Open Self-Organizing Network: a continuous development for Radio Access Network performance optimization

Open  Self-Organizing Network:  a continuous development for Radio Access Network performance optimization
 

Open Self-Organizing Network: a continuous development for Radio Access Network performance optimization

 

TIM RAN digitalization Path

The evolution of radio access requires a new approach for network creation and management, assuring quality-of-experience to the customer, even with increasing traffic and network complexity, and sustainable costs for the operator.

Digitalization and automation are key elements to address this challenge.

TIM is engaged in a transformation process which, applied in the radio access domain, involves the introduction of open and cloud software architectures that enable the automation of network functions.

Two main initiatives were exploited in the last few years:

  • Starting in 2015, TIM defined an “Open SON” reference architecture based on API (Application Programming Interface) availability for the exchange of information among the involved functional blocks [1];
  • Starting in 2016, TIM, as a worldwide pioneer, tested and deployed Virtual RAN (vRAN) with several partners [2] [3].

The “digitalization” (i.e. automation and softwarization) path of TIM is represented in Figure 1

 

Figura 1 - TIM RAN digitalization path

It is worth noticing how digitalization proceeds “from the center to the edge”, i.e. from the centralized NMS (Network Management Systems) going down to the radio access sites, leveraging on cloud infrastructure and operators’ software development and integration capabilities.

Open SON represents a first automation step: starting from already existing design and optimization tools, an automation framework was built by:

  • Introduction of centralized SON (cSON) software modules, integrated with existing RAN architectures (and existing distributed SON features - dSON) and providing API for “closed loop” interaction with TIM design and optimization tools;
  • Usage of TIM cloud infrastructure for onboarding both centralized SON and design/optimization tools. In particular, NFV-I (Network Function Virtualization Infrastructure) was exploited.

With RAN virtualization (second step), part of the baseband components of the BBUs (Base Band Units) nodes become virtualized functions integrated on NFV as well (indicated as “Centralized Units”), while lowest protocol layers (e.g. physical level) still exploit physical dedicated HW, together with the RRUs (Remote Radio Units), installed next to the antennas. Network Management Systems, on the other hand, evolve to enable orchestration and workflow management within RAN domain.

The digitalization path completes with the third step, aligned with O-RAN (Open RAN) architecture [4]:

  • Virtualization of the whole base-band protocols on NFV (Network Function Virtualization), including physical layer. Baseband is split into Centralized units (CU) and Distributed units (DU).  The only physical component is the remote Unit implementing Radio frequency (RF) transmission and reception;
  • Architecture with open interfaces, from the fronthauling (i.e. interconnection between RU and DU) and the mid-hauling, up to the management interfaces, thus enabling the growth of “an ecosystem of innovative new products that will form the underpinnings of the multi-vendor, interoperable, autonomous RAN” [4];
  • RAN intelligent controller introduction, that can be considered as an evolution of Centralized SON module in previous TIM architectures, providing Open APIs towards higher level systems (Management & Orchestration according to ONAP architectures) [5].

The “cloudification” of RAN protocols down to the physical layer implies the availability of distributed data centers, designed taking into account band/latency requirements of fronthauling interconnections. This requirement can be managed in synergy with Edge Computing, also leading to a reduction of the distance between the network computational capability and the customer.  For such a reason, RAN virtualization and Edge Computing can be seen as main drivers for distributed data centers realization.

Open RAN introduction is the lever to transform Network Management systems and processes, by fully exploiting open environment and HW/SW decoupling. Automation based on the evolution of SON paradigm is even more crucial in a multi-vendor environment that will require enhanced control by the Operator. In this context the capability of “in house development” of SW modules by the operator can be a crucial asset, enabling also introduction of advanced “machine learning” capabilities, exploiting an “augmented intelligence” approach (see [6] for details)

 

The Open SON architecture

The SON (Self-Organizing Network) concept was introduced by 3GPP since the initial releases of LTE standardization, in order to help mobile access network operators to cope with the increasing complexity of configuration, optimization and assurance processes.

TIM has played an active role in 3GPP SON standardization activities since their beginning, combining the innovation vision with the “in field” experience in 2G/3G/LTE radio access design and optimization [7]. The work of TIM has been mainly focused in promoting open, interoperable and flexible solutions.  

Based on this experience, in the last years TIM defined the “Open SON” approach, including “in house” software development, as a key initiative in the path toward virtualization and automation of the radio access network.

The “Open SON” architecture can be applied to both the legacy 2G/3G/LTE radio access networks and the evolution towards the so called “Virtual RAN”, based on the NFV  paradigm within the radio access domain.

The “Open SON” approach enables two different “closed loop” optimization processes:

  • “Automatic Closed Loop” addressing basic optimization and configuration activities, for which a complete automation can be envisaged;
  • “Human Closed Loop” addressing more complex activities, where TIM radio access specialists are supported by software applications (mainly developed “in house”), enabling also an effective control of automatic functions (through API).

