The invention relates to service management in communication networks.
Operators are increasingly faced with the challenge of how best to set up their network in order to satisfy dynamic service demand from their customers. This is increasingly so with the advent and introduction of more complex messaging services and architectures (e.g. in the context of IMS or the overall context of personalised messaging services and experience).
These challenges are further compounded by the fact that operators are increasingly simultaneously managing multiple heterogeneous networks such as traditional GSM, and newer IP based IMS networks.
Existing network capacity measurement systems tend to focus in isolation on measuring capacity of particular network service point sub-systems such as for example endeavouring to determine sizing/capacity of specific node(s) of an SMS sub-system, whereas the challenge the operator is facing is optimizing the entire messaging network infrastructure, including sub-systems.
The invention is directed towards providing an infrastructure for improved service management.
According to the invention, there is provided a service management framework for operation in a communication network having a service delivery infrastructure with network elements providing services to subscribers, the service management framework comprising:
In one embodiment, the analysis means is adapted to generate at least one service predictive model for each network element, group of network elements, or subscriber segment or subscriber segments.
In one embodiment, the analysis means or the collecting means is adapted to execute a set of data adapters for initial processing of collected data.
In one embodiment, the analysis means is adapted to use agreed quality of service requirements when determining an action to be implemented which may affect quality of service. In one embodiment, the analysis means is adapted to determine an action to be implemented which affects quality of service for a subscriber segment and not other subscriber segments, the determined actions being selected from one or more of re-prioritizing a subscriber's messages in delivery queues, adjusting bandwidth for a subscriber, message re-direction to particular service centres, or re-allocation of hardware resources across multiple services.
In one embodiment, said service predictive models track network infrastructure utilisation parameters including current number of active subscribers, traffic demand, projected number of active subscribers, and expected demand.
In one embodiment, the analysis means comprises means for generating network element models defining characteristics of network element configurations. Preferably, the analysis means comprises means for performing analysis of relationships between said segmentation, service predictive, and network element models. In one embodiment, the analysis means comprises means for managing subscriber objects each including at least a subscriber identifier, and service objects each defining a service, and for managing associations between these objects to develop the models.
In one embodiment, the analysis means is adapted to manage subscriber segment objects, each defining a subscriber behaviour segment, and to manage network element objects each defining a network element of the service delivery infrastructure and its associations with service objects.
In one embodiment, the analysis means is also adapted to manage trend objects, each trend object representing dynamic aspects of at least one parameter and being executed to track behaviour of a subscriber object, a service object, a network element object, and a segment object.
Preferably, at least one trend object tracks a basic trend and at least one trend object tracks a derived trend including service demand which is derived from at least one basic trend object.
In one embodiment, each segmentation model comprises a segment object and related subscriber objects and the subscriber objects are related to trend objects representing subscriber behavioural characteristics.
In one embodiment, a service predictive model comprises a segment object and related trend objects.
In one embodiment, the analysis means and the acting means comprise means for determining if a subscriber exceeds fair usage beyond a flat fee tariff, and for performing an automatic action of re-prioritizing that subscriber's messages in delivery queues, or adjusting bandwidth for the subscriber.
In one embodiment, the analysis means and the acting means comprise means for performing automatic subscriber re-provisioning including changing a subscriber's service centre address.
In another embodiment, the analysis means and the acting means comprise means for notifying a subscriber of an action which has been performed or needs to be performed by the subscriber. In one embodiment, the analysis means and the acting means comprise means for performing automatic pushing of information to subscriber devices. In one embodiment, the analysis means and the acting means comprise means for performing automatic pushing of information for Over-the-Air device re-configuration.
In one embodiment, the analysis means and the acting means comprise means for performing automatic re-configuration of a network element.
In one embodiment, the analysis means and the acting means comprise means for performing automatic re-configuration of a network element to operate with a different technology stack, such as transfer from GSM to IMS (SIP enabled).
In one embodiment, the analysis means and the acting means comprise means for automatically adjusting operation of a network element for service delivery optimisation. In one embodiment, the analysis means and the acting means comprise means for automatically adjusting network control parameters including message delivery attempts per time period or number of active subscribers.
In one embodiment, the control parameters are associated with a license for a network element or group of network elements.
In one embodiment, the analysis means and the acting means comprise means for performing automatic re-configuration by activating software resident on network element hardware servers to achieve intelligent service provisioning.
In one embodiment, the analysis means and the acting means comprise means for performing automatic re-configuration by installing software on network element hardware servers to achieve intelligent service provisioning.
In one embodiment, the analysis means and the acting means comprise means for performing automatic activation of hardware servers to augment processing capacity of certain network elements.
In one embodiment, the analysis means and the acting means comprise means for performing re-configuration of a server to allow or prioritise delivery of content from or to a content provider or a subscriber.
In one embodiment, the analysis means and the acting means comprise means for performing modification of a network element or link settings including allocated in-flight capacity, content validity periods, network connections, network connection bandwidth, or quality-of-service parameters.
In one embodiment, the collecting means comprises an interface for receiving network element transactional data including call detail record (CDR) streams.
In one embodiment, the collecting means comprises an interface for receiving network performance objects containing information concerning network element utilisation, including SNMP performance objects.
In one embodiment, the collecting means comprises an interface for receiving network element monitoring information.
In one embodiment, the collecting means comprises means for receiving real time data feeds.
According to another aspect, the invention provides a computer readable medium comprising software code for performing operations of a framework of any embodiment described above when executing on one or more digital processors.
The invention will be more clearly understood from the following description of some embodiments thereof, given by way of example only with reference to the accompanying drawings in which:
An intelligent service management framework of the invention collects network usage data, analyses this data, and then dynamically acts in response to the data in an automatic feedback loop to modify service delivery infrastructure in order to optimise service delivery. The framework is dynamically adaptive to service usage. The framework is particularly important for network operators as it provides the ability to automatically adhere to or manage agreements such as service level agreements (“SLA”) and licenses governing capacity concerning factors such as numbers of active subscribers and throughput of network elements. Also, because the framework optimises the technical capability of the network the operator can optimise the number of active subscribers and the throughput of network elements. Also, they can optimise quality of service (“QoS”) for services provided to the active subscribers. The framework executes segmentation and service predictive models to dynamically and pre-emptively modify the service delivery infrastructure of the network to optimise service delivery. This is described in more detail below.
In the following detailed description of the framework the subscriber represents the entity that uses a service, i.e. has some form of subscription. The subscriber can be either a mobile user or an application (with a subscription service). The subscriber consumes various products and services, offered by the service provider. There is usage of network-based products and services. There are: bearer-based services (e.g. voice, SMS, MMS) and value added service, delivered over the bearer services (e.g. content delivered over SMS).
The service delivery infrastructure includes a set of hardware and software resources having configurations and underlying communication links. An example of a delivery network is a set of service centre and gateway platforms connected in a way to support messaging services over multiple bearers. “Service demand” is a quantified level of service that subscribers intend on consuming (e.g. SMS demand is 10 million messages a day). “Demand forecast” is the prediction, projection or estimation of expected demand over a specified future time period.
The framework automatically collects service usage data and service delivery infrastructure utilisation data. It automatically analyses this data, and then acts in an automatic feedback loop to optimise the network. As part of the analysis phase, the framework performs subscriber segmentation, which is a process of identifying distinct groups of subscribers with common service usage behavioural characteristics. Behavioural-based segmentation is important; therefore the segment definitions could include segments such as “Trend Followers”, “Intensive Users”, “Low Users”, and “Irregular browsers”. The subscriber segmentation provides a set of segmentation models representing behavioural characteristics of subscribers; and the framework derives service predictive models from a combination of segmentation models and other collected data concerning the network resources. The service predictive models are used to generate predictions and on this basis the framework automatically optimizes messaging infrastructure to for example introduce new services. The predictive aspect of the analysis is very important—allowing automatic pre-emptive network modification.
The segmentation models represent service usage patterns, specific to each subscriber segment. Service demand levels are determined relying on segment-specific service usage patterns. Forecasting techniques are used to produce differentiated service demand models (to reflect specific characteristics of individual subscriber segments). Service demand models are an example of service models.
Service predictive models are developed and optimised by correlating the service demand levels (produced as a result of subscriber analysis) and collected data concerning service delivery infrastructure utilisation. Service delivery infrastructure is optimised based on the models, produced for respective services, which optimisations in the main are in real-time but depending on the service optimisation can also be in near real time or performed over a longer period. The automated operation of the framework effectively results in automated understanding of subscriber preferences and behaviour to dynamically maintain a service delivery infrastructure accordingly.
Referring to
Usage information is collected by the intelligent framework components to create a complete profile of the subscribers. The framework analyses subscribers and builds the segmentation models, based on which a set of behavioural characteristics specific to various subscriber segments are created. Those characteristics provide an historical view of the behaviour of subscribers and are used to generate a set of predictive models. Alongside the subscriber-centric view, service platform utilisation information is aggregated by the framework to model how the service delivery infrastructure network elements perform.
The acting phase allows definition/deployment/adjustment of those services across the network in the most efficient way (which depending on the service adjustment can be in real-time), so that available platform resources are used optimally.
For data collection, the framework interacts with the service delivery network to collect real-time transactional data, such as call detail record (CDR) streams, and monitoring information, such as performance objects. The real-time data collection and analysis (as opposed to the traditional data warehousing type of off-line post-processing, producing results in weeks/months time) supplies other analytical modules of the framework with up-to-date information.
Subscriber segmentation is an important aspect. In the example of
The subscriber analysis module provides real time information based on the service usage data, collected from the service delivery network components (further details on service usage data analysis are provided below).
In one example, automatic analysis of the service demand carried in the GSM and IMS networks allows evaluation of the level of utilisation of the service centres (SC1, SC2, and SC3) and the gateway (GW). Following the trend lines and looking at the sizing points at June 2007 (06.2007), end of 2007 (12.2007) and end of 2008 (12.2008) the required capacity for each type of service is determined (“Service Delivery Infrastructure Usage Analysis” component in
Referring to
Thus, dynamic transformation of particular network nodes is achieved, whilst maintaining seamless service continuity for subscribers.
In-line with the other flexible service management considerations, the framework allows control of the effective network capacity dynamically allocated to the delivery network components, allowing optimisation of delivery of the service to the required degree (in terms of capacity).
The framework allows flexible licensing management and so is particularly advantageous for license agreements between operators and equipment suppliers, which are typically based on the maximum hardware capacity of the network elements, or estimated required capacity, or number of active subscribers. The framework performs adjustment of network element control parameters associated with licenses. Examples of control parameters are number of message delivery attempts per time period and number of active subscribers.
Considering the example with the SIP and GSM service centres above, the capacity is dynamically managed based on the actual monitored and predicted usage patterns. In the example above the subscriber analysis module provides the actual number of active subscribers, making use of the service over SIP versus GSM. The anticipated number of active subscribers (based on the predictive behavioural models) could be used to manage agreements between the operator and mobile virtual network operators (MVNOs), virtual network operators (VNOs), or value added service (VAS) providers or the service delivery infrastructure equipment/component vendors in a real-time manner.
In
It is important to note that similar to provisioning service-related attributes of the service delivery infrastructure components, the subscriber-related attributes are equally provisionable. Direct subscriber provisioning is performed in a comparable intelligent manner.
Referring to
A set of databases (the “subscriber profiles data store”, “service delivery infrastructure data store”, “raw usage data store” and “service model store” in
Service usage data collection, sanity checks and data assurance type of activities are performed as part of the “collect” phase: flows 1 (1a, 1b, 1c), and 3 (3a, 3b, 3c and 3d) reflect the actual data collection process (whereby either a pull or push or streaming mechanism could be used to transfer the usage data to the framework platform (the push mechanism is shown in the example). CDRs are used for information about individual transactions in the depicted example (flows 3a, 3b, 3c and 3d). Note the example message flow labelled “Msg Flow 2” in
The “analysis” phase consists of a number of stages. First of all, the transaction level data (flow 5) is processed and aggregated to the subscriber level (block “subscriber analysis: Segmentation and Predictive Modelling”), whereby a set of basic behavioural and derived trends are calculated for each subscriber (flow 6). Subscriber segmentation is performed within the same step: subscriber segments are identified based on a set of behavioural rules (block “subscriber analysis” and flow 6). A set of segment-level trends which represent behavioural characteristics of all subscribers, belonging to a segment, are calculated as well. These aggregated segment-level trends (basic and derived, which hold results of predictive models) form input for further service demand analysis.
