Mobile network optimization specifically Radio network optimization is a very dedicated and specialized activity which takes care of maintaining utilization level, configuration, optimization and healing of radio resources. Moreover, it has a direct impact on the CAPEX and OPEX of a telco network. Historically mobile network optimization was executed manually. It became a herculean job for optimization engineer to optimize the radio network to meet the end user requirement as well as reduce CAPEX and OPEX due to various factors. These factors include parallel operation of different networks (e.g., 2G, 3G, 4G, Circuit switched voice call, data call, VoLTE, VoWifi), increased number of parameters that must be monitored, and complexity of call flows. Self-Organizing Network (SON) facilitated to a great extent optimization of the network in autonomous mode to achieve a cost-efficient and well-performing network.

SON with its in-built capability of performance monitoring, autonomous optimization and configuration as well as healing facility took a significant role to plan, operate and optimize the radio network and helped operator to obtain substantial financial benefits.

Existing SON solution typically follow one of the three architecture models mentioned below

Architecture Model Advantages Disadvantages
Centralized Solution

All decision-making capabilities and processes are performed at the central Management layer.

 ·   End to end network optimization is possible.

·   It facilitates Vendor independent third-party SON solution since processes are executed at management layer and not at network element layers which need vendor specific solution.

·   Conflict Management is comparatively easy since resolutions are executed only at management layer.

 ·   Massive dataflow from site to NOC which need additional infrastructure and backbone investment

·   Since all the resolutions are getting executed at central level, it will introduce longer response time and substantial latency within the system.

·   With increase of number of sites, actual advantage of SON is very hard to achieve.
Distributed Solution

Decision-making capabilities are with network element and network elements communicate with each other directly to execute the resolutions

 ·   Very dynamic in nature and very quick response time since resolutions are being executed within network elements.

·   No additional infrastructure or backbone is required to adapt this solution.

·   Scaling of the solution is one of the major advantages.

 ·   E2E network optimization is not possible since no visibility of overall network while executing the decision.

·   Vendor independent third- party solution is not feasible since decisions are executed within network elements.

·   Conflict management is very difficult.
Hybrid Solution

It’s a mixed solution of Centralized & Distributed SON where part of the resolutions is getting executed at central location and remaining at the network elements.

 ·   Comparatively better response time than Centralized SON.

·   Scaling of this solution is moderate compared to distributed solution.

 ·   This solution is not very clearly defined among vendors.

·   Vendor independent third- party solution will be very complicated, practically not feasible since some resolutions are executed within network elements.

·   Disadvantages of both Centralized and Distributed solution are genetic.

Considering the above pros and cons, we can conclude that existing SON solutions has the following challenges based on the deployment architecture:

  • Vendor specific proprietary solution.
  • Lack of scalability.
  • Additional Infrastructure and Backbone requirement.
  • Slower response time and substantial latency within the system.

Operators are still adopting Centralized SON solution to optimize their radio network resources conceding some degree of overall system latency and absorbing additional infrastructure and backbone cost.

Many operators across the globe have now started to evaluate and adopt a different way to operate & optimize their radio network leveraging next generation technologies. These include Cloud Native Architecture, Open API based interfaces, and disaggregated network architecture. A new approach is needed to deal with gigantic traffic growth which has resulted in more network elements getting integrated in the network, complexity of multi-layer networks and call flows. O-RAN has come up with the new concept of RAN Intelligent Controller (RIC) for RAN programmability and for SON to run as an external application.

Empowered by AI/ML and open communication, the below mentioned O-RAN architecture is based on standards defined by O-RAN ALLIANCE. This architecture is fully supporting and complementary to 3GPP and other industry standards.

O-RAN logical architecture. Image used courtesy of O-RAN ALLIANCE

RAN Intelligent Controller (RIC) along with its application xApps is a new virtualized function which enables RAN programmability to existing or new RAN networks.  It uses open-source tools with open interfaces towards RAN (E2 interface) and can be deployed either centralized in the cloud or distributed at the cell site. It is also connected with Service Management and Orchestration (SMO) layer through A1 and O1 interfaces. Some major features and functionality of RIC are as mentioned below:

  • RIC can host different application for different functionality. RIC applications consist of one or more micro-services and are assigned with the database as per designated function during on-boarding.
  • RIC applications are totally independent of RIC, any third-party application can be on-boarded.
  • The Near-Real Time (RT) RIC delivers a robust, secure, and scalable platform that allows for the flexible on-boarding of third-party control applications.
  • RIC will use these applications to provide SON functions for Near-RT Application, conflict mitigation, security services and management services.
  • The O-RAN architecture provides next generation Radio Resource Management (RRM) with its in-built intelligence powered by very strong AI/ML. Moreover Near-RT RIC is completely compatible with legacy RRM such as per-UE controlled load-balancing, radio bearer management, interference detection, and mitigation.
  • RIC use the interface towards SMO for Non-RT Application. SMO makes RIC more powerful, scalable through its Non-RT RIC functions which includes service and policy management, RAN analytics, and model-training for the Near-RT RAN functionality. The key capabilities of the SMO are FCAPS interface to O-RAN Network Functions, Non-RT RIC for RAN optimization and O-Cloud Management, Orchestration and Workflow Management.
  • Trained models and real-time control functions produced in the Non-RT RIC are distributed to the Near-RT RIC for runtime execution.
  • The Non-RT RIC monitors long-term trends and provides QoS or QoE policy guidance to the Near-RT RIC. Empowered with ML techniques and RAN data reported from E2 nodes, the Near-RT RIC can dynamically predict the QoS or QoE performance of the application and then change the RAN behaviour to assure the user experience.

Considering above features and functionalities of RIC assisted by SMO functions, we can conclude that RIC can address most of the shortfalls of existing SON functions across networks and can autonomously operate and optimize the radio network to reduce CAPEX and OPEX of the operators. The following table summarizes how RIC addresses some of the challenges with existing SON solutions.

Issues with existing SON Solution Approach to address the same in RIC
Vendor specific proprietary solution. Due to its Open Architecture, RIC is vendor agnostic which will enable technology agility and grants operators’ freedom from “vendor lock-in”.
Lack of scalability. Since real time and non-real time functions are intelligently distributed between Near-RT RIC and SMO, we can effortlessly scale the overall solution based on the operator’s requirement.
Additional Infrastructure and Backbone requirement. Based on network requirement, we can deploy the solution at the network edge which will avoid additional investment in infrastructure and backbone capacity.
Slower response time and substantial latency within the system. RIC will provide very quick response time and overall solution will be very dynamic in nature since decisions are getting executed close to the network elements.

References: