WO2013149870A1

WO2013149870A1 - Method and system for managing actions implemented on a network element within a telecommunications network

Info

Publication number: WO2013149870A1
Application number: PCT/EP2013/056242
Authority: WO
Inventors: John Keeney; Sven VAN DER MEER; Gabriel HOGAN
Original assignee: Telefonaktiebolaget L M Ericsson (Publ)
Priority date: 2012-04-05
Filing date: 2013-03-25
Publication date: 2013-10-10

Abstract

A method for recommending actions to a management agent (20) for implementation on a telecommunications network (19), the method including the steps of: generating a list of actions from an action catalogue (12), each of said actions being a candidate for implementation on the network element (22); ranking the list of actions in accordance with one or more ranking parameters (17) indicative of the suitability or perceived of said actions; and providing at least a portion of the ranked list of actions to a management agent (20) arranged to manage said network element (22) to enable the management agent (20) to select and implement one, none or some of the actions. Also disclosed is a recommender system (10) for performing the above-described method.

Description

METHOD AND SYSTEM FOR MANAGING ACTIONS IMPLEMENTED ON A NETWORK ELEMENT WITHIN A TELECOMMUNICATIONS NETWORK

Technical Field

The present invention relates to a method and system for managing actions implemented on a network element within a telecommunications network, and particularly, but not exclusively, to a method and system for recommending certain actions to a network element to enable the network element or associated management agent to select one or more actions based on the recommendations.

Background

Management processes in telecommunications networks are typically planned and implemented manually. Manual implementation requires a user to write a series of workflows and manually target one or more workflows to individual telecommunication network characteristics. Accordingly, a manually authored management process is unable to adapt to the evolution of the characteristics in the behavior. In particular, within a given network there is typically one 'best practice' response to a given problem and the management system cannot deviate from this response or learn from evolving expertise about the network or other 'best solutions' employed in other telecommunication networks. It will be appreciated that the process of manually implementing management process and updating the 'best practice' response is time-consuming and an inefficient use of human resources.

One known improvement to this manual approach is the implementation of management processes through rules in the form of statically encoded workflows, codebooks and runbooks, or in an extended for as policy rules. The use of rules enables a basic level of automated response to incidents in the managed telecommunications network. Whilst providing an advantage over wholly manually implemented management processes, the rule-based, policy-based or workflow- based approach nonetheless suffers a number of the problems afflicting manually authored systems such as the inability to adapt to the evolution of the characteristics in the behavior of the network.

Furthermore, the use of rule-based, policy-based, or workflow-based systems for telecommunications network management has several other serious disadvantages. Firstly, in traditional policy-, rule- and workflow-based management, the action selected by the management system may be directly and immediately applied without the intervention of a human operator, thus removing the ability for a human operator to have fine-grained authority or control over the managed system. Accordingly, the expertise of the user cannot be exploited to optimize the management of the telecommunications network. Secondly, rule, policy- and workflow-based systems require a careful selection of the rule language and a constant maintenance of the defined and used rules, including contradictions and conflicts in the maintained rule set; even where the rules are automatically generated, their maintenance constitutes a major overhead. The sophistication of the rule or policy based approach can be improved by the addition of an inference engine for management decisions. The combined system of an inference system and rules is known as an expert system and is designed to emulate the decision making of human experts. This is achieved through the use of logic such as propositional logic, first order logic, temporal logic, fuzzy logic or the like. However, expert-based systems do not avoid the problems described above in relation to rule-based, policy-based and work-flow based systems. In addition, the logic employed in the inference engine constraints the ability to infer solutions from the rule (or knowledge) base. It has been found that expert-based systems are also brittle in terms of their ability to cope with unknown cases where existing rules or knowledge is not available.

More complex automated self-management approaches such as Artificial Intelligence (Al) techniques for telecommunications network management have been proposed. One example of this approach is described in US Patent Number 5666481 , which discloses a Case-Based Reasoning (CBR) approach comprising a trouble ticket data structure to describe and store communications network faults and, in the event of a new incident, search through the stored trouble tickets to select trouble tickets based on the similarity of the reported incident to the previous incidents encoded in the trouble tickets. Other Al techniques such as Artificial Neural Networks (ANN) and Bayesian Belief Networks (BBN) have also been proposed for fault filtering, root case analysis, failure prediction and correlation in telecommunications network management. However, all of the known complex automated self-management have serious deficiencies and have not been widely adopted. For example, whilst CBR, ANN, and BBN approaches are able to progressively learn to rank historical incident reports and the defined solutions for those incidents, this improved ranking is only ever based on an increasingly fine-grained comparison of incident feature or parameters; Al systems do not automatically learn to rank proposed solutions based on uptake by the management agent. Accordingly, such techniques are useful for fault filtering, correlation, root cause analysis, failure prediction and the like, but are not typically adapted to present a solution to a fault.

Summary

In accordance with the present invention as seen from a first aspect, there is provided a method for assisting management of network element within a telecommunications network, the method comprising the steps of:

(a) generating a list of actions, each of said actions being a candidate for implementation on the network element;

(b) ordering the list of actions in accordance with one or more ranking parameters;

(c) providing at least a portion of the ordered list of actions to said network element or to a management agent arranged to manage said network element.

The ranking parameters may be indicative of suitability of said actions, which may be a perceived suitability based on available knowledge and/or rules. The suitability may be determined in accordance with past behavior of the telecommunications network, and in particular may be determined in accordance with past behavior of the network element. The network element or the management agent arranged to manage the network element may review the ordered list of actions and select the appropriate action for implementation on the network element. The choice of the network element or management agent arranged to manage the network element may be wholly or partially influenced by the order of the list. One advantage of the latter case is that the management agent maintains oversight and control of the network, thereby allowing for special cases and exceptions.

The step of generating a list of actions may comprise consulting an action catalogue. The step of providing at least a portion of the ordered list of actions may comprise providing the top n actions in the ordered list, where n may be any number other than one. Alternatively, the step of providing at least a portion of the ordered list of actions may comprise providing the top m% of actions in the list, where m may be any number other than 100. Alternatively, the step of providing at least a portion of the ordered list of actions may comprise providing the full list of actions.

The one or more ranking parameters may be determined in accordance with past behavior of the network element. The past behavior may comprise past behavior of the network element in relation to the action, and in particular may comprise past approval of the action by the network element. Said past approval may comprise explicit approval such as a notification of approval received from the management agent. Alternatively, or in addition thereto, said past approval may comprise implicit approval, which may comprise past adoption of the action. Alternatively, or in addition thereto, the one or more ranking parameter(s) relating to a given action may be determined in accordance with past disapproval of the action by the network element. Said past disapproval may comprise explicit disapproval such as a notification of disapproval sent by the management agent arranged to manage the network element. Alternatively, or in addition thereto, said past disapproval may comprise implicit disapproval, which may comprise non-adoption of the action by the network element in spite of the action being provided to the network element as part of the ordered list of actions.

Alternatively, or in addition thereto, the past behavior may comprise past behavior of the network element in relation to other actions similar to said action. As used herein, the term similar action is defined as an action having a 'distance' to said action below a defined or identified threshold, where 'distance' is a quantitative measure of the similarity between actions. In particular, the one or more ranking parameters relating to a given action may be determined in accordance with past approval of a similar action by the network element, which may comprise implicit and/or explicit approval. Alternatively, or in addition thereto, the one or more ranking parameter(s) relating to a given action may be determined in accordance with past disapproval of a similar action by the network element, which may comprise implicit and/or explicit disapproval. Advantageously, the inclusion of past behavior of the network element in relation to similar actions provides a larger data pool from which to calculate the one or more ranking parameters and improves the accuracy of the ordering of the list of actions.

A similar action may comprise an action having at least one characteristic in common with said action. The at least one characteristic in common may comprise one or more of:

(i) general action category, for example: escalate, hold, ignore, forward, report;

(ii) action target i.e. the network entity/entities that the action changes/impact on, for example: a particular network element, a particular network configuration, a particular service, a particular user;

(iii) action subject i.e. person/entity that issues the action/owns the action, for example: first level support, second level support, incident manager;

(iv) management category, for example: performance, security, accounting, configuration, fault;

(v) expense of the action, for example: on-site maintenance has a high expense, software updates has a moderate expense, protocols or reports have a low expense;

(vi) time required to issue the action, for example: as soon as possible, now, soon, later, never;

(vii) perceived impact an action has on operations, for example: critical update, non-critical update;

(viii) the time it will take the action to complete, for example: simple reports are fast while site maintenance can take a significant length of time.

Alternatively, or in addition thereto, the one or more ranking parameters relating to a given action may be determined in accordance with past behavior of other network elements having at least one characteristic in common with the network element. Said past behavior may comprise past behavior of other network elements in relation to the action. In particular, the one or more ranking parameters relating to a given action may be determined in accordance with past approval of the action by another network element sharing at least one characteristic with the network element, which may comprise implicit and/or explicit approval. Alternatively, or in addition thereto, the one or more ranking parameter(s) relating to a given action may be determined in accordance with past disapproval of the action by a network element sharing at least one characteristic with the network element, which comprise implicit and/or explicit disapproval. The characteristic shared between network elements may comprise physical location, physical environment, hardware, software and/or functionality.