A similar activity is performed with regard to the service delivery infrastructure analysis—the raw usage data is analysed, whereby the platform utilisation data, is combined with transaction information to get a complete picture as needed, (flow 7). A set of trends is produced for individual service centres, i.e. components of the service delivery infrastructure (flow 8).
The final analysis step is correlation of the following three elements (provided by previous stages):
(Refer to the intelligent service management framework modules in
The service predictive model is adjusted based on the analysis results, whereby a recommended set of characteristics specific to network elements is determined and activated (flow 12).
In the “act” phase the recommended service configuration (flow 13) is propagated to the actual service delivery infrastructure components (flows 14a, 14b and 14c).
In more detail, the following explains how the network is dynamically modified because of the actions of the framework. As described above with reference to
Flow 14a shows a re-configuration action consisting of activating SC application software. In the given example—SIP enabled SC software is activated on the SC3 platform.
As shown in flow 14b, the GW connectivity (network end point) parameters are set up to point at SC1 and SC3 as SIP SCs, and SC2—as a GSM SC. Using the service usage daily pattern information the maintenance window (i.e. a time interval when the service demand is low) is determined. During this time interval the GW configuration is temporarily adjusted to utilise SC1 and SC2 only; the SIP SC software packages are activated (installed, if needed, and activated, using the service provisioning commands/primitives, supported by the SC platform/software and ‘known’ to the service provisioning module); and, when completed, the GW configuration is adjusted again to utilise all three SCs as mentioned above.
Flow 14c shows that, due to service demand, the framework acts to provide an increase in the capacity of SC2 to cater for higher message throughput.
Architecture of the Framework
Heretofore the challenge of dynamically managing a service delivery infrastructure has not been addressed in a highly automated manner apparently because of the perception that there would be an excessive amount of processing required and indeed complexity in development and maintenance of the system. The invention provides an architecture for the framework and particularly the analysis means which achieves automatic feedback and action without excessive overhead.
Referring to
Subscriber: The subscriber represents the entity that uses services, i.e. has some form of subscription. The subscriber can be either a mobile user or an application. It will typically be uniquely represented by an ID (for example, MSISDN, short code, potentially an IMSI or a set of IP addresses). Additional objects (the “Subscriber Info” object in the overview), related to Subscriber, could be added if relevant data is available (e.g. “Equipment” Object, phone, or PDA, “Subscriber Services”, “Tariff' to represent the tariff associated with the subscription type and detail the charges which are applied for particular services, “Promotion” to represent activities such as campaign or special offer or bundles applied against a user).
The “Subscriber Info” object has interfaces to external systems such as CRM, customer care, or provisioning modules.
Service: the object represents various services offered by the service provider (in case of Value Added Service—potentially over multiple bearers). Usage of a service by a subscriber, i.e. the fact that a subscriber consumes various services, is represented by the association between the objects. In practice, the link between the Service and Subscriber objects is used to analyse the Service Demand. Similarly, a “Service Info” object holds available Service details (whereby interfacing with the Service Provisioning module is implemented).
Platform (network element): the object represents a network element of the service delivery infrastructure (e.g. a service centre or a gateway platform). The fact that services are delivered over a multiple set of platforms is represented by the relationship between the objects. Similarly to the Service—Subscriber objects relationship, the link between the Service and Platform objects is used to determine the level of utilisation of the service delivery infrastructure components and determine optimal service configurations over the set of platforms.
Trend: To represent the dynamic aspects of various elements the Trend object is executed to track behaviour of associated objects, for example “subscriber”, “service”, “platform” and “segment” objects. Trends, or behavioural variables, could be basic and derived (to keep the objects overview diagram simple the related data objects are left out). Basic trends represent information, directly related to the usage such as “Number of messages received over a particular bearer”, “Total charged amount” per subscriber per time period or per service per time period. Various granularity levels are supported to efficiently aggregate and maintain data, e.g. hourly trends, daily trends, weekly trends. Derived trends are used to hold results of various analytical models that are typically built using a number of basic trends. Examples are: predictive trends “Growth decile score”, “Browsing propensity”, “IMS usage propensity” or “Churn Decile score” per subscriber or per segment, where each of those trends is calculated using a number of basic trends aggregated for various time periods, “Average Revenue” or “IMS usage level” per segment. Derived trends are primarily used to determine the anticipated level of the service demand (which is one of the main input parameters for service management).
