Page 32 - ITU Journal Future and evolving technologies Volume 2 (2021), Issue 6 – Wireless communication systems in beyond 5G era
P. 32
ITU Journal on Future and Evolving Technologies, Volume 2 (2021), Issue 6
interacts with and manages the speci ic assisted systems
distributed across the network. This is made possible via
Intelligent Intelligent a speci ic API [107], [108].
policy service
control deployment The recent document [109] de ines a clear list of the re‐
quirements for the ETSI ISG ENI architecture. These re‐
quirements are important to evaluate and control how
Intelligent AI works within the network and the applications to im‐
analysis NETWORK and SERVICES
and Intelligent prove network operations, management and service pro‐
prediction resource visioning. Fig. 12 also shows these requirements and how
management they are grouped. Finally, the use cases that have been
identi ied for the ETSI ENI system can be categorised into
Intelligent
monitoring ive main groups [110]: infrastructure management, net‐
work operations, service orchestration and management,
assurance, and network security. The irst considers the
Service and Network Requirements processes related to the management of the network in‐
frastructure such as maintenance, planning, and alloca‐
General requirements tion of services. The second concerns with the operations
Service orchestration and management of the network so, the extraction and analysis of the run‐
Network planning and deployment time contexts and the optimisation of management oper‐
Network optimisation
Resilience and reliability ations. The third handles the orchestration and manage‐
Security and privacy ment of orders and services, taking into account speci ic
different Service Level Agreement (SLA) of verticals. The
Functional Requirements
fourth deals with network monitoring and prediction of
Data collection and analysis future network states in order to ensure optimal main‐
Policy management tenance and continuous service delivery. Finally, the de‐
Data learning ployment of AI also targets network security.
Interworking with other systems
Mode of operations Side by side, ETSI has also been investigating a reference
Model training and iterative model for autonomic networking, cognitive networking
optimisation
API requirements and self‐management of networks and services. This spe‐
Non-Functional Requirements ic architecture is called ETSI Generic Autonomic Net‐
working Architecture (GANA), published in 2016 [111].
Performance requirements This architecture is directly inspired by the idea of SON,
Operational requirements nevertheless it provides a more general reference which
Regulatory requirements
is able to interoperate with complementary technolo‐
gies such as SDN, NFV, and big data analytics for Au‐
Fig. 12 – Circle of functionalities of the ETSI ENI system. Groups and list tonomic Management and Control (AMC). The idea of
of ETSI ENI requirements regarding how intelligence is applied to the AMC relies on the de inition of Decision‐making‐Element
network and applications to improve network operations, management (DE), which is an autonomic function that is a cognitive
and experience of service provisioning.
control‐loop in centralised/distributed management and
control planes. The DE owns self‐* features such as self‐
The scope of the ETSI ENI framework is to continuously
iguration, self‐optimisation, self‐healing, etc. Each
capture network system’s con igurations in order to tak‐
DE is an adaptive entity, which dynamically monitor and
ing actions to dynamically change the system’s character‐
manage its respective management entity. Practically, a
istics, according to the targeted objectives and KPI. As al‐
DE is placed within a network node at a speci ic layer of
ready mentioned before, while we were discussing the la‐
the protocol stack. Additionally, each DE can be either a
tency issues of 6G, ETSI ENI intelligent architecture re‐
real or a softwarized entity.
lies on big data mining and analysis, which are input for
ML to train, learn, decide, and act. According to these as‐ This brief excursus on ETSI ENI and GANA has been im‐
pects, ETSI ENI can automate complex network human‐ portant to underline the concept that the complete inte‐
dependent decision‐making processes to increase net‐ gration of AI into 6G cannot neglect the experience pro‐
work performances. Fig. 12 depicts the intelligent func‐ vided by ETSI in the last ive‐six years. A signi icant part
tionalities of the ETSI ENI architecture. By considering of research on intelligence in 6G has not even been men‐
the pure architectural characteristics, the ETSI ENI sys‐ tioned in this research. For example, the only article deal‐
tem considers the heterogeneity of the existing and fu‐ ing with AI and 6G, cited in Section 4, just brie ly mention‐
ture hardware network infrastructure together with the ing ETSI ENI is [69]. None of the others even cite either
full virtualisation obtained via the ETSI MANO SDN‐NFV ETSI ENI or GANA. However, ETSI ENI and GANA architec‐
architecture. On top of that, there is the AI of ENI, which tures will be pivotal for 6G standardization as ETSI MANO
SDN-NFV architecture has been fundamental for 5G.
20 © International Telecommunication Union, 2021