Alternatively, or in addition thereto, the past behavior of other network elements having at least one characteristic in common with the network element may comprise past behavior of another network element having at least one characteristic in common with the network element in relation to actions having at least one characteristic in common with the action. In particular, the one or more ranking parameters relating to a given action may be determined in accordance with past approval of another action sharing at least one characteristic with the action by another network element sharing at least one characteristic with the network element, which may comprise implicit and/or explicit approval. Alternatively, or in addition thereto, the one or more ranking parameters relating to a given action may be determined in accordance with past disapproval of another action sharing at least one characteristic with the action by a network element sharing at least one characteristic with the network element, which comprise implicit and/or explicit disapproval.

Alternatively, or in addition thereto, the one or more ranking parameters relating to a given action may be determined in accordance with one or more rules, which may be accompanied by an inference engine for management decisions to form an expert system. The rules may comprise an initial definition of best practice.

The action catalogue and/or the one or more ranking parameters may be updated periodically. Alternatively, or in addition thereto, the action catalogue and/or the one or more ranking parameters may be updated in response to a change in one or more of the criteria upon which the suitability is determined. For example, the response of the management agent and/or network element to a list of actions provided to it may be used to update the one or more ranking parameters relating to the actions in said list and/or actions other actions having at least one characteristic in common with one or more actions in said list. The response of the management agent and/or network element to a list may be tagged with the time and date of the response, and more recent responses may be allocated a greater weight than older responses.

In one embodiment, the one or more ranking parameters may be initially defined by one or more rules and may subsequently evolve in response to behavior of the network element and/or other network elements sharing at least one characteristic with the network element. The evolution may be based on acceptance or rejection of actions belonging to previous lists provided to the network element or the management agent arranged to manage the network element.

The action catalogue may evolve in response to said past behavior. New actions may be added to the action catalogue and/or existing actions may be amended or deleted.

The one or more ranking parameters may be determined by an entity other than the network element or management agent arranged to manage the network element. For example, the one or more ranking parameters may be determined by the management system.

The actions may be management actions.

The actions may be in response to a given incident. In such an embodiment, the actions may be alternative responses to the incident or may be complementary responses to the incident. The method may further comprise the step of detecting an incident and extracting one or more incident parameters from the incident. This step may be performed before step (a) above. The action catalogue may be queries in accordance with the extracted incident parameters.

The method may be implemented on a manager such as a Network Management Station (NMS).

In accordance with the present invention, as seen from a second aspect, there is provided a method executed on a management agent arranged for managing a network element, the method comprising the steps of:

(a) receiving an ordered list of actions;

(b) selecting an action belonging to the list partially in accordance with the order of the list and partially in accordance with one or more rules specific to the network element and/or the management agent arranged to manage the network element; and,

(c) implementing the action on the network element.

The method in accordance with the second aspect of the present invention may be implemented on the network element and/or on a management agent arranged to manage the network element. It will be appreciated that a method in accordance with the second aspect of the present invention provides the network element and/or the management agent with the opportunity to implement an action that is not positioned first in the ordering of the list. Some instances in which a management agent may wish or need to overrule an order provided to it are provided below by way of example:

(i) the ordered list of actions may violate best common practice of the management agent's experience;

(ii) the ordered list of actions might collide with other actions taken by the management agent at the same time;

(iii) general operational policies of the management agent might have evolved but the ranking parameters used to create the ordered list may not have been amended to account for the change in the general operational policies;

(iv) other contextual information that was not accounted for in the ranking parameters used to create the ordered list may render the recommendation invalid; such information may comprise scheduled maintenance work that was not accounted for in the ranking parameters used to create the ordered list.

The method in accordance with the second aspect may be performed after the method in accordance with the first aspect, the ordered list provided in step (c) of the method in accordance with the first aspect being equivalent to the ordered list received in step (a) of the method in accordance with the second aspect. The method in accordance with the second aspect may comprise updating the one or more ranking parameters in accordance with the action selected in step (b) in the method in accordance with the second aspect and/or other actions belonging to the ordered list but not selected in step (b) in the method in accordance with the second aspect. The method may comprise selecting more than one action belonging to the list and implementing all of the selected actions. Alternatively, the step of selecting an action may comprise selecting one action only and ignoring and/or rejecting all other actions in the list.

The method may comprise the step of selecting an action not belonging to the list, which may replace step (b) above. This step may be appropriate if the network element or the management agent arranged to manage the network element does not consider the actions provided to it to be suitable. Alternatively, the method may comprise the step of selecting an action not belonging to the list, which may be performed in addition to step (b) above.

