AUTONOMOUS NETWORKS IN 5G

Stephen S. Mwanje and Christian Mannweiler

Nokia Bell Labs, Germany

ABSTRACT

Cell densification and addition of new Radio Access Tech-
nologies have been the solutions of choice for improving
area-spectral efficiency to serve the ever-growing traffic
demand. Both solutions, however, increase the cost and com-
plexity of network operations for which the agreed solution
is increased automation. Cognitive Autonomous Networks
(CAN) will therefore use Artificial Intelligence and Machine
Learning (ML) to maximize the value of automation. This
paper develops the models for cognitive automation and
proposes a CAN design that addresses the requirements for
5G and future networks. We then illustrate the benefit of this
approach by evaluating ML models that learn a network’s
response to different mobility states and configurations.

Keywords – Cognitive Autonomous Network, Network
Management Automation, 5G Figure 1: BSS and OSS deliver value from the network,
but Network Management maximizes automation value.
1. INTRODUCTION network through the OSS. Network Management (NM) as a

part of the OSS (see Figure 1) is responsible for the
Demand for mobile communication services has grown knowledge about, control of, and (re-)configuration of the
unabatedly for at least two decades, recently driven by network’s devices to match operational objectives. It carries
mobile Internet connectivity and the related broadband ser- the biggest burden of network complexity as it defines which
vices, especially video. The solutions hereto have been two- network devices, at what locations and for which customers
fold: (1) to deploy new Radio Access Technologies (RATs) should be used in which way and to which degree.
2
intended to improve area-spectral efficiency (in bits/Hz/m ) Prioritizing the automation of NM processes will as such
amidst ever higher spectrum demand; and (2) densification maximize the impact of the CSP’s overall automation efforts.
in each RAT, by deploying ever more cells to meet users’ This includes processes in network configuration and
Quality of Service (QoS) needs at all locations and times. So operation optimization and healing as described in the Fault,
mobile networks are characterized by heterogeneity and high Configuration, Accounting, Performance & Security
Base Station densities, which directly translate into high Management (FCAPS) framework of the International
Capital Expenditures (CapEx) and Operational Expenditures Telecommunication Union’s Telecommunication
(OpEx) as well as high complexity of network design and Standardization Sector recommendation M.3400 [1].
operation. 5G, which at the least adds another radio layer,
will further complicate these networks. Yet, although net- Today, a degree of automation has been achieved through
works can be very complicated, the individual devices need Self-Organizing Networks (SON) [2]. SON is the first
to be rather simple and cost-efficient to be scalable to global- generation of NM automation, where SON Functions (SFs),
scale networks. Thus, the complexity is transferred to the realized as closed-loop control systems, address specific NM
network operability layers which must, at all times, retain the problems like balancing load among cells [2]. The SFs
broader view across the network (see Figure 1). exhibit static, rule-based behavior that maps observed
network states, e.g., changes in Key Performance Indicators
The major solution to these challenges is automation, espe- (KPIs), to (re-)configurations of individual Network Config-
cially automation of network management. For a given uration Parameters (NCPs) or entire network configurations.
network, Business Support Systems (BSS) and Operations Since the multitude of SFs need to be coordinated, their
Support Systems (OSS) ensure to deliver value from the net- management and coordination is performed in a hierarchical
work, both to the customer through BSS applications and to manner according to rather fixed rules, or through policies
the Communication Services Provider (CSP) operating the

978-92-61-26921-0/CFP1868P-ART @ 2018 ITU – 56 – Kaleidoscope