Self-Organizing Networks (SON) are collections of AI/machine learning driven functions for automatic configuration, optimization, diagnosis and healing of cellular networks. It minimizes the cost of running mobile networks by eliminating the manual configuration of network elements and troubleshooting operational problems. The ultimate goal of SON is to reduce capital expenditure (CAPEX) and operational expenditure (OPEX) which could pose a problem with the ever involving network operations.
Types of Self-Organizing Networks
There are three major SON architecture options.
Centralized SON Architecture: In centralized SON architecture, commands, requests and parameter settings data are executed at the network management level and flow to the network elements while measurement data and reports flow from the elements to the network management. The SON algorithms are thus able to take information from all parts of the network and optimize parameters. Centralized SONs are also better protected against network instabilities. Third-party SON network solutions are also possible because the functionality can be added at the network management level.
However some of the problems associated with Centralized SON are longer response times, increased backbone traffic which is as a result of the bi-directional flow of data from the network management to network elements. This in turn prolongs the time the network takes to adapt to changes.
Distributed SON Architecture: Here, commands are distributed across the edge of the network where each node exchanges SON related messages directly with other nodes. This architecture is more flexible than the Centralized SON and eliminates the problem of longer response times. It also scales better as the number of cells in the network increases.
However, the individual node optimization does not necessarily translate to optimum network operation. The implementation is also vendor-specific, making third-party solutions difficult.
Hybrid SON Architecture: This means that part of the SON algorithm is run on the network management level while another part is run on the network elements in a bid to combine the advantages of the centralized and distributed architectures. However, while the advantages are combined, the disadvantages do not cancel each other out. This means that the SON related traffic in the backbone associated with centralized SON will be proportional to the number of network elements, making it difficult to scale. Third-party solutions are also difficult to implement.
Why is SON important?
1. Automated Network Configuration: The number of network elements and parameters that need to be manually set has increased significantly from one generation of wireless network to the next. From 4G to 5G and even the theoretical 6G, these elements keep increasing and it may become impossible to keep up with manual configuration. The advancement of wireless networks from 2G to the current 5G also comes with running parallel operations on the part of the service providers. The advent of 5G does not eliminate 2G to 4G operations as they're still mainstay in many places. Thus, operators need to coordinate their network operations for increased efficiency, which is where SON comes in using the Inter-RAT (Radio Access Technology) protocol.
2. Reduced installation time and cost: With SON, there is reduced OPEX due to reductions in manual efforts in connection with protection, monitoring, optimisation, diagnosis and healing of the network. There is also reduced CAPEX due to more optimized use of network elements and spectrum.
The adoption of SON technology has improved the services of early adopters by reducing the rate of dropped calls and network congestion, savings of OPEX and increase in service revenue. It is also playing a role in the increase in the adoption of 5G networks through technologies like network slicing, dynamic spectrum management, predictive resource allocation and automated deployment of virtual 5G network functions.