In accordance with the present invention as seen from a third aspect, there is provided a recommender system for managing a network element within a telecommunications network, the system comprising:

an action selector for generating a list of actions, each of said actions being a candidate for implementation on the network element;

an action ranker for ordering the list of actions in accordance with one or more ranking parameters;

an output for providing at least a portion of the ordered list of actions to said network element or to a management agent arranged to manage said network element.

The ranking parameters may be defined in accordance the suitability or perceived of said actions to the network element. The suitability may be determined in accordance with past behavior of the telecommunications network, and in particular may be determined in accordance with past behavior of the network element.

The system may comprise an input for receiving an indication that an incident has occurred. In this embodiment, the actions are potential responses to said incident, which may be mutually exclusive or may be executable in conjunction with one another. The input may be received from the managed telecommunications network. Alternatively, or in addition thereto, the input may be received from a management agent, the management agent being human or automated. The system may comprise a list manipulator for manipulating the list of actions. The list manipulator may be arranged for selecting certain actions belonging to said list, which may comprise the top n actions within the ordered list where n is an integer not equal to one, or the top or the top m% of actions within the ordered list, where m may be any number. The actions selected by the list manipulator may be those output to the network element or management agent.

The system may comprise an action catalogue comprising an inventory of actions, the action selector being arranged to select actions from the action catalogue.

The system may comprise an updater arranged for updating the action catalogue and/or the ranking parameters. The updater may be arranged to update the action catalogue and/or the ranking parameters periodically. Alternatively, the updater may be arranged to update the action catalogue and/or the ranking parameters in accordance with a response of the network element or management agent arranged to manage the network element to one or more of said actions.

The system may be arranged for managing a plurality of network elements within the telecommunications network.

The system may be implemented as software. Alternatively, or in addition thereto, the system may be implemented as discrete hardware circuitry. Alternatively, or in addition thereto, the system may be implemented with discrete hardware components. Alternatively, or in addition thereto, the system may be implemented as an integrated chip or may be implemented as an arrangement of chip modules, which may be controlled by a software routine or a program stored in a memory. Alternatively, or in addition thereto, the system may be implemented as a signal processing device or computer device. In accordance with the present invention, as seen from a fourth aspect, there is provided a management agent for managing a network element, the management agent comprising:

an input for receiving an ordered list of actions from a recommender system;

a memory for storing one or more rules; and, a processor for selecting an action belonging to the list partially in accordance with the order of the list and partially in accordance with one or more rules specific for the management agent arranged to manage the network element. The management agent may comprise an output for outputting instructions to the network element to implement the action selected by the processor.

In accordance with the present invention, as seen from a fifth aspect, there is provided a telecommunications network comprising a recommender system as hereinbefore described and a management agent arranged for managing a network element, the recommender system being arranged for providing said output to said management agent.

The management agent may be arranged for selecting one or more of said actions. Alternatively, or in addition thereto, the management agent may be arranged for selecting one or more actions not belonging to said list.

The network may comprise an actioner for causing the selected action(s) to be invoked on the network element. The management agent may be arranged to control the actioner.

The management agent may be arranged to notify the updater of the response of the management agent to the output from the recommender system. In particular, the management agent may be arranged to notify the updater of the action(s) selected by the management agent. Alternatively, or in addition thereto, the management agent may be arranged to notify the updater of the action(s) rejected by the management agent.

Brief description of the drawings

An embodiment of the present invention will now be described by way of example only and with reference to the accompanying drawings, in which:

Figure 1 is a high level schematic block diagram illustrating a recommender system in accordance with an embodiment of the invention; Figure 2 is a high level schematic block diagram of a telecommunications network managed by one or more management agents, assisted by the recommender system illustrated in Figure 1 , the components that form part of the recommender system being shown within the dashed box; and,

Figure 3 is a high level flowchart illustrating a method according to an embodiment of the invention.

Detailed description

With reference to drawings, the particulars shown are by way of example only for illustrating alternative embodiments of the invention. No attempt is made to show structural details of the invention in more detail than is required for fundamental understanding of the invention. The drawings and description make apparent to those skilled in the art how the invention may be embodied in practice.

The present description explains certain embodiments of the invention but it will be appreciated that the present invention is applicable to other embodiments or being practiced or carried out in various ways. With particular reference to Figures 1 and 2 of the drawings, there is provided a recommender system 10 for assisting management of a telecommunications network 19. In particular, the recommender system 10 is arranged for presenting a ranked list of recommended actions in response to indications of incidents occurring in the managed telecommunications network 19. The ranked list of recommended actions is received by a management agent 20 comprising a processor 25 for selecting an action partially in accordance with the ranked list of recommended actions and partially in accordance one or more rules stored within a memory 24 of the management agent 20. The approval and/or disapproval of certain actions by the management agent 20 is communicated to the recommender system 10 in order to refine and improve future rankings.

