Page 165 - Proceedings of the 2018 ITU Kaleidoscope

Session 2: Artificial Intelligence and 5G
S2.1 Self-Healing and Resilience in Future 5G Cognitive Autonomous Networks
Janne Ali-Tolppa (Nokia Bell Labs, Germany); Szilárd Kocsis, Benedek Schultz and Levente

Bodrog (Nokia Bell Labs, Hungary); Márton Kajó (Technical University of Munich, Germany)

In the Self-Organizing Networks (SON) concept, self-healing functions are used to detect,
diagnose and correct degraded states in the managed network functions or other resources. Such
methods are increasingly important in future network deployments, since ultra-high reliability is
one of the key requirements for the future 5G mobile networks, e.g. in critical machine-type
communication. In this paper, we discuss the considerations for improving the resiliency of future
cognitive autonomous mobile networks. In particular, we present an automated anomaly detection
and diagnosis function for SON self-healing based on multi-dimensional statistical methods, case-
based reasoning and active learning techniques. Insights from both the human expert and
sophisticated machine learning methods are combined in an iterative way. Additionally, we present
how a more holistic view on mobile network self-healing can improve its performance.

S2.2 AI as a Microservice (AIMS) over 5G Networks
Gyu Myoung Lee (Liverpool John Moores University, United Kingdom); Tai-Won Um (Chosun
University, Rep. of Korea); Jun Kyun Choi (Korea Advanced Institute of Science & Technology,
Rep. of Korea)

As data-driven decision-making services are being infused into Internet of Things (IoT)
applications, especially at the 5G networks, Artificial Intelligence (AI) algorithms such as deep
learning, reinforcement learning, etc. are being deployed as monolithic application services for
autonomous decision processes based on data from IoT devices. However, for latency sensitive
IoT applications such as health-monitoring or emergency-response applications, it is inefficient to
transmit data to the Cloud data centers for storage and AI based processing. In this article, 5G
integrated architecture for intelligent IoT based on the concepts of AI as a microservice (AIMS)

is presented. The architecture has been conceived to support the design and development of AI
microservices, which can be deployed on federated and integrated 5G networks slices to provide
autonomous units of intelligence at the Edge of Things, as opposed to the current monolithic IoT-
Cloud services. The proposed 5G based AI system is envisioned as a platform for effective
deployment of scalable, robust, and intelligent cross-border IoT applications to provide improved
quality of experience in scenarios where realtime processing, ultra-low latency and intelligence
are key requirements. Finally, we highlight some challenges to give future research directions.

S2.3 Multifractal Modeling of the Radio Electric Spectrum Applied in Cognitive Radio Networks
Luis Tuberquia-David and Cesar Hernández (Universidad Distrital Francisco Jose de Caldas,
Colombia)

The work discussed in this article is framed within the context of cognitive networks in America,
showcasing the scenario of the radioelectric spectrum of the city of Bogotá, Colombia. The
objective is to model the traffic of the wireless network, since it is underused in this region of Latin
America. Hence, some tools are studied to allow the structuring of the type of traffic seen in the
network. Based on stochastic tools such as the log-scale diagram, the linear multiscale diagram,
and the multifractal spectrum, this research aims to verify the multifractality of traffic series

collected on the electric radio spectrum of Bogotá, Colombia in 2012. In fact, the study reveals
that all the channels of the network have a multifractal behavior with 90% of them presenting a
Hurst parameter in the 0.5 to 1 range. The evidence suggests that the traffic in this region could be
modeled as multifractal time series. Therefore, the analysis carried out intends to provide a new
modeling method for the Colombian radioelectric spectrum in the form of a multifractal-based
analysis.

– 149 –