WO2014008915A1 - Network management systems for controlling performance of a communication network - Google Patents

Abstract

There is provided a network management system (13) for a communication network (1). The communication network (1) has one or more sub-networks (2) each being managed by a sub-network management system (3'). The network management system (13) comprises a KPI calculator (20) for receiving performance data from said sub-network management system (3'), and for determining KPI values on basis of the performance data. The system (13) also comprises a KPI comparator module (22) configured for receiving said KPI values from said calculator (20), comparing said KPI values with KPI target values and outputting identifiers of off-target KPIs not meeting their KPI target values. A controller module (24) is present for receiving said identifiers of said off-target KPIs, and selecting a selected rule set based on the off-target KPIs, and outputting a set indicator of said selected rule set. Also a rule set repository (26) is present for storing a plurality of rule sets, each of said rule sets being associated with a rule set identifier, and for providing an output rule set out of said plurality of rule sets for which the rule set identifier matches said set indicator of said selected rule set, to each of said sub-network management systems (3').

Description

Network management systems for controlling performance

of a communication network

Technical field

The invention relates to communication networks, and more specifically to network management systems for controlling performance of these networks. It also relates to methods of managing such networks.

Background

Modern communication networks are a conglomerate of interacting sub-networks. Users of the communication network attach with their user terminals to the communication network and the communication network provides connection to other user terminals or provides communication services to a user via the user terminal. User terminals are attached to the communication network via access (sub) networks. Access networks are based on an access technology. Access technology can be wire-line like ISDN, DSL, LAN/IP, fiber to the home, Docsys, but also the old fashion 2 wire telephone lines.

Another group of access networks is wireless based on radio technology; radio access networks characterized in that a user terminal is connected to the access network via a radio link. Radio access networks are also denoted cellular networks as the radio link has only a limited reach. Radio access networks therefore have mostly a plurality of access points to which a terminal can establish a radio link. The geographic area around the radio access point in which the radio link to the access point can be established with a certain probability is referred to as a cell of the radio access network. A moving terminal can change the radio link from access point to access point while moving through different cells. Examples of radio access networks are GSM, WCDMA, LTE, WIFI/WLAN but also satellite networks.

Central in the communication network is the core network. Access networks and the core network are interconnected via access nodes. The core network arranges connections between terminals and provides basic communication services. The connection can be message based (SMS, Mail, internet surfing), voice/audio based, picture/video based, Examples of the basic communication services are setting up connections between two or more terminals, alternative connection when a user terminal is occupied or not reachable like connection to voice mail recorder or callback functions. Other basic services are the charging and billing of incurred cost to users, handling of prepaid accounts, etc.

On top of the core, network service networks like IMS or IN provide more advanced telecommunication services like Virtual private networks, call barring and closed user groups.

Underlying to access networks, service networks and the core network is the transport network. The transport network can be based on a technology like Ethernet/IP, ATM or TDM. A transport network may contain more than one sub-network based on such a technology.

GSM, WCDMA and LTE are the most widespread radio access network technologies today. These networks usually coexist and provide together mobile network services in the same area. Network terminals are capable of connecting to and roaming between multiple network types and utilize without interruption. The different network technologies operate in different carrier frequency bands. The frequency distributions among the operators and among the different technologies are controlled by the local authorities. One technology can support more carrier frequencies in the same time, which multiply the network capacity. These networks are called multi-carrier networks.

The cell structure in a multi-technology and multi/carrier networks can be quite complex. The initial cell plan is determined by planning tools based on mostly geographical and radio propagation information. The main goal of the cell planning is to provide a full coverage and appropriate cell relations in the service area. The initial plan many times is prepared for each network type separately and the inter-network relations are added later in order to make handover between the different networks possible. When the network usage increases, additional carriers may be added to the networks. When the cells are on the air, it is possible to make radio coverage and radio quality

measurements, which can also be used to plan network expansions.

One of the main achievements of cellular networks is that they provide mobility for the user. The user can connect to a network anywhere and the user can move in the service area without interruption of the service. Mobility is achieved by handovers between the cells. In a certain area usually there are multiple overlaps of cells belonging to the same or different networks. The handover is possible between cells for which a neighbor relation is configured. The handover decision during a call may be made by signal strength measurements and handover policy. One can distinguish intra-frequency, inter- frequency and inter radio technology (IRAT) handovers (HO). Intra-frequency HO is handover between cells using the same frequency band. Inter-frequency HO refers to handover between different carriers of the same technology (e.g. WCDMA or LTE). IRAT HO is handover between two different networks, e.g. GSM to WCDMA or LTE.