In detail, the recommender system 10 comprises an input in the form of an indication receiver 1 1 arranged for receiving an indication that an incident has occurred within the managed telecommunications network 19. The indication may be received from the managed telecommunications network 19 itself. Alternatively or additionally, the indication may be received from the management agent 20, which may be human or automated.

An action catalogue 12 is provided within the recommender system 10, which contains an inventory of actions. The action catalogue 12 is structured such that each action is stored alongside an indication of its suitability in relation to given incidents. This suitability is determined on the basis of the response of the management agent 20 to the action when the action was previously recommended by the recommender system 10. The action catalogue 12 is initially populated from a catalogue of best practice within a best practice store 21 via a best practice persister and loader 13. The catalogue of best practice may be derived from the persisted state of another recommender system (not shown) using a using the best practice persister and loader of that system. Alternatively, the catalogue of best practice may be derived from best-practice workflow specifications such as the Ericsson Managed Services Organization MS-TOP library.

Entries within the action catalogue 12 may be amended or new entries added as the recommender system 10 learns from the decisions of the management agent 20. For example, a new entry may be added if the management agent 20 chooses to respond to an incident by executing an action that was absent from the action catalogue 12 at that time. Amendment or augmentation of the action catalogue 12 is implemented by an updater 14, which is arranged to receive notification of the response of the management agent 20 to the recommended actions provided by the recommender system 10. In particular, the updater 14 is arranged to receive notification of the action(s) selected by the management agent 20 and/or the actions(s) rejected by the management agent 20. The response of the management agent 20 to the recommended actions presented to it is tagged with the time and date of the response, and more recent responses may be allocated a greater weight than older responses. This enables older selections to decay as the managed telecommunications network 19 evolves. In the preferred embodiment, updating of the action catalogue 12 is performed automatically upon the response of the management agent 20 to the recommended actions.

An action selector 15 is arranged to receive a signal from the indication receiver 1 1 that an indication of an incident has been received. The action selector 15 is adapted to extract one or more incident parameters and select certain actions from the action catalogue 12 based on these incident parameters. It is envisaged that the action selector 15 performs an item-to-item comparison between first and second lists, the first list comprising signals from the network and second list comprising known actions. This comparison can be done using a cosine-based algorithm, a conditional- probability algorithm or any other algorithm suitable for this purpose. One example algorithm implemented on the action selector 15 may comprise data-mining trouble tickets that provide information about signals and associated (previously selected) actions and define a similar matrix. Another example algorithm may comprise implementing rule-based actions based on the best common practice of the management agent 20, the algorithm further comprising the step of evolving the rules in accordance with user feedback or the like. The skilled person will appreciate that other algorithms are possible, including those relating to expert systems and machine learning.

The actions selected by the action selector 14 form a list of candidate actions, which is then passed to an action ranker 16.

The action ranker 16 is arranged to rank the list of candidate actions according to the suitability of each action in relation to the incident and the network element(s) 22 upon which the action is to be implemented. The suitability is parameterized by one or more ranking parameters 17, which are initially loaded from the best practice store 21 via the best practice persister and loader 13. The action ranker 16 is provided with appropriate models associated with relationships between items in first and second lists, the first and second lists associated with the action ranker 16 being different to those associated with the action selector 14. The first list of the action ranker 16 comprises actions belonging to the list of candidate actions and the second list of the action ranker 16 comprises users' preferences or learned user selections. It is envisaged that the action ranker 16 comprises an algorithm that considers all possible combinations of items up to a defined or identified size. The action ranker 16 then selects a subset of these combinations as higher-order similarities in order to prevent exponential growth.

The ranking parameters 17 are updated via the updater 14 according to the response of the management agent 20 to actions recommended in relation to past incidents. In particular, the ranking parameters 17 are updated according to the previous approval and/or disapproval of actions in relation to past incidents. For example, if the management agent 20 disapproves a given action that is presented to it by the recommender system 10 in response to a given incident then the ranking parameters 17 are amended such that that particular action will in the future be ranked lower in relation to that particular incident or other incidents similar to that incident. In addition, other actions similar to that action may in the future be ranked lower in relation to that particular incident or other incidents similar to that incident. The disapproval by the management agent 20 may be explicit, namely the management agent may notify the recommender system 10 that it disapproves of the action. Alternatively or additionally, the disapproval by the management agent may be implicit, namely the management agent does not adopt the action when the action is presented to it by the recommender system 10. Conversely, if the management agent 20 approves a given action that is presented to it by the recommender system 10 in response to a given incident then the ranking parameters 17 are amended such that that particular action will in the future be ranked higher in relation to that particular incident or other incidents similar to that incident. Again, the approval may be explicit, namely the management agent may notify the recommender system 10 that it approves of the action. Alternatively or additionally, the approval may be implicit. For example, the management agent implicitly approves the action by adopting the action.