In case of platform related trends, levels of resource utilisation (hardware resources, e.g. CPU or network bandwidth, or software buffers and available contexts) are monitored using the Trend objects. Those types of trends are used as input, reflecting utilisation of the service delivery infrastructure components, for the service management.
Segment: A subscriber is associated with a segment based on various behavioural characteristics, i.e. the rules defined based on Subscriber level trends. A subscriber can change segment over time although the segments themselves are relatively static.
Pyramid: an aggregation of all defined segments forms a full customer pyramid.
The combination of objects is utilized by the framework to achieve an automatic feedback loop. The following describes this in more detail.
The following objects are identified in
The Subscriber Profiles and related trend analysis (created directly based on collected CDRs: SC1 CDR (Msg ID, Orig: A, Recip: C), GW CDR (Msg ID, Orig: A, Recip: C), SC2 CDR (Msg ID, Orig: A, Recip: C) and similar CDRs, generated for other call flows, shown in
The segmentation model and an overview of subscribers allocated in segments is represented by the following objects:
Looking at the service predictive models at the segment level, the following trends are used to capture the model results (“Expected Demand SIP” and “Expected Demand GSM”—for the predicted values of the service demand levels of respective services):
Clearly, given a different behaviour of subscribers in different segments, the predictive models produce different results for individual segments (e.g. the trend followers are likely to have different dynamics of behaviour in the future than the intensive users, even if both segments have similar starting positions in terms of service utilisation levels).
Once the service level demands are understood (modelled) at a per segment level, proper prediction for the expected service demand level is done:
Consider the service delivery network components utilisation. The active service definition models and analysis of levels of utilisation of individual platforms, which rely on the performance metrics related trends, are as follows:
The initial Service delivery model definition:
The Service Delivery Infrastructure platforms characteristics:
Referring to the predicted demand level for both GSM and SIP services, which are (as outlined above):
The framework automatically determines that there is not enough capacity, allocated to serve the predicted SIP service element demand according to the current service predictive model (i.e. the maximum capacity of SC1 is 4 msg/time_period versus the expected SIP service demand being 5 msg/time_period). It also determines that there is under-utilised GSM capacity (the expected GSM service demand is 4 msg/time_period versus both SC2 and SC3 having 4 msg/time_period as their maximum throughput capacity). The framework therefore adjusts the Service predictive model from:
As such the required increased SIP service demand is met by properly utilising available delivery components, i.e. SC3, being re-configured from a GSM type of service centre to SIP. Additionally, the capacity of SC2 is increased to be able to deal with the predicted service level demand.
The configuration of individual platforms is adjusted accordingly:
Thus, the actual model objects and their values are translated into the platform configuration commands and activated on all affected platforms (e.g. the SC type becoming “SIP” is translated into the SIP SC software packages activation commands, the configured licensed throughput capacity is adjusted using appropriate provisioning commands, the GW routing configuration results in a set of routing tables related commands.). Similar to this example, where the throughput capacity is managed relying on the expected demand levels, the number of active subscribers (and resources allocated to them on individual service delivery infrastructure platforms) is controlled. Also, the invention allows control of the number of active subscribers and the resources allocated.
In the above detailed example only a very limited number of subscribers is referred to, for clarity. However, in practice there would in general be- many more subscribers, as indicated for example in
The following examples illustrate further the flexibility and extensive applicability of the framework.
Marketing Campaign Management Example
The same framework is used in the campaign management case (e.g. an advertising campaign). Refer to
The ability to dynamically adjust the configuration of service delivery infrastructure to cope with campaigns (which depending on the service adjustment can be in real-time), is an important aspect of the framework. In the given example, if a marketing campaign (to stimulate the usage of the Messaging Service over IMS/SIP—e.g. a “free IMS access for the whole month of June” offer for a set of target subscribers, selected based by the Subscriber Analysis module) is run, the GW and SC configurations are adjusted in near “real-time” mode (based on the constantly monitored uptake level of the campaign). Similarly, during the post campaign monitoring mode, the new Service Delivery components are reconfigured to accommodate the needs.
Control service usage data collection is constantly performed to validate active configurations and allows demand forecast creation based on the actual levels of utilisation, thus allowing the subscriber, service infrastructure components utilisation, and service demand models at all levels to be constantly tuned.