The two measures that substantially determine the service quality and user perception are grade of service (GoS) and quality of service (QoS). GoS parameters are call or session level quantifiers, which are related to the access to the services, e.g. call establishment success rate, while QoS parameters are packet level quantifiers that characterize the quality of the ongoing sessions.

In order to maximize the revenue, operators would like to provide differentiated services to different users groups. Distinguished users, named premium or gold subscribers, are ready to pay extra money for receiving better services. The expectation of these customers is to receive excellent service without compromise anytime and anywhere. The operators would like to provide reasonable or optimum service to the average users or silver customers. The most important for subscribers in the so-called bronze or economic category is the favorable price. In return they tolerate lower service levels. In order to support the above mentioned services categories, the network should be able to provide differentiated service both in the radio and transport part of the network. GoS and QoS are end-to-end measures and therefore the radio and transport network should provide GoS and QoS together. In the network the following mechanisms provide GoS and QoS separation: admission control, congestion control, scheduling methods and delivery priorities. In an IP transport network DiffServ routers are able to deliver packets with different cause codes with different priorities. The different session types and sessions belonging to different subscriber categories can be mapped to different cause codes in the edge routers. LSP and MPLS are available to separate traffic via different routes in the IP transport network. In the radio network different sessions belonging to different session types are mapped to different radio bearer types and/or marked by different QoS indicators. The different radio bearers are admitted to the radio network with different priorities, policies and the radio frames are delivered by different priorities. State of the art network management is performed by operation and maintenance, O&M, subsystems or Operational support subsystems, OSS. A regular communication network has a multitude of these O&M and OSS. Below the term sub-network

management system (SNMS) is used to refer to both an O&M and an OSS.

Normally each sub-network in the communication network, like IP transport, ATM transport, GSM access, LTE access, DSL access, but also sub-networks like charging and billing in the core network has its own SNMS. A state of the art SNMS manages the performance of the sub-system and has no or very limited interaction with other SNMS. Tendency for operators is to change from system performance to user experienced performance. The user experience is related to many of the sub-networks/systems in the communication network. To express user experienced performance the operator defines so-called Key Performance Indicators (KPIs) and target values for these KPI. The target value of a KPI defines the "green" condition as a minimum, maximum value or as a bandwidth in which the KPI may fluctuate. Whereas the KPI themselves are related to user experience, the target values are related to agreements (and related cost) like in a user subscription. Hence KPI can be a conglomerate value for all users, but also differentiated to user groups (i.e. service levels), types of services like SMS, voice call, video call or IP-bandwidth for internet browsing, geographic areas like city or rural access or even single cells.

In the state of the art communication networks, information on the user experience related KPIs cannot be used automatically to improve the configuration of the subnetworks/systems in the communication network.

Summary of the invention

It is an object of the invention to improve the configuration management of

communication networks.

This object is achieved by providing a network management system for a communication network, the communication network having one or more sub-networks each being managed by a sub-network management system, wherein the network management system comprises - a KPI calculator configured for receiving performance data from the sub-network management system, and for determining KPI values on basis of the performance data;

- a KPI comparator module configured for receiving the KPI values from the

calculator, comparing the KPI values with KPI target values and outputting identifiers of off-target KPIs not meeting their KPI target values,

- a controller module configured for receiving the identifiers of the off-target KPIs, and selecting a selected rule set based on the off-target KPIs, and outputting a set indicator of the selected rule set,

- a rule set repository configured for storing a plurality of rule sets, each of the rule sets being associated with a rule set identifier, and for providing an output rule set out of the plurality of rule sets for which the rule set identifier matches the set indicator of the selected rule set, to each of the sub-network management systems.

By determining off-target KPIs and selecting the proper rules to be executed by the subnetworks, an automated optimization of the configuration of the network is possible, thereby also gradually optimizing the KPIs.

