Mobile operators around the world are facing major challenges that cannot be tackled with conventional network management solutions. The existing network technologies that are being used by operators almost reached their theoretical limits. In order to deliver higher capacity and better coverage while maintaining the customer experience, operators are forced to invest further into their networks. In order to deal with this challenge, visionary mobile operators are investing in network automation technologies, so that they can manage costs and maintain quality by utilizing their network infrastructure more efficiently.
However, once 5G is introduced, today’s automation techniques will not be adequate to manage the highly complex and the sophisticated network infrastructure as well as the fast-growing subscriber base. So, the proposed automation solutions should be self-learning, adaptive and proactive to meet the evolving network and customer requirements. As such, AI becomes a highly critical component of the next generation network management platforms.
P.I. Works’ Focus on AI and the Recent IEEE Research Project
P.I. Works has already started incorporating certain components of AI into its advanced network automation technology. Machine learning, a subset of AI, is one of the key research topics for P.I. Works and is a strategic element of the product roadmap.
As a part of these efforts, P.I. Works recently contributed to a research study (Mobility-Aware Cell Clustering Mechanism for Self-Organizing Networks) that proposes a novel fusion of Self-Organizing Networks (SON) and AI. SON is the 3GPP defined standard for the automated management of the routine network management activities. The research paper explains the benefits of the mobility-aware clustering mechanism (a subset of machine learning) and its impact on the conventional SON functions.
The mobility-aware clustering mechanism basically groups the cells in the network based on their relational activities; hence the cells with higher mobility activities are grouped under a single cluster. This method is an upgrade to the conventional distance-based clustering algorithms. It detects and groups the cells based on the frequency of their interaction with each other, rather than looking at the physical cell distances only.
The method boosts the accuracy and efficiency of the conventional SON functions. This means improvement at every level in the network KPIs and KQIs. The SON modules that can benefit from the clustering mechanism are the ones with direct impact on mobility. PCI, CODC, ANR, ACO, MLB as well as some LTE-advanced / 5G functions (e.g., CA, M-MIMO) are examples to the modules that benefit from an advanced clustering mechanism.
This is an unconventional approach, yet an effective one for managing complex networks with millions of handover relations and different coverage and capacity requirements. It improves the efficiency of network optimization by analyzing a smaller number of cells rather than the entire network. More details related to this approach can be found in the IEEE research paper published on www.ieee.org.
AI and machine learning will be instrumental in managing highly complex fast growing 5G networks that demand seamless low-latency and ultra-high-speed mobile services. P.I. Works has been working closely with prominent mobile operators around the world and has been helping them manage complexity more effectively on the path to 5G. For more information please contact firstname.lastname@example.org.
Sanayi Mah. Teknopark Bul.
No:1/3A 103 34906 Pendik/Istanbul
T: + 90 216 999 1099
F: + 90 216 912 1099