The access components in heterogeneous networks consist of Home NodeB, Home eNodeB and corresponding gateways that interconnect the NodeBs with the operator core. Here are a few unique characteristics of heterogeneous networks:
1. Architectural Changes
In HetNet, there are 5 basic access mechanisms that are defined by the 3GPP standard. This includes pico cells, relay nodes, Wi-Fi access nodes, Home NodeB and Home eNodeB. The Home NodeB and Home eNodeB are connected to the core network through the security gateways and Home eNodeB gateways. For the Home eNodeBs, since the X2 interface used primarily for interference control is optional, there is a dependency on the OAM (operations, administration, and maintenance) and SON (self-organizing network) functions to carry out the optimizations. For the Wi-Fi access, there is a trusted and non-trusted model. In the trusted model, the UE (user equipment) is transparently switched to the Wi-Fi network since the access points will be trusted and verified by the operator core. But with the non-trusted model, the user equipment needs to establish a secure IPSEC connectivity with the core over a Proxy Mobile IP tunnel after authenticating the credentials with an AAA server through the packet data gateway. In relay node option, as the name indicates, donor eNodeB coverage is extended through the use of relays over a wireless backhaul. These architectural changes subsequently impact the call flow and eventually impact the traditional management architectures, functionalities, and management system development road maps. There’s now a different level of classification for small cells that’s popularly called metro cells, which are deployed indoors or outdoors. They have a higher processing capacity and are typically managed by the CSPs to enhance capacity and coverage of the macro cells.
2. Self-Organizing Networks (SON)
Self-Organizing Networks primarily consist of three mechanisms—selfoptimization, self-healing, and self-configuration which are components of an automated service assurance system. In a network ecosystem like HetNet, it’s essential to leverage SON because traditional monitoring and management might not be efficient and could lead to high operational expenditures. For traditional service management solutions, SON will be an enabler that impacts operational procedures implemented by these solutions.
One notable advantage of HetNet is the flexibility it offers in terms of using multiple radio access technologies like 3G, LTE, Wi-Fi or traffic rerouting/offloading options related to LIPA (Local IP Access), SIPTO (Selected IP Traffic Offload) with or without mobile operator support. With multiple access options available, 3GPP also offers a mobile IP based network address assignment, which provides flexibility to seamlessly switch from one access mechanism to another while ensuring session mobility. This is going to impact service management systems since the service can flow through any of the available access mechanisms. Until the operator has visibility into all the supported access options, there will be gaps in determining service quality.
4. Deployment Model
With the heterogeneous network, multiple deployment options are possible. Among the emerging deployment options are Femto as a Service (FaaS) and Small Cell as a Service (SCaaS) where the access terminals and antennae systems are owned by a hosting provider (Home (e)NodeB Hosting Party) and multiple mobile operators leverage the access infrastructure built by these hosting providers. Similar characteristics exist in deployments having Distributed Antennae System (DAS) owned by a hosting party or the building infrastructure provider, which is shared by multiple CSPs. RAN (radio access networks) sharing, or Mobile Operator Core Network (MOCN) sharing, is another area that will greatly impact management because the RAN infrastructure will be shared by multiple operators, and collecting and distributing performance and fault metrics will involve additional overheads. Typically in these types of shared infrastructures, the SON may not work as effectively because the optimizations carried out for one operator may not function properly for others. Phantom cells, currently in the research stage, is another type of deployment model where the control and bearer follow different paths – i.e. the control path is handled by a macro cell and the bearer path is handled by a small cell. The choice of deployment model varies, depending on each operator’s business requirements. The deployment model also impacts management frameworks since multiple usage scenarios will need to be monitored and measured.
Service Management of Heterogeneous Networks - Challenges
In service management, service modeling is a critical activity which defines association metrics for a service at a high level. Three main types of associations are defined – services to resources mapping, service performance metrics to resource performance metrics, and resource level alarms/notifications to service level alarms/notifications.
Mapping services to resources depends on the inventory management system. With the introduction of heterogeneous networks, many access nodes need to be accounted for by the inventory management system in terms of their physical and logical information. It’s also essential to have an automated reconciliation mechanism that can discover configuration changes and update the inventory.
When mapping service to resource metrics, it may not be that challenging to define the metrics, but four fundamental factors need to be considered to handle the flood of performance data—latency, performance data size, collection frequency, and capacity. Since HetNet supports a parent-child cell model, where one macro or micro cell acts as an aggregator for multiple small cells, it’s essential to define the aggregated performance metrics, which need to be mapped to the service metrics that enable overall service performance assessment.
For mapping between service alarms/notifications and resource alarms/ notifications, one challenge is the reporting of small cell level faults to the network operations center and the subsequent impact assessment on service, such as when a local power outage switches the small cells in the region to back up power. In service modeling, the alarms need to be carefully chosen and designed to avoid alarm flood and alarm filtering rules at the resource layer so that service impact is assessed without imposing a load on the management infrastructure.