In an embodiment, the controller module comprises a state machine, a state of which is defined by a state table comprising in each row an indicator for the current state, the new state, the condition for changing the state, and an indicator of a rule set.

The use of a state machine allows to handle complex 'if-then-else-situations' taking the history of the system behaviour of the communication network into account. The condition field of the table may contain a KPI identifier, resulting in a simple embodiment with the controller module acting on one prioritized KPI at a time. Alternatively the condition field contains a Boolean expression of KPI identifiers. This option allows that different KPIs are treated with the same priority by the controller module.

In an embodiment, the KPI comparator comprises:

- a PID module for performing PID processing on the KPI values,

- a threshold comparator for comparing an output of the PID module with KPI target values.

The KPI comparator may further comprise a prioritizer configured for selecting one off- target KPI out of a number of off-target KPIs having the highest priority. In an embodiment, the KPI calculator is configured for calculating KPI values for each item in one or more categories. A category may be one of service types, service levels or geographic areas of the sub-networks. This makes service level and service type differentiation possible across different sub-networks which may be sub-networks using different technology. Furthermore load balancing is possible across cells in the same network and across cells of different sub-networks. A geographic area can be a location area, or a radio cell.

In an embodiment, the KPI comparator, the controller module and the rule set repository are configured to process one item of one category at a time provided by the KPI calculator.

In a further embodiment, the controller module and the rule set repository are configured to process one item of one category at a time provided by the KPI comparator.

In yet a further embodiment, the KPI comparator, the controller module and the rule set repository are configured to process each item in each category in parallel. This embodiment is also referred to as the whole matrix approach.

In an embodiment, rules comprised in the rule sets are supplemented with sub-network indicators indicating for which sub-network management system the rules are valid. This embodiment has the advantage that one rule set can be sent as one set to all sub- network management systems, which will then recognize and process the rules indicated to them.

In an embodiment, the rule sets each comprise several subsets of rules, the subsets each being associated with a sub-network indicator so as to be sent to the management system of the indicated sub-network.

In a further embodiment, the rule set repository is configured to store for each rule set a version number indicating which version of a rule set is stored by each specific subnetwork management system, and configured to send rules set identifiers to the specific sub-network management system to indicate which rule set shall be executed. If a rule is updated in the rule set repository, it only requires sending an update to the specific sub- network management system that might use a previous version of that the rule set.

In an embodiment a rule comprises a place holder that is substituted by the rule set repository before the rule is send to the sub-network. In this embodiment less different rules are needed in the repository of the network management system. Alternatively, a rule may comprise a place holder that is to be substituted by the receiving sub-network management system. In this case less different rules in the repository are needed and the network management system does not need to know all details in each sub network. In an embodiment, a rule in the output rule set provides one or more of the following:

- a criterion for I RAT HO,

- a dividing algorithm between GBR and BE traffic of available bandwidth,

- a selection rule for traffic diversion on MPLS,

- a intra cell HO criterion,

- a communication service grant limiting algorithm.

According to an aspect, a method of managing a communication network on basis of Key Performance Indicators (KPIs) is provided, wherein the communication network comprises one or more sub-networks each having a sub-network management system, and wherein the method comprises:

- receiving performance data from the one or more sub-network management systems,

- calculating values for one or more KPIs from the performance data,

- comparing calculated KPI values with target values, to render off-target KPIs,

- if one or more off-target KPIs occur:

o selecting a set of rules associated with the off-target KPIs,

o providing the set of rules to each sub-network management system.

Brief description of the drawings

Figure 1A schematically shows a communication network according to the state of the art;

Figure 1 B schematically shows a network management system and a communication network according to an embodiment;

Figure 2A schematically shows a KPI calculator according to an embodiment;

Figure 2B shows an example of a set of KPI values used when optimizing the

configuration of the sub-networks;

Figure 2C shows another example of KPIS arranged in a KPI matrix having values based on service level and service type;

Figure 3 schematically shows a more detailed view of the KPI Comparator 22 according to an embodiment; Figure 4 schematically shows an example of a proportional-integral-derivative controller used in an embodiment of the invention;

Figure 5A schematically shows a more detailed view on the controller module according to an embodiment;

Figure 5B is an example of a state diagram of a state machine used in the controller module according to an embodiment;

Figure 6 shows an example of a network comprising GSM, WCDMA , LTE and WIFI radio access networks connected to a core network;

Figure 7 shows another example of a state diagram to be used by a controller module according to an embodiment;

Figure 8 schematically shows three users travelling through three cells of a cellular network working in the same frequency band according to the state of the art,

Figure 9 shows another example according to the state of the art;

Figure 10 shows an example for the operation in a multicarrier network wherein three frequency bands are available;

Figure 1 1 shows another example for the operation in a multicarrier network, where rules, affecting both cell selection and handover selection are applied;

Figure 12 is a flow chart of a method of managing a communication network on basis of KPIs according to an embodiment of the invention.

