Machine Learning Techniques: A Survey Of Machine Learning Techniques Applied To Self-organizing Cellular Networks

MACHINE LEARNING TECHNIQUES A SURVEY OF MACHINE LEARNING TECHNIQUES APPLIED TO SELFORGANIZING CELLULAR NETWORKS

CONTENTS • The Era of 5G • Introduction • Overview of Machine Learning Algorithms

• Analysis Of Machine Learning Applied in SON • Future Research Directions • Conclusions

THE ERA OF 5G • Approaches to 5G: • Massive MiMO

• mm Waves

• Problems is with • Maintenance • Optimization • Failure Detection and Recovery

• Suitable Solution

SELF ORGANIZING NETWORKS(SON’S) • Overview of Self Organizing Networks

• Machine Learning In SON

• Paper Objective and Contribution

WHY SON ? (NICE QUESTION)

• Reducing manual intervention or least possible human interaction.

• Atomization of repetitive process. • Improve runtime operation/optimization based on real time data analysis. • Improving user experience and network performance.

BASIC FEATURES OF SON

SELF CONFIGURATION • All the configuration procedures necessary in order to make the network operable. • Self Configuration Use Cases • Plug & Play : IP Address allocation, Authentication, Software and Configuration Download from OAM.

• Planning radio parameters of a new eNB: handover & cell selection thresholds, power settings, etc. • Establish connectivity with other nodes: Automatic Neighbor Relation (ANR)

SELF OPTIMIZATION • Continuous optimization of the BSs and network parameters in

order to guarantee a near optimal performance. • Self Optimization Use Cases • Backhaul Optimization

• Coverage and Capacity optimizations • Antenna parameters optimization • Interference management

• HandOver (HO) parameters optimization • Load balancing

SELF HEALING • Continuously monitor the system in order to ensure a fast and seamless recovery. • Self-diagnosis: create a model to diagnose, learning from past experiences. • Self-healing: automatically start the corrective actions to solve the problem. • Significantly reduce maintenance costs. • Self Healing Use Cases • Cell outage detection and compensation : Antenna tilt and the cell transmit power

• Self-healing of board faults

3GPP PROPOSED NETWORK

OVERVIEW OF SELF ORGANIZING NETWORKS

OVERVIEW OF MACHINE LEARNING ALGORITHMS

K-NEAREST NEIGHBOR

NEURAL NETWORKS

• Used as a Classifier • Self-Optimization • Self-Healing

DECISION TREES

SUPPORT VECTOR MACHINE

K-MEANS

SELF ORGANIZING MAPS

CONCLUSION • Reduction of human intervention. • Real time optimization of network. • Control in OPEX and CAPEX for operator.

• SON are flexible, adaptive, resilient.

THANK YOU !!!! ANY QUESTIONS ?? • References: • A Survey of Machine Learning Techniques Applied to Self-Organizing Cellular Networks Paulo Valente Klaine, Student Member, IEEE, Muhammad Ali Imran, Senior Member, IEEE, Oluwakayode Onireti, Member, IEEE, and Richard Demo Souza, Senior Member, IEEE