One critical impact on service management resulting from the introduction of SON will be the self-healing mechanism. In the current SON, the self-healing mechanism is oriented towards resource level problem identification and rectifying problems like cell outage, coverage holes, etc. The efficiency of the self-healing depends on how quickly and accurately problems are detected and fixed, which, in turn, depends on the vendor products and corresponding element management systems.
Like self-healing, self-optimization and self-configuration will also impact service management systems. Self-optimization detects optimization issues and works with configuration modules to fine-tune end device parameters. Some of these optimization issues include handover thresholds, power settings, antennae parameters, neighbor lists, cell identity parameters, and tracking area association. Since this is done dynamically, until the inventory system gets updates on such configuration changes, there will be data mismatches that impact service models and data aggregation mechanisms in the service management systems.
Another challenge with enabling SON is that the legacy 2G/3G portion of the network will continue to be managed manually while the small cells will be monitored and controlled through automated SON functions. This creates additional operational overheads, especially in service scenarios involving mobility across different mobile technologies and cell variants. It will also be operationally challenging to maintain separate procedures and systems for the management of the legacy and small cell RANs.
On Demand Management
In 3GPP specification 32.835—pertaining to HetNet management— one of the critical requirements is on-demand management, which is context-specific management of the deployed infrastructure. The majority of Common-Off-The-Shelf (COTS) OSS components, especially those marketed in the service assurance space, have limited flexibility in terms of defining management contexts. This limits the operations team from implementing custom management use cases and often this leads to heavy dependency on the network vendor or managed service partner.
Management of Converged Network
With respect to HetNet, one particularly interesting use case, “Femto Access Management,” (described in 3GPP specification 32.833), describes a combination of fixed and mobile access elements in a residential deployment scenario, which provides connectivity to the operator’s mobile core network. The femto network is deployed to provide a convenient, high-quality connection to the mobile core when the mobile user tries to access the services from home. Here the service management system should have the intelligence to determine the access mechanism used by the end user when he connects from home or outside, or seamlessly moves into a femto connection zone. To enable this, a separate set of resource metrics and the corresponding service dependency need to be ascertained as part of the service modeling activity.
Integration of 3GPP RAN with Non-3GPP RAN
controlled by a partner operator gets connected to the primary operator mobile core, which is quite possible in non-3GPP access mechanisms, such as HetNet-supported Wi-Fi and WiMAX. When the user switches from 3GPP to Wi-Fi access, the control layer messaging may switch over to the ePDG, assuming a non-trusted Wi-Fi access. The service management layer needs to have visibility across both Wi-Fi and 3GPP access layers, with information on the mobility parameters, to identify the sessions that are switched from one technology to another. Similarly, the session quality of service also needs to be measured regardless of the access technology being used.
Recommendations and Implementation Approach
This section looks at service management, specifically service assurance of heterogeneous networks, from an end user’s point of view.
Assuming that the majority of brownfield CSPs across the globe have a base network/resource management architecture in place, more are now moving to the next stage in telecommunication network management—from resource level management to service level management. While the TM Forum Business Process Framework and associated process areas can be used as a reference for defining the service management architecture, there are equally important frameworks like TOGAF and the relatively new 3GPP management architecture. Since TOGAF defines the enterprise architecture, which is broader in nature, it’s relevant to consider the more focused 3GPP proposed management architecture and its implications for HetNet service management.
As shown in Figure 2, 3GPP 32 series specifications prioritize information management through the use of information models rather than network elements as physical resources. The advantage of this management architecture is that it’s vendor agnostic and it emphasizes functional elements and the management of information.
This architecture relies on three standardized components:
- Network Resource Model: The management information inside network elements that need to be managed.
- Interface Model: The management information exchange between the network resources and management systems.
- Common Data Model: the common data that can be used by the network resource and interface models.
These three models are represented by integration reference points, which consist of:
- Generic: A logical model called information service, and
- Specific implementation models called solution sets
Northbound and southbound interaction between the entities is carried out through interface N, which has associated generic information and implementation specific solution sets defined using interface model specifications.
The only options currently available from a service management point of view are subscription management (3GPP 32.17x) and user data convergence (3GPP 32.181/182). Some use cases, like TCP optimization, Location/Proximity detection, media broadcast, and Service Session Tracing, may require direct interaction by the services layer with the resources. Per the 3GPP architecture defined above, this requires that the functionality of the network manager be extended to serve service specific management requirements.
Also, service sessions, especially those involving Home eNodeB, pass through the security and Home eNodeB gateways (GWs) on the RAN side. In this case, the management systems for Home eNodeBs and respective gateways may be from a single vendor or from separate third-party vendors. Interacting with multiple vendor elements will impose additional overhead because the management interfaces and SON specific algorithms utilized by each vendor may be different.