Thus modelling service demand levels and service delivery infrastructure utilisation enables the identification, activation of and efficient running of specific marketing campaigns to optimize the service delivery infrastructure. This can include for example campaigns to encourage/enable better utilization during off-peak hours, or for example different pricing strategies to decrease usage (or increase revenue) during peak hours.
SLA Management Example
Another example serves to illustrate the dynamic aspect of configuration control.
If all agreed QoS levels are to be implemented on the delivery network in a static manner, the total level of required resources could be higher than the actual level of available network resources. However, understanding the actual level of utilisation, based on the monitored and modelled service demand (“content propensity model” as shown on the figure), the services can be dynamically provisioned over the available platforms utilising the resources (e.g. allocated in-flight capacity and content validity periods, application network connections and their bandwidth, dedicated to a particular content provider, routing resources between the application GW, Service Centres and SS7 Gateway).
The example illustrates how the invention facilitates the management of multiple services, delivered by a set of shared service delivery infrastructure components, (where multiple co-existing services can also be managed on a single shared platform or multiple shared platforms) which is a widely deployed approach for service delivery infrastructure setup. The “??” annotation in the flows from the application gateway to SC1 and SC2 and from SC1 and SC2 to the SS7 GW in
In the QoS context, individual service requests, originated by or destined to subscribers (or originated from subscriber to VASP or originated from VASP to subscriber), for subscribers belonging to the top segments, could be granted a higher QoS than to those from/to the subscribers from the low value segments (e.g. in case of a limited license capacity or bandwidth, SMS delivery attempts could be prioritised based on dynamically provisioned subscriber characteristics, such as the segment it belongs to).
Fair Use Charging Example
Another example serves to illustrate the dynamic aspect of configuration control, coupled with direct automatic customer care interaction with a particular subscriber.
A subscriber can subscribe to a particular flat fee tariff. Flat fee tariffs can be configured associated with subscriber segments. The framework is particularly adapted to configuration of flat fee tariffs as it provides a comprehensive view of subscriber behavioural and usage characteristics. This can be used to implement fair use policies which can be associated with particular subscriber segments. Thus if a subscriber exceeds usage beyond the flat fee tariff, automatic action can be taken by the operator such as re-prioritizing of that subscribers messages in delivery queues, or adjusting the bandwidth on the IP network for a particular subscriber based on usage. The system could generate a message such as an SMS to the subscriber indicating that the subscriber has exceeded the fair use policy applicable to them.
MMS Use Case Example
Another example serves to illustrate the dynamic aspect of configuration control, coupled with direct automatic customer care interaction with a particular subscriber.
If a subscriber has normally been a high MMS user and suddenly MMS use ceases, the framework is adapted to detect such a drop in usage. Automatic Action can be taken such as the system proactively generating an SMS to the subscriber indicating that customer care help is available if for example there are handset issues. Additionally, a work item for an customer care person to follow up on can be generated.
The framework dynamically combines collect, analyse, and act phases whereby the full service management cycle is performed in the main in a real-time manner (although depending on the nature of the “act” phase this phase can be in real time, near real time or over a longer period).
The framework advantageously is also very suitable as a means or input for network capacity planning, where for example subscriber segmentation and the behaviour of subscribers within a segment can be used for network capacity planning. Depending on the required capacity changes this can be a long term process.
In addition, the invention is applicable to service management of a single service on a dedicated platform, or to service management of one or multiple co-existing services on a single shared platform or multiple shared platforms.
The invention is not limited to the embodiments described herein but may be varied in construction and detail. For example, the models may be implemented other than as objects, for example, via functional programming. Also, the invention may be implemented with actions involving technology stacks other than GSM and IMS, such as CDMA or TDMA. Also, the invention may be implemented with technology stacks required to deliver content such as video streaming to mobile devices. Further, actions other than those described may be implemented by the framework. Also, although the description of the invention has focused on SMS-related services, it is equally applicable to MMS, IM, mobile browsing, and other related services or technologies.
Filing Document | Filing Date | Country | Kind | 371c Date |
---|---|---|---|---|
PCT/IE2008/000108 | 10/23/2008 | WO | 00 | 8/19/2010 |
Number | Date | Country | |
---|---|---|---|
60996026 | Oct 2007 | US |