Control the Chaos
Operating concepts for 4G mobile networks rely on self-organization.
Dr. Ken Roberts | Josef Thormann | Murugaraj Shanmugam
Evolution in the mobile telecommunication networks is inevitable! 3GPP LTE is the key word here — in the future, these networks will support an enormous number of automation functions which simplify the operation of the network The cost savings which become possible increase the attraction of network operation for mobile network operators. The purpose of this article is to identify the need for the 4G OSS systems to support the automation capabilities and to propose a solution approach for the 4G OSS in order to support the SON concepts.
The demand for anywhere, anytime communication is increasing, and the end users (subscribers) are demanding broadband speed coupled with seamless mobility. Consequently, new technologies are being developed and deployed to accomplish the vision of all-IP and broadband – like performance in mobile networks.
The Third Generation Partnership Project (3GPP) is working on a new radio interface, which provides high bandwidth e.g., 100Mbit/s downlink termed as “Long Term Evolution (LTE)”. The 3GPP’s System Architecture Evolution (SAE) initiative is working on network architecture to achieve the vision of All-IP based Packet Network. This network architecture is characterized as a simple and a flat IP network infrastructure. The combination of LTE’s higher bandwidth capabilities and SAE’s flat IP network architecture fulfills the requirements of Next Generation Mobile Networks (NGMN), also referred to as 4G networks.
Additional to providing higher throughput and all IP network concepts, the 4G NGMN networks are capable of supporting numerous additional automation functionalities such as; Self Configuration, Self healing etc, collectively termed “Self Organizing Networks (SON)”. The SON functionality is thus very appealing from a mobile operator perspective due to its direct impact on the “costs” of the operations.
The current networks (2G, 3G) in contrast have poor automation and, configuration changes to the network require manual intervention and result in high operational effort. So, it is of great interest for operators to minimize such operational efforts by reducing the need and effort of manual intervention. The SON concepts are primarily focused on fulfilling automation requirements and provide a vital value-proposition, reduced “operational cost” by introducing self-organizing mechanisms, for the mobile network operators.
To provide an overview of the SON concept, we will briefly describe the capabilities of the 4G network with SON functions:
- Self-planning: This procedure facilitates finding the optimal location and basic radio/configuration parameters required to deploy a new eNodeB (radio base station) installation.
- Self-Configuration: To enable a eNodeB to automatically configure the fundamental and, basic configuration information to support “plug and play” behavior.
- Self-optimization and self-tuning: To support certain performance measurements that are used to (auto-) tune the network parameter to increase quality and performance.
- Self-testing and self-healing: To validate the configuration information and to execute periodic performance tests in the eNodeB. If a problem is detected, execute recovery procedures to return the node to its “normal” operational state.
The objective of the SON is to automate operational activities through extended and more intelligent network elements and OSS. As a positive effect, the network should be operated in a highly efficient and effective manner with a minimal human intervention. Moreover, strategically the SON will provide the following benefits: Reduced development, testing and maintenance costs, decreased design to deployment times, increased reliability of network and services.
Certainly, the SON features have the potential to assist mobile networks operators to achieve the required network optimization and to provide maximum performance. So, if the SON is considered as one of the important functions, the 4G networks should be able to monitor, manage and control the SON enabled nodes through its Operational Support Systems (OSS). So far, little work has been performed in this area and the requirements of the 4G OSS are still unclear.
In this article, we focus on the OSS aspects of the SON enabled 4G networks. If the underlying network elements are capable of performing autonomic functions, what is the role the management of such networks i.e., OSS needs to play? As a mobile network operator, it is of great importance to have controlled automation to avoid chaotic network environment and we argue that such controlled automation should be a fundamental supported by the 4G OSS.
It should be noted that the scope of this document is restricted only to OSS; with respect to SON features, and, any other NGNM specific details are intentionally left out of scope. The purpose of this article is to identify the need for the 4G OSS systems to support the automation capabilities and to propose a solution approach for the 4G OSS in order to support the SON concepts.
SON Concepts Heighten Performance Capability
Besides the bandwidth and high throughput improvements in the 4G LTE network, it postulates an innovative and powerful approach to improved network performance. These innovations are collectively termed “Self Organizing Networks”. The Self Organization concept enables or provides the capability for the (/a set of) network elements automatically adjust its (/their) parameters (according to the actual network situation) in order to achieve maximum performance.
The SON philosophy refers to entities ability, in the network, to take autonomous action in response to an extant condition or trigger. Autonomous action is generally accepted to be some form of automation based on a set, or sets, of constraints defined by the user (i.e., organization).
