A self-organizing network (SON) is a framework for automating planning functions in radio access networks (RAN) and mobile networks. The goals of a SON are to enable self-configuration, self-optimization, and self-healing of radio network elements. It also can minimize interference between adjacent cells and between macro and small cells. SON was initiated by the 3GPP for the automated operation and maintenance of LTE networks. Specifically, it was designated in 3GPP's Release 8 and Release 9, with a later Release 10 defining more features of SON for enhanced and upcoming networks.
Originally developed as an operational approach to streamline cellular RAN deployment and optimization, SON has been increasingly focused on integrating new capabilities, such as self-protection against digital security threats and self-learning through artificial intelligence and machine learning. This allowed the scope of SON to extend beyond RAN and include mobile core and transport network segments. This extension allows SON to address requirements for 5G networks, such as end-to-end network slicing. Similar solutions have also emerged for WiFi, where SON can simplify wireless networking in home and enterprise environments.
Although vendors often provide their SON components, most networks have multi-technology, multi-vendor deployments that require an Independent Software Vendor (ISV) to ensure an optimized cross-network state and simplified control. In the case of self-organizing networks, there are three types of architectures or deployments, dependent on the user's approach to SON. These included distributed SON, centralized SON, and Hybrid SON.
A distributed SON architecture is where numerous automation components are across the network, which usually has the main advantage of using real-time data for near-real-time actions or near-immediate results. However, as many distributed SON components are provided by various equipment vendors, they can be difficult to integrate and collaborate across technologies to achieve full network optimization. Third-party solutions can also be difficult to integrate with a SON algorithm, which can cause an algorithm to execute on network elements, but not actually occur.
In a centralized SON architecture, algorithms are executed at a network management level, meaning commands, requests, or network settings come from a network management level to the network elements, with data and reporting flowing in the opposite direction. The benefit of this approach is the SON algorithm can take information from an entire network into consideration, making it possible to optimize parameters of centralized SON functions to become more globally optimized for varying network characteristics. Further, this approach offers users a chance to add functionality for a multi-vendor or third-party SON solution on the network management level. Drawbacks of this network architecture include longer response times, increased backbone traffic, and a single point of failure. The longer response times include network instabilities, and the two-way traffic increases the network traffic, which can increase again when more cells are added to the network.
Hybrid SON, as the name suggests, is an attempt to combine the advantages of centralized and distributed SON solutions—the centralized coordination of functions and the ability to respond quickly at the network element level. This solution runs part of the algorithm at the network management level and the other part on the network elements. But this also brings the drawbacks of both centralized and distributed SON, with the related traffic bottlenecks through scaling networks and the related processing power on the network. Also, since the SON algorithms run on network elements, the interface between centralized and distributed SON functions will need to be proprietary; therefore, third-party solutions will be complicated.
Following the CGPP releases, including Release 8, Release 9, and Release 10, the features for the SON network as intended to be deployed are laid out. The last release further details use cases for new networks and enhancements to existing use cases, covering macro and metro networks, as well as interoperability with existing cellular networks.
SON features by release
SON offers a variety of functionality for networks, including self-configuration, self-optimization, self-healing, and self-protection, which are made possible through artificial intelligence, predictive analytics, and pre-optimized software algorithms. These functions strive to make complicated network administrations simple, offering a plug-and-play environment for simple and complex network tasks and reducing the necessary administration for the network.
This means the SON automatically recognizes and registers new access points and base stations as part of the radio network. Neighboring radios automatically adjust their emissions and other technical parameters to avoid interference and maximize coverage and capacity.
As the name suggests, this functionality allows the network to optimize base stations' technical parameters, based on purpose. For example, a self-optimizing network can optimize airtime resources for service level agreements (SLA) per device, while application groups can be maintained regardless of congestion, high device density, and changing spectrum availability.
Self-healing allows SON to heal itself automatically when a base station fails or connectivity is lost. Self-healing networks adjust adjacent cells' parameters to provide continued service or minimize the degradation of service to affected users.
As the name suggests, a SON is capable of automatically defending itself from penetration by unauthorized users, with the primary goal of maintaining the security of the network and the confidentiality of data sent over that network.
Automatic neighbor relations for SON help facilitate smooth signal transitions from cell to cell as a device moves through a network. This has, previously, been a complicated or laborious task for human operators but can be handled automatically through SONs. ANR works to analyze and communicate with neighboring cells to ensure handovers are seamless.
The automation of SON is often its main benefit as it reduces the need for manual, human attention for installation and network management. This can reduce the cost of maintaining a network and offer a network that is quicker to install and easier to maintain while offering better performance, only because it offers less service interruption. Further, SONs can help enhance networks in many ways:
- Improving network performance
- Reducing network downtime
- Increasing user experience over private cellular networks
- Reducing overall capital expenditure
- Improving IT staff efficiency
Since SONs runs on algorithms and artificial intelligence, rather than scripts, the self-healing and performance management of a cellular network can be without human intervention while also responding to an ever-changing network environment, rather than automating fixes or solutions with premade scripts capable of responding to specific problems. This flexibility allows SONs to take account of all elements, or all necessary elements, of a network before applying a change or setting a configuration.
Self-organizing network vendors
With the evolution of telecommunication infrastructures and networks towards 5G, the proposition for a SON, or a further move away from traditional and static processes toward the dynamic and automatic processes offered by SON have been suggested, increasing the overall speed of the networks and providing self-adaption capabilities for mobile networks. For 5G networks, SON offers use cases and problem-solving pertinent to emerging, associated technologies, including spectrum management and sharing, user association, multi-radio access technology optimization, and direction cell search for millimeter wave (mmWave) networks.
This is partially enabled by the machine learning included in SONs that are key for enhancing the operations, administration, and maintenance (OAM) activities. It can also assist in advancing new cellular use cases, such as ultra-reliable low latency communications for the support of combinations of frequencies. SONs are included in the 5G Standalone (SA) system, which is considered a key enabler as it can improve the efficacy of the throughput at the edge of the network. Key drivers for the deployment of SON for 5G and future generation networks include the following:
- The network parameter design, tuning, and management lead to high complexity in terms of the coordination among existing networks and complex architectures. Network management in this case has been considered unfeasible without the use of automated SON functions operating in real-time and taking feedback from the network.
- Standardization of network traffic characteristics and parameters to support basic 5G use cases can be embedded in SON systems and applications, including target setting, such as low latency and high reliability for the network.
- The AI and ML in SON can benefit operation of networks and offer platform advancements and network automation for complex implementations.
- 5G use cases and intelligent techniques, such as C-RAN, mmWave, cognitive radio, network slicing, spectrum sharing, automated backhauling, and massive MIMO, could be combined with SON platforms to increase network performance.
More of a proposal than a possibility, it has already been suggested that, given an introduction of self-coordination capabilities for a SON platform, SONs could be used for continued advancements in network technology, both keeping SON platforms relevant for networks, and continuing to increase the automation in network environments and increase the efficiencies and the speed of the networks, providing a more seamless user experience. Further, in new networks, they could be used as hybrid coupled SONs, which could provide greater standardization across larger networks and reduce network weaknesses.