The response of the management agent 20 to the recommended actions presented to it is tagged with the time and date of the response, and more recent responses may be allocated a greater weight than older responses. In the preferred embodiment, updating of the ranking parameters 17 is performed automatically upon the response of the management agent 20 to the recommended actions.

The recommender system 10 further comprises an output in the form of a recommendations presenter 18 for providing the ranked list of selected actions, or a portion thereof, to the management agent 20.

It will be appreciated that the system may be implemented as software, discrete hardware circuitry, discrete hardware components, a single integrated chip, an arrangement of chip modules, a signal processing device or computer device. It will also be appreciated that the present invention is not limited to applications with only one management agent 20; input from different network management agents is within the scope of the present invention and will allow the recommender system 10 to evolve the recommendation list and ranking to widen the scope of response to a telecommunication network incident and evolve the 'best practice' based on inputs from multiple network management agents 20.

With reference to Figure 3 of the drawings, there is illustrated a method 100 of managing a telecommunications network 19, the method utilizing the recommender system 10 illustrated in Figure 1 .

The method commences at step 101 by loading best practice rules and/or policies in the form of a catalogue of best practice from the best practice store 21 . This step of loading best practice rules and/or policies is implemented by the best practice persister and loader 13. The catalogue of best practice may in variations of the preferred embodiment be derived from the persisted state, such as the action catalogue and/or ranking parameters, from another recommender system (not shown) using a best practice persister and loader of that system. In an alternative variation, the catalogue of best practice may be derived from best-practice workflow specifications such as the Ericsson Managed Services Organization MS-TOP library.

The method continues at 102 with the step of initializing the action catalogue 12 and ranking parameters 17 in accordance with the catalogue of best practice.

In an alternative embodiment (not shown), the steps of loading best practice rules and/or policies are absent and the method commences at 103 with the step of waiting for an indication of an incident. In variations of the present invention, the step of waiting for an indication of an incident 103 comprises actively monitoring the telecommunications network 19 or one or more network elements 22 therein. In an alternative variation, the step of waiting for an indication of an incident 103 comprises waiting to receive a signal from the telecommunications network 19, one or more network elements 22 therein and/or an entity arranged for monitoring all or part of the telecommunications network 19 such as an Element Management System (EMS) (not shown). The recommender system 10 may receive notification of an incident via an alarm notification, which comprises a 'probable cause' field containing information about the probable cause of the incident. The codes within the probable cause field may be standardized, for example ITU, ETSI, 3GPP standardizations. Alternatively, or in addition thereto, vendors and operators also use their own causes or cause codes within the probable cause field.

Incidents may relate to properties of the system or physical problems affecting the system. Examples of such incidents include but are not limited to: intrusion detected; loss of signal/frame/transmission/redundancy/synchronization; delayed information; software error; temperature out of allowed bounds; transmit failure; receive failure; performance degradation; out of memory; door open (on a site); call setup failure; power problem/failure; host unreachable.

Other incidents might be more related to users and services. Examples of such incidents include but are not limited to: handover failure; registration failure on communication nodes; call drop; congestion; ping pong problems such as mobile telephones continually registering and re-de-registering at multiple cells; heartbeat failure i.e. problems that occur when a node is not dead but also does not send any information or carry traffic; authentication errors of users and/or user equipment.

Another class of incidents may be those related to service assurance. Examples of such incidents include but are not limited to: KPI violation or MOS value degradation such as noticeable decreased performance; SLA violation in which contractually agreed network behavior is violated.

Yet another class of incidents may be those related to the introduction of new equipment, which may send any type of unexpected signal to the management system.

The skilled person will appreciate that the present invention is not limited to providing a list of actions in response to the above list of incidents; the invention may also be applicable to other incidents not listed above. Once an indication of an incident is received, the parameters of the incident are extracted by the recommender system 10 at 104. The extracted parameters are used to query the action catalogue 12 at step 105 to find actions that have are considered suitable in response to at least one of the incident parameters. In one variation of the present invention the suitability of a given action to a given incident parameter is parameterized by a suitability value and the action is selected from the action catalogue 12 if the suitability value is above a certain threshold for at least one of the incident parameters. A list of candidate actions is compiled by the action selector 15 at 106. The list of candidate actions comprises all of the actions that are considered suitable in accordance with the specific criteria of the recommender system 10. For example, if the suitability value of an action in relation to an incident parameter extracted 104 exceeds the threshold suitability value described above then the action is added to list of candidate actions.