The current OSS systems in the mobile network (2G, 3G) are not designed to handle such autonomic behavior of the network elements as these networks do not have “automation” or “self organization” at their core. So, if the future network elements are able to perform such “automation” then the following question arises. As a 4G network operator: Is it necessary to control the automation? And if so, how to control them?
For example, it is already technologically possible today for the SON-enabled eNodeB (enhanced Node B) to detect the inter-cell interference between two base stations. The SON’s objective of this use case is to provide the possibility for the network elements (base stations) not only detect the cell interference but also automatically adjust the base station parameters to achieve optimum performance.
Technically, the ability of the base stations to achieve the optimum network performance sounds exciting and promising. However, individual network elements do not have the overarching network knowledge and, tuning the network parameters without such “global’ network knowledge may not be preferable and may result in chaotic network behavior.
In order to govern such autonomic behavior, rules have to be carefully designed and be enforced. These rules are not only based on the pure technical measurements but also subject to environmental conditions.
For example, it may be technically possible that a base station is allowed to maximize its transmitter power, environmental conditions (regulation) may not permit it. These regulations may also be dynamic and changes according to the environmental conditions, for example transmit power may be limited between the hours of 08:00 and 18:00 due to the close proximity of a kindergarten (certainly this is NOT a technical limitation however it may be a politically expedient ‘rule’). So the rules have to be designed and enforced according to the local context (current situation of the base station) and subsequently, fitting to the global context (current organizational policies). As a consequence, for the network operator, it is extremely important that the automation be handled in a controlled manner.
To effectively manage the SON features of the underlying 4G LTE network elements, the 4G OSS has to be aware of the autonomic capability of the individual network elements and apply governance to achieve the expected behavior. We argue that the SON features, i.e., automation capabilities, will have a direct influence on the OSS architecture and will change the way the OSS have been designed.
This is due to the fact the 4G OSS processes and systems should consider the automation possibilities at its core. The OSS processes (e.g., Fault management) should accommodate the automation features and be ready to handle the changes that are executed as a result of the SON. Subsequently, the underlying OSS systems should be aware of what changes are required and more importantly, how to control the changes to achieve the expected behavior.
Starting from this basis, this article argues that the 4G LTE SON influences mainly both the operations and maintenance (O&M) processes — in that SON mandates modifications of existing operational activities to handle the autonomous behavior of the network elements — and the OSS architecture — because the SON features require specific functional capabilities and interfaces between OSS components.
The following section will discuss the possibilities for the 4G OSS in order to handle the SON features.
OSS with Integrated Policy Management
SON incorporates autonomic behavior into the Network Elements (NEs) in response to achieve the required behavior stimulus. A behavior stimulus can be described as an event (or incident) requiring action to correct or normalize. An example for such event or incident could be, actual network performance exceeding a ‘threshold’ and the action may be to establish a particular quality of service to improve the network performance.
The required performance metrics and thresholds can be either statically configured within the network device (factory provisioning) or dynamically set overtly by an OSS function. So, whenever the actual performance exceeds the threshold, a local NE resident function will be triggered to address the situation.
In the traditional network (2G, 3G), such ‘actions’ are defined and triggered by their respective OSS. However, in the 4G LTE SON, the NEs have the capability of triggering an action and thereby bypassing the OSS. For network stability and integrity reasons, such actions should be performed only with the knowledge of the OSS. So, the 4G OSS incorporates the ‘governance’ capability in order to handle and control the SON based NEs. Such a governance capability can be realized in the form of Policy based Management (PbM) and this document discuss PbM as a potential solution for the 4G OSS. PbM can be described as a governance paradigm whereby the ‘rules’ applicable to an entity are stored external (to the entity) and applied as required by OSS.
Figure 1 depicts the role change in 4G OSS to manage ‘smart’ devices.
This role change is required because the device is itself capable of processing and the OSS is responsible to ensure that the actions taken do not impact or create ‘knock-on’ situations within the managed environment.
Policy Management based OSS
A policy enabled environment provides functionality enabling decision criteria (and the decision itself) to be defined external to the system the decision affects. The incorporation of policy allows the application of external decision criteria to be applied to situations as they arise.
Figure 2 shows this capability is positioned as an integrated but separate component of the OSS/ BSS system.
The principle artifacts associated with policy architecture are:
- Policy Decision Point (PDP): A PDP provides the functionality associated with making a decision. It retrieves the set of ‘rules’ associated with the decision and applies operational considerations to the rules and establishes [set of] actions that address the operational situation.
- Policy Enforcement Point (PEP): PEP provides the functionality associated with rule execution. Rules specify [set of] actions established by the PDP.
- Policy Repository (PR): A PR contains the policies, typically possesses the storage functionality.