Detailed description of the embodiments

Figure 1 A schematically shows a state of the art communication network 1 . The communication network comprises a number of sub-networks 2, for example a GSM access network, a WCDMA access network, a WIFI access network, a core network and an IP transport network. Each of the sub-networks 2 is controlled by an O&M or an OSS, referred to as O&M/OSS 3. To each of the O&M/OSS 3 an operator terminal 9 is connected allowing an operator to change configuration parameters used by the corresponding O&M/OSS 3. An O&M/OSS 3 further comprises a monitor function 4 that retrieves data from the sub-network 2 and provides this to a control function 6, see arrow 5. The control function 6 has fixed algorithms for controlling the sub-network and uses the provided data 5 as well as the parameter values to provide reconfiguration commands to a configurator 8, see arrow 7. The configurator 8 translates the

reconfiguration commands into reconfiguration data which is send by the configurator 8 to the correct entities in the sub-network, see arrow 1 1 . The monitoring function 4 is configured to collect statistical, counter and event-based data from its corresponding sub-network. Monitoring can be differentiated per radio cell/area, service level service type etcetera. Typically accessibility, retainability, integrity and traffic related

performance e.g. load and mobility parameters are monitored.

The accessibility parameters may comprise: attempt, success, fail rate of accessing the requested radio resources and radio channels, per service types.

The retainability parameters may comprise: drop rate due to pure radio environment or radio related changes, preferably measured per service types.

The integrity parameters may comprise: block or frame error rate for circuit switched (CS) services, throughput parameters for Packet Switched services.

Packet level QoS parameters may also be monitored, such as packet loss, delay, and jitter. Mobility parameters may comprise handover attempt, success, failure rates per service types.

The load parameters may comprise the actual number of users in idle and active states, as well as the number of active connections, which may be expressed in Erlang. For packet switched traffic the average user level throughput can be monitored.

In case of a cellular radio access network the monitoring function 4 may also collect radio environmental data, such as measured signal strength and interference (signal to noise ratio). Radio environment parameters may be measured by the user equipment (e.g. mobile phones) and base stations of the cellular network(s), examples of radio environment parameters are:

• signal level, (WCDMA: Received Code Power (RSCP), Received total wide band power, LTE: Reference Signal Received Power (RSRP)),

• interference measures (WCDMA: Ec/No, Signal to Interference Ratio (SIR), LTE: Received Interference Power, Reference Signal Received Quality (RSRQ)).

Other standard radio environment data specified by 3GPP TS 36.214, 3GPP TS 25.215, or any vendor specific radio environment measurement can also be used. The radio environment parameters are used primarily to measure the coverage and the radio quality.

As stated in the background operators tend to shift from system performance to user experienced performance. This leads to a greater spectrum of KPIs. There may be different types of KPIs. One type of the KPIs may be related to the actual traffic load per services and per user types. Another type of KPIs may be related to the QoS (e.g.

speech quality, packet delay, jitter), or GoS (e. g. calls setup success/failure rate, call drop rate, handover success/failure rate). Specific examples of KPIs which may be used are:

• KPI1 : Number of active connections per cell

• KPI2: Total throughput per cell

• KPI3: user throughput

• KPI4: Speech quality

· KPI5: jitter

• KPI6: average packet delay

• KPI7: calls setup success rate

• KPI8: Call drop rate

• KPI9: Handover success rate

· KPI10: signal level

• KPI1 1 : interference level

Figure 1 B schematically shows an embodiment of a network management system 13 (NMS) in communication with a communication network 1 . As compared to Figure 1 A, some changes are made to the O&M/OSS 3 of each subsystem 2. The control function 6, see Figure 1 A, is replaced by a rule engine 6A and a rule repository 6B. The rule repository 6B holds a current set of rules to be enforced by the O&M/OSS 3'. The multiple O&M/OSS 3' are further adapted in that they now provide performance data to a KPI calculator 20 comprised in the network management system 13 by their monitoring modules, see arrow 28. The KPI calculator 20 determines a value 21 for each KPI and provides that value to a KPI comparator 22. The KPI comparator 22 compares KPI target values with the received KPI values and determines if a certain KPI exceeds its target value and which KPI has priority if more than one KPI exceeds. The identity of the exceeding KPI, or a value 0 if no KPI exceeds, is send to a controller module 24, see arrow 23. The controller module 24 determines the identity of a set of rules required to optimize the configurations of the sub-network 2, and provides this, see arrow 25, to the rule set repository 26. The rule set repository provides the new set of rules to the

O&M/OSS of the sub-networks 2, see arrow 27. Each O&M/OSS having received a new set of rules stores them in the rule repository 6B after which the rule engine 6A starts enforcing the new rules. Please note that as the rules are no longer hardcoded, rules can be created or modified runtime. Figure 2A provides a more detailed view on the KPI calculator 20 according to an embodiment. The KPI calculator 24 comprises a receiver 201 , a transmitter 203 and a calculating unit 202. The receiver 201 is provided with performance data 28 from the monitor functions 4' in each of the O&M/OSS 3'. The O&M/OSS 3' may send this data based on an internal timing, the receiver 201 of the KPI calculator 20 may request the data when needed, or the receiver 201 subscribes to data and may in that provide a time interval. If data is complete the calculating unit 202 then calculates the values for the KPIs. Completeness of data may be done by a trigger from the receiver 201 or based on heartbeat timing. The output of the KPI calculator 20 is a set of KPI values. This set can be a single value for each KPI as indicated by 21 1 in Figure 2B or differentiated for each KPI as indicated by the example 212, where for each KPI a matrix of values is provided here based on service level and service type, see Figure 2C. Instead of a 2-dimensional matrix, higher dimensions may be used if higher order differentiation is used. For the remainder of the detailing of the modules of the network management system first the list of single KPI values is used as basis. The more complex situation of matrices requires some additional measures and as discussed separately under differentiated KPI.

Figure 3 provides a more detailed view on the KPI Comparator 22 according to an embodiment. The KPI comparator 22 comprises a receiver/input 221 , a comparator unit 222 and the transmitter/output 230. Input 221 receives the KPI values, see arrow 21 . The output 230 provides the identity of a target exceeding KPI or a value 0 if none exceeds. The comparator unit 222 comprises a repository 225 for storing target values for each KPI. The target value may have an upper and lower limit which can be used to set a minimum, a maximum or a bandwidth to which the KPI shall adhere. The upper and lower limit value for each KPI is provided to a threshold circuit 227. The threshold circuit 227 compares in parallel the value of each KPI 224 with the limits provided by

225, see arrow 26. The output 228 is per KPI 0 or 1 representing no or yes exceeding. A prioritizer 229 then checks if a KPI exceeds the target and if so if there are more than one exceeding. The output of the prioritizer 229 is provided to the output 230 as a KPI indicator. If no KPI exceeds its target value, the KPI indicator has value 0. If one KPI exceeds the number of that KPI, and if more than one KPI exceeds, the number of the KPI having the highest priority is output.

It is noted that in the above the words 'exceeding a target value' are used, but that this should be read as 'not meeting its target value', since a KPI value may as well be too low to meets its target. A KPI which does not meet its target is also referred to as an off- target KPI.