The architecture defined by TIM is summarized Figure 2.

 

Figura 2 - TIM RAN Automation Architecture

Centralized SON Solution was specified as an open platform: modular, flexible, programmable (through API). Baseline platform was integrated (through NFV) in 2017, with the first use cases. Since then, new use cases can be easily introduced exploiting Open API and leveraging on development of radio access design and optimization tool in TIM domain. In particular:

  • TIMplan server, described in [6] providing radio design data;
  • IRMA: implementing configuration policies (macro-optimization);
  • TIMqual: providing micro-optimization (i.e. cell-by-cell optimization) and overall process orchestration.

The interworking between design, optimization and configuration tools is a crucial aspect of the proposed architecture, due to the intrinsic characteristics of Radio Access management: each new site leads to changes in mobility management, traffic distribution and interference conditions also for the neighbouring sites. In this context , the exploitation of accurate coverage/interference simulations [6], specific peculiarity of TIM approach, is fundamental.

 

Figura 3 - SON use cases integration

The following use cases have been considered so far:

  • ANR+PCI (Automatic Neighbours Relations): neighbouring lists (defined by distributed SON) are controlled by centralized SON and PCI conflicts (caused by new neighbours) are automatically managed;
  • ACM (Automatic Configuration Management): automatic check and enforcement of configuration parameters (based on configuration guidelines);
  • MLB (Mobility Load Balancing): Inter-cell load balancing to redistribute traffic to avoid/delay sites congestion;
  • CCO (Coverage/Capacity Optimization): based on tilt adjustment and interference control;
  • Automatic lock/unlock: automatic execution of lock/unlock of cells/nodes.
  • Auto-provisioning of new nodes (also known as “zero touch” provisioning).

As a further use case, MDT (Minimization of Drive Test) was introduced, also exploiting the same reference architecture.

 

Optimization use cases

TIM methodology for Open SON use cases implementation is depicted in Figure 4, where three interconnected loops are represented:

  • “Inner” loop, describing core “DevOps” activities carried out “in house” by TIM for the Continuous Development (CD) of design, configuration and optimization tools (i.e. TIMplan, TIMqual, IRMA). SW modules by external providers can be also integrated. SW solutions are developed starting by a “Minimum Viable Solution” that is tested in field, followed by an iterative CI/CD process, tuning both algorithms performance and tools usability;
  • “Outer” loop, describing the exploitation of SON use cases, leading to process evolution (through automation): new use cases introduction is typically triggered by field activities feedbacks and/or technology innovation opportunities;
  • “External” loop, representing overall market and ecosystem evolution. TIM contributes also to this global evolutionary path through participation in standardization bodies (e.g. 3GPP, O-RAN…) and other innovation initiatives (European projects, trials….), but also through continuous co-operation with partners (selected through scouting and evaluation activities).

To better clarify the process described in Figure 4, specific application examples can be considered. The workflow for setting up a use case can start from a requirement coming from field activities, such as in the case of MLB, which addresses over-loaded cells (identified through KPI analysis), or when new opportunities are made available by technology evolution, as in the case of MDT feature, which provides a new methodology for “in-field” performance monitoring.

The second step is the feasibility study, which analyses the matching between field requirements and evolutionary “know-how” in order to decide whether and how to proceed with the implementation (GO / No GO decision).

Several aspects are evaluated during this stage:

  • Stakeholders: involved subjects for each phase of the work are individuated, analyzing whether they belong to internal structures of the organization, beneficiaries of the solution or development structures, or external subjects that provide tools and / or re-usable knowledge;
  • Technical feasibility: this point focuses specifically on the technical aspects. The technical viability of proposed use case is assessed by evaluating the architecture, the availability of “actionable” network parameters, the KPIs to be considered for evaluating the benefits of the use cases;
  • High level solution definition: the high-level design of the software solution is carried out. Needed SW modules may be already available or may need to be added to the existing framework: new modules are typically implemented leveraging on “in house” development capabilities. External SW modules may be integrated as well, selected in cooperation with procurement departments;
  • Cost sustainability: it assesses the economic viability of a proposed use case by evaluating the set-up costs, operating expenses and required efforts with respect to the benefits (Balanced Capex /Opex /Efforts and Benefits) The “cost of not doing” is also taken into account in this phase.

For example, in the case of MDT the feasibility was based on:

  • the availability of the MDT feature at the radio level (included in RAN vendors solutions);
  • the development of a platform for the collection and processing of measures;
  • the definition of a methodology, leveraging internal SW tools, for managing the processed data.

Once the feasibility study is completed (i.e. with a “go” decision”), the “inner” loop is activated, exploiting the paradigm of “continuous development”.