Figure 3 shows how decision making is accomplished by using an external source.
It must be noted that the invoked processes (shown extreme right) are not aware of the classification criteria; they process only information relevant to their design. The association of a specific action response (OK, Alternative or NOK) is external and does not have any design impact on the application process, that is an application process is designed for a specific process and is ignorant of any ancillary condition or situation that may exist that is not directly relevant to their design.
The linking between individual elements of either application process or policy process is accomplished by publishing or subscribing to messages. These control flow messages are generated and set as process post conditions (published), or consumed (subscribed) triggering a process pre condition.
The combination of PDP and PEP provide the functionality to encode, evaluate and execute appropriate rules based on observed behavior within the managed environment. The triggers for rule execution are observed behavior and this requirement is predicated on a mechanism to evaluate conditions within the managed environment. Typically, these triggers are specified as notifications issued by a NE (i.e. managed object) describing a situation or condition within the NE.
Implications of Policy elements in OSS
It is apparent that intelligence within the NE poses a new situation that a ‘centralized’ OSS may not need to operationally address an event or incident. This is due to the capability of the NE to execute particular actions to overcome the subject event or incident. However, it is foreseeable that the OSS still would like to control and manage the autonomous behavior of the entities in order to avoid chaotic situation. To achieve that, special capabilities (in form of policy server) need to be incorporated within the 4G OSS.
In this context, the introduction of policy management presumes the appropriate “know-how” within the organization, e.g., so that rules and policies can be created and established, as well as the adaptation of operational processes to include the policy management, plus the introduction of PDP as a decision-supporting system within the 4G OSSS and the implementation of policy elements for the entities that need to be governed by the 4G OSS.
Moreover, these policy elements (PEP, PDP) within the 4G OSS are an additional task when compared to the current OSS. Although the policy elements aim to simplify the relatively mundane operations of the network, their implementation requires substantial expertise and effort. This "know-how" must first be acquired by the operator before the automation potential can be exploited.
With the introduction of automation, the 4G OSS takes a new responsibility that governs the autonomous actions of the network, since the network elements are capable of performing certain functions (e.g., fault prevention and remedy) that were initially performed by the OSS. And finally, the OSS functionality moves from “what” must be done to address the incident/situation to “when” must this action be done. This transition requires a deep understanding of the action to be performed and its causal effects, not only in the NE itself, but for the collective environment (network and organization policy) within which the NE exists.
Advantages of a Policy-Based SON
Policy-based SON in the 4G LTE environment minimizes the workload on a “central” OSS by reducing and ultimately eliminating manual configuration of network operational parameters. Another decisive advantage is that the policy-based SON network enables agility and complete control over the network. Such a control helps the network operator to react faster according to the current network situation and organizational requirements. We envision that the policy-based SON technology in an LTE network will enable the network elements to automatically (more importantly, it enables flexibility) configure, change, optimize and even extend network coverage, capacity and topology based on network & organizational policies. Moreover, the network will support flexible frequency and bandwidth allocation, based on predefined rules which allow the node to independently react to interference, changes in signal strength, location of end users, and other volatile environmental criteria.
The policy based self organizing networks clearly offers the potential to further automate operational processes in the mobile network and to operate the network in a highly efficient and effective manner with minimum human intervention.
The 4 G LTE SON technologies allow the implementation in different degrees: from basic SON to full blown SON. The degree of complexity increases depending on the level of SON implementation. Operators intend to implement SON could start with basic SON where procedures like Self-Configuration for the eNodeB can be introduced and move towards the full-blown where Self-Optimization and planning procedures can be introduced. Note that the time line and the preference of the implementation level of SON is dependent on the requirements of the network and the organization.
Figure 4 depicts the influence and the impacts of SON within the organization.
SON is an innovative concept that will ultimately reduce the costs in the mobile network, especially “OPEX” in the long term. This new technology will minimize operating expenses by reducing the overt management involvement day-to-day required for running the network.
It should be noted that the SON technology has its own challenges. For one, the greater the SON capability, the higher the CAPEX will be. However, this is a one-off investment which will help the organization to reduce the OPEX in the long term. Secondly, the complexity of shaping the correct policies for running a network will dramatically increase. And thirdly, with more automation and less human intervention, the vulnerability of the network is likely to increase and this needs to be carefully addressed.
For the stakeholders in the organization, SON aims to improve performance and efficiency by changing the traditional nature of operation & evolution in the mobile network. For the successful SON implementation, it requires close cooperation between the stakeholders in the organization. The required level of cooperation between the stakeholders (e.g., planning and operations) in the organization should be defined & properly established. In conclusion, by introducing automation, SON ultimately aims to achieve the goal of business efficiency within the organization.