Once compiled, the list of candidate actions is ranked at step 107 by the action ranker 16 in accordance with ranking parameters 17, the ranking parameters 17 being derived from guidelines of best practice and/or the historical approval or disapproval of actions in response to similar past incidents. In one variation of the present invention the ranking may be in accordance with the suitability value described above for a certain incident parameter. Alternatively, or in addition thereto, the ranking may be in accordance with the sum of suitability values for all incident parameters extracted at 104.

Once the list of candidate actions is ranked, the top n actions of the ranked list are selected at step 108, where n is an integer such as 10. In one variation of the present invention, the integer n may be chosen during the initial set-up of the recommender system 10. In an alternative variation, the best practice store 21 may comprise the best practice value for n. In an alternative embodiment of the present invention, step 108 may be replaced with the step of selecting the top m% of actions. It will be appreciated that the top n actions or the top m% of actions could comprise all of the actions within the list of candidate actions. The recommender system 10 presents the top n actions within the ranked list to the management agent 20 at step 109, the ranking of the list being maintained within the reduced list. Accordingly, the management agent 20 receives the ranked top n actions from the recommender system 10 at step 1 10. At step 1 1 1 , the management agent 20 reviews the ranked top n actions and selects one, some, or none of the recommended actions. It is envisaged that this selection will be based partially on the ranking attributed to the actions by the recommender system 10 and partially on criteria specific to management agent 10 and/or the network element(s) involved in the incident and/or action.

If the management agent 20 does not consider any of the recommended actions to be appropriate then it may select none of the actions presented to it and instead select an action that does not form part of the list presented to it at step 1 12. Alternatively, if more than one action may be applied in response to the incident then the management agent 20 may select one or more actions from the ranked top n actions presented to it at step 1 12 in addition to selecting one or more actions that do not form part of the list presented to it at step 109.

In this way, whilst the recommender system 10 identifies recommended solutions, it does not by default automatically apply those solutions; the management agent 20 retains the ultimate decision to ignore or select from the recommendations presented. Thus, the management agent 20 maintains overall control of the managed telecommunications network, but with the assistance of the Recommender System 10. This allows for evolving best practice to be presented to the management agent while still allowing and supporting special cases and exceptions.

Once the management agent has selected one or more actions, it causes these actions be invoked on the managed telecommunications network 19 or certain network element(s) 19 therein. This is achieved by way of an actioner 23.

The management agent 20 notifies the updater 14 of the actions(s) that it has chosen in response to the incident at step 1 13. The updater 14 may amend the action catalogue 12 and/or the ranking parameters 17 according to the actions chosen by the management agent 20 in steps 1 1 1 and 1 12. For example, if at step 1 12 the management agent 20 selects an action that was not recommended to it and it is found that this action is missing from the action catalogue 12 then the updater may add this action to the action catalogue 12. As another example, if the management agent 20 rejects the action ranked top by the recommender system 10 then the updater 14 may reduce the suitability value of that action in relation to all or some of the incident parameters of the given incident. This step of updating the action catalogue 12 and/or ranking parameters 17 enables the recommender system 10 to adapt in response to network agent interactions, allowing older selections to decay as the managed telecommunications network 19 evolves.

From the foregoing therefore, it is evident that the present invention provides an effective system and method for selecting actions to be implemented on a managed telecommunications network.

Claims

1 . A method for assisting management of a network element within a telecommunications network, the method comprising the steps of:

A method as claimed in claim 1 , wherein the ranking parameters are indicative of suitability of said actions.

A method as claimed in claim 1 or claim 2, wherein the step of generating a list of actions comprises consulting an action catalogue.

A method as claimed in any preceding claim, wherein the step of providing at least a portion of the ordered list of actions comprises providing the top n actions in the ordered list, where n may be any number greater than one.

A method as claimed in any preceding claim, wherein the one or more ranking parameters are determined in accordance with past behavior of the network element or other network element sharing at least one characteristic with the network element.

A method as claimed in claim 5, wherein the one or more ranking parameters are determined in accordance with past behavior of the network element, said past behavior comprising one or more of:

(i) past adoption of the action by the network element;

(ii) past rejection of the action by the network element;

(iii) past adoption of another action sharing at least one characteristic with the action by the network element;

(iv) past rejection of another action sharing at least one characteristic with the action by the network element.