In order to control the adaptive speed of the communication network or prevent overshoot and oscillations, the KPI value over time may be subject to proportional, integrating and differentiating adjustment of the KPI value. This function is provided by the PID regulator 223. Figure 4 schematically shows an example of a PID regulator 223. The PID regulator has 3 parallel branches 32,33 and 34 that are fed by an input signal 31 which is a function of time, so x(t) = KPI (t). This input signal is processed by a proportional multiplier 32, an integrator 33 and a differentiator 34. The outcome of the three calculation modules 32, 33, 34 is summed up to render on output 36 being a signal y(t) which a function of time. The values for y(t) are calculated on subsequent points in time. Heuristically, these values can be interpreted in terms of time: P depends on the present error, I on the accumulation of past errors, and D is a prediction of future errors, based on current rate of change. The weighted sum of these three actions forms the output of the PID controller. The PID controller 223 is able to eliminate undesirable effects due to KPI fluctuations and will achieve more robust closed loop operation. It might occur that a KPI exceeds a threshold only temporarily due to fluctuation. It would lead unstable operation if these fluctuation trigger rule changes. The integrating aspect prevents that peaks in the KPI value reach the output. On the other hand the rate of change in a KPI value can be so high that alternative rules should be applied before the KPI value reaches the threshold as it is then probably already too late. The differentiating branch will increase the KPI value above the value it actually has. It is noted that instead of a PID regulator also more advanced statistical and heuristic regulator principles could be used. E.g. the average and standard deviation of the KPIs are continuously calculated and the regulator determines if the KPI change is within its natural fluctuation and adjusts the KPI value accordingly before it is provided to the threshold circuit. It should be noted that typically several KPIs are measured and compared in parallel. In that case a plurality of PID regulators 223 may be present in the KPI comparator 22, one for each KPI. Figure 5A provides a more detailed view on the controller module 24 according to an embodiment. The KPI indicator is provided, see arrow 23, to the receiver/input 241 of the controller module 24. The controller module 24 provides a rule set indicator 25 via the transmitter/output 243. The controller module 24 further comprises a selector 242. In a relatively simple embodiment the selector 242 is a table with in each row a KPI identity and a rule set identity.

Alternatively, more advanced selection mechanism may be used for identifying rule sets. For example, if different rule sets must be selected in case of a KPI4 exceeds on its own or a previously KPI2 exceeded but KPI4 has a higher priority than KPI2. A state machine solution is a very adequate way of implementing this but other solutions may be used as well. An example of a state machine description is provided in the table 1 below, and in Figure 5B. Each row in table 1 represents a possible state change. The first column provides the current state, the second column the state to change to, and the third column the condition under which the state change is allowed. The last column provides the rule set that should be used in the state changing to. If no KPI exceeds its target then a condition called KPI0 is armed and the state will change back to state called S00 with an accompanying Rule Set RS 00 representing normal operation. When KPI2 is still exceeded when KPI4 is no longer, then the state changes from RS 23 to RS 13 instead of RS 00. KPIx is any KPI exceeding, so all but KPI0.

Current state next state Condition rule set ID

S12 S23 KPI 4 RS 23

S12 SOO KPI 0 RS 00

S12 S13 KPI 2 RS 13

SOO S12 KPI x RS 12

S13 S25 KPI 4 RS 25

S13 SOO KPI 0 RS 00

S25 SOO KPI 0 RS 00

S23 S13 KPI 2 RS 13 S23 soo KPI 0 RS 00 table 1

The rule repository 26 receives a rule set identity 25, see also Figure 1 B. The repository 26 can be implemented in several ways which may have an impact on the distribution method of the rule set repository.

In a first implementation each rule set is identified by a rules set identifier and contains all the rules for all the O&M/OSS 3'. The rule set repository 26 sends the whole set in total to the O&M/OSS 3' based on the received rule set identity matching the rule set identifier received from the controller module 24. Each rule in the rule set may have an indicator applicable for a certain O&M/OSS 3'. The O&M/OSS 3' will then only activate rules that are applicable to it.

In a second implementation the rule set is divided into sub sets. Each sub set may have an additional identifier pointing to a specific O&M/OSS. The rule set repository sends each sub set to the specific O&M/OSS pointed to, based on the received rule set identity matching the rule set identifier.

A third implementation is an addition to the two previous implementations. One or more rules of the rule set have placeholders that need to be filed before the rules are actual send. This method is advantageous when differentiated KPIs are used like for each service level.

A fourth possible implementation arranges also the transmission function in co-operation with O&M/OSS 3' that handle differentiated rule sets, like a rule set per radio cell/area. The rules in the rule set comprise an indication of the differentiation. Alternatively the differentiation is provided to the rule set repository. The transmission is arranged the same way as in the first or second implementation.