 

Figura 4 - SON use cases implementation through CI/CD (Continuous development, Continuous integration) process

As a matter of fact, fast evolving technologies lead to a major shift in the design and delivery of SW solutions. The waterfall paradigm cannot be applied anymore, because the context itself is continuously evolving, leading to the need of a SW production flowchart based on continuous development, which means continuous integration, continuous testing, continuous delivery. For the same reason, a strict cooperation is needed, even in the SW design/development phase, among developers and final users (typically belonging to the Operations departments).

For such a reason the “DevOps” paradigm, already adopted for the development of planning and optimization tools, quite perfectly matches the methodology adopted by TIM for SON use cases definition  (Figure 5).

During the Design phase, a “Minimum Viable Solution” is defined to start as soon as possible with “in field” activities and then to learn and to continuously improve the solution. This way of working conflicts with the linear, iterative design process that works towards a perfect problem-solution fit. It is a remarkable cultural change, which requires fast implementation rather than risk reduction, because the real use of the solution and the evaluation of its benefits are visible only in field, since it is impossible to capture “a priori” all the possible behaviors.

In the “digital world”, sub-optimal solutions may be released and then adjusted in continuous way, considering feedbacks from the users, and thus making the solutions more and more adherent to the operational needs.

“Learning in field” is the key to assess the real “added value” for which the solution was initially designed. This working method involves:

  • a fast development methodology, making available a preliminary release of the use case;
  • a fast testing carried out together with operation departments (does the solution work?);
  • an efficient “knowledge sharing” between the development team and the operations team for the functional testing (does the solution provide the expected benefits?).

As a final step, ”Operation guidelines” phase includes the final solution deployment and the drafting of the guidelines for recurrent use, encompassing both the operational steps and the related responsibilities (e.g. through the definition of a proper Responsibility assignment matrix). This phase leads to a continuous cycle too, since in-field deployment, in an evolving scenario, continuously requires process adjustments and SW solutions upgrades.

The methodology application results are reported in the “insights” sections, providing network performance improvements (e.g for MLB: Mobile Load Balancing) and operational efficiency improvements (e.g. for MDT exploitation in Certification Activities). A third application example related to VoLte performance improvements derived from PCI optimization can be found in [1]

 

Figura 5 - DevOps Reference Process

Conclusions and acknowledgments

TIM Open SON initiative was described, by focusing on three main aspects:

  • Automation is a crucial factor in TIM RAN digitalization path and “in house” development of design, optimization, configuration and orchestration tools is a key asset for “taking the control” in the Operator domain, both in the present and in the evolutionary “Open RAN” scenarios.
  • Continuous development, based on DevOps approach, has been adopted also for Open SON to fully exploits TIM’s know how on Radio Access and to evolve automation tools based on real in-field “experience”
  • For each use case introduced, network performance and operational efficiency benefits have to be monitored in order to enable a continuous tuning/improvement of deployed solutions.

The described methodology and the reported use cases are the results of an extensive cross-functional activity (still on-going) which involved many TIM colleagues, belonging to several TIM areas (innovation, planning, procurement, engineering, SW development, NFV implementation, network development, network maintenance, in field activity coordination…). The authors gratefully acknowledge the valuable contribution of each of the involved colleagues

 

References

“DigiRAN: il valore dell’automazione nell’accesso radio” Notiziario Tecnico TIM. N. 1 2018. https://www.telecomitalia.com/tit/it/notiziariotecnico/edizioni-2018/n-1-2018/N6-DigiRAN-valore-automazione-accesso-radio.html

https://www.telecomitalia.com/tit/en/archivio/media/note-stampa/market/2016/TIM-Altiostar-VRAN.html

https://www.telecomitalia.com/tit/en/archivio/media/note-stampa/corporate/2018/Nota-Stampa-TIM-Ericsson.html

https://www.o-ran.org/ [o link ad articolo nello stesso notiziario tecnico]

https://www.onap.org/architecture

“Managing complexity: augmented intelligence for 5G radio access design and optimization” Notiziario Tecnico TIM. N. 2 2019 https://www.telecomitalia.com/content/tiportal/it/notiziariotecnico/edizioni-2019/n-2-2019/N5-Managing-complexity-augmented-intelligence-5G-radio-access-design-optimization.html

“Mobile trend: Self-Organizing Networks” Notiziario Tecnico TIM, N.2 2014. http://www.telecomitalia.com/tit/it/notiziariotecnico/archivio/2014-2/capitolo-06.html

 

Notes

Automatic Method For Mobility Load Balancing In Mobile Telecommunications Networks – Patent No US 10.433,209,B2 – Oct. 1, 2019  and Patent No EP 3 318 082 B1

Example of T-MAST installed at Tirrenia for Italian acrobatic team exhibition: https://www.telecomitalia.com/content/dam/telecomitalia/it/archivio/documenti/
media/note_stampa/mercato/2019/Carrato_Tirrenia1.jpg