A method as claimed in claim 5 or claim 6, wherein the one or more ranking parameters are determined in accordance with past behavior of another network element sharing at least one characteristic with the network element, said past behavior comprising one or more of:

(i) past adoption of the action by another network element sharing at least one characteristic with the network element;

(ii) past rejection of the action by another network element sharing at least one characteristic with the network element;

(iii) past adoption of another action sharing at least one characteristic with the action by another network element sharing at least one characteristic with the network element;

(iv) past rejection of another action sharing at least one characteristic with the action by another network element sharing at least one characteristic with the network element.

A method as claimed in any preceding claim, wherein the one or more ranking parameters are determined in accordance with one or more rules.

A method as claimed in claim 8 as appended to any one of claims 5 to 7, wherein said one or more ranking parameters are initially defined by one or more rules and subsequently evolved in response to said past behavior.

A method as claimed in any one of claims 5 to 7 as appended to claim 3, wherein the action catalogue is evolved in response to said past behavior.

A method as claimed in any preceding claim, further comprising the step of detecting an incident and extracting one or more incident parameters from the incident.

A method as claimed in any preceding claim, wherein the method is implemented on a manager such as a Network Management Station.

A method executed on a management agent arranged to manage a network element, the method comprising the steps of:

(a) receiving an ordered list of actions;

(b) selecting an action belonging to the list partially in accordance with the order of the list and partially in accordance with one or more rules specific for the management agent arranged to manage the network element; and, (c) implementing the action on the network element.

A method as claimed in claim 13, wherein the method comprises selecting more than one action belonging to the list and implementing all of the selected actions.

A method as claimed in claims 13 or claim 14, wherein the method comprises the step of selecting an action not belonging to the list.

A method for managing a network element, the method comprising the steps of:

(a) a recommender system generating a list of actions, each of said actions being a candidate for implementation on the network element;

(b) the recommender system ordering the list of actions in accordance with one or more ranking parameters;

(c) The recommender system providing at least a portion of the ordered list of actions to a management agent arranged to manage said network element;

(d) the management agent receiving an ordered list of actions;

(e) the management agent selecting an action belonging to the list partially in accordance with the order of the list and partially in accordance with one or more rules specific for the management agent arranged to manage the network element; and,

(f) implementing the action on the network element.

A method as claimed in claim 16, further comprising the step of updating the one or more ranking parameters in accordance with the action or actions selected.

A recommender system for assisting management of network element within a telecommunications network, the system comprising:

an action ranker for ordering the list of actions in accordance with one or more ranking parameters; an output for providing at least a portion of the ordered list of actions to said network element or to a management agent arranged to manage said network element.

A system as claimed in claim 18, further comprising an input for receiving an indication that an incident has occurred.

A system as claimed in claim 18 or claim 19, further comprising a list manipulator for manipulating the list of actions.

A system as claimed in claim 20, wherein the list manipulator is arranged for selecting certain actions belonging to said list, said actions selected by the list manipulator defining said at least a portion of the ordered list of actions.

A system as claimed in any of claims 18 to 21 , further comprising an action catalogue, the action catalogue comprising an inventory of actions and the action selector being arranged to select actions from the action catalogue.

A system as claimed in any of claims 18 to 22, further comprising an updater for updating the ranking parameters in accordance with a response of the network element or management agent arranged to manage the network element to one or more of said actions.

A system as claimed in claim 23, wherein the updater is arranged to update the action catalogue in accordance with a response of the network element or management agent arranged to manage the network element to one or more of said actions.

A management agent for managing a network element, the management agent comprising:

an input for receiving an ordered list of actions from a recommender system;

a memory for storing one or more rules; and,

a processor for selecting an action belonging to the list partially in accordance with the order of the list and partially in accordance with one or more rules specific for the management agent arranged to manage the network element.

26. A management agent according to claim 25, further comprising an output for outputting instructions to implement the action selected by the processor.

27. A telecommunications network comprising the recommender system of any one of claims 20 to 26 and a management agent of claim 25 or claim 26, the recommender system being arranged for providing said at least a portion of the ordered list of actions to the management agent.

28. A network as claimed in claim 27, wherein the management agent is arranged for selecting one or more of said actions provided by the recommender system and/or one or more actions not provided by the recommender system.

29. A network as claimed in claim 28, further comprising an actioner for causing the one or more selected actions to be invoked on the network element.

30. A network as claimed in claim 28 as appended to claim 23 or claim 29 as appended to claim 23, wherein the management agent is arranged to notify the updater of the one or more selected actions.