The flexible rule framework described above makes it possible to add rules based on operation experience new sales/technological strategies. Easy and gradual

implementation of new rules is possible. New rules can be added to rule sets or rules can be adapted via a user interface to the rule repository 26. It allows the operator of the NMS 13 or a service provider to change network behavior in real time. Furthermore, new or changed rules can be time stamped and the rule repository 26 activates the rules at the time stamped. The same can be used for time stamped removal of rules. Differentiation of KPI

As stated before instead of a single list of KPI, matrices of KPI may be used allowing KPI to be differentiated like the example given in Figure 2C. The working of the KPI comparator 22 and selector 242 are substantially identical but now on a per matrix cell basis. So for example, the KPI values for a service level being 'economy' and for a service type being 'SMS'. The basic difference to the single KPI list of Figure 2B is that the operation needs to be done for each cell in the matrices of Figure 2C. This may be implemented by parallel processing for the KPI comparator 22 and the selector 242. Another example is differentiation in service level and area/cell. In that case also parallel processing may be used but the problem arises how to select the large amount of different rule sets. In this case matrix cell by cell processing may be used. A first implementation is based on a heartbeat provided by the KPI calculator 22 that provides the KPI 1 to n belonging to a first cell of the matrix and then the next cell and so on to the KPI comparator 22 and selector 242. This option allows using a single target value per KPI valid for all cells of the matrix. When however also KPI target differentiation is wanted, then the second implementation is better. The KPI comparator acts on whole matrix basis and provides the heart beat and cell by cell KPI identity to the selector. A third option is also possible in that also the selector 242 works parallel on the whole matrix (basically one state machine per matrix cell). In the first and second option the previous state shall be remembered per matrix cell by the selector 242. As indicated by the dotted line 244 in Figure 5A, the indices of the current matrix cell that is processed are indicated to the rule set repository 26. The output of dotted line 244 has a further advantage in that the number of rule sets can be limited. The rule set repository 26 also receives the cell indices and can adjust the rule set if required by using place holder principles in the rules that are filed on basis of the received indices and the rule set repository 26 can send the rule set including a representative for the indices (like for a specific radio cell/area) to the O&M/OSS 3' which stores the rules applicable for that radio cell/area. This method may be used also for other options where the O&M/OSS 3' are capable of storing differentiated rule sets.

The embodiments described above may be applied to manage multi-technology and multicarrier cellular networks, where the cells using different radio technologies or carriers overlap. Several cellular networks can be controlled by the same network management system 13. Knowledge of neighboring cells or overlap may be implemented in the rule sets in the repository 26 but may also be left to the O&M/OSS 3'. A rule may specify only the fact of a neighboring cell and the O&M/OSS 3' may fill in which radio cells it concerns. Likewise the knowledge on overlap of carrier or radio technologies can be rule based or left to the O&M/OSS systems 3'.

In the following a number of possible embodiments are described to further explain the invention. These examples will probably not provide an optimal solution for the communication network 1 but are discussed to provide a good understanding of the working of the invention.

Example 1

In the first example, an operator owns a network comprising GSM, WCDMA, LTE and WIFI radio access networks connected to a core network. Transportation is performed by an IP transport network. The situation is provided in Figure 6.

GSM and WCDMA are collocated and not all radio cells have WIFI coverage. The operator provides for 2 types of subscriptions:

- Premium with a maximum access rate of 7Mb/s connection LTE, GSM, WCDMA.

- Economy with a maximum up to 7Mb/s, connection GSM/WCDMA and optional LTE.

When enough bandwidth is available higher data rates are provided. Both economy and premium user may request premium content for which is paid extra like movie on demand or live soccer games. Terminals used by users are not all multimode, so some might be WIFI enabled and others not. The operator requires that each category gets a fair share but also that premium content has highest priority as it provides additional income. A user experience KPI is defined as bearer establishment success rate being the percentage of bearer requests that lead to established bearers. The KPI is calculated per category (economy, premium or premium content) and per radio cell as the situation may fluctuate largely amongst radio cells. Bearer requests above the 7Mb/s limit are not included and premium content rates are not considered for the 7Mb/s.

The target KPI values are determined in that the success rate for premium content must be higher than 95%. For premium users the success rate must be higher than 60% and for economy users the success rate should not drop below 10%. As the O&M/OSS 3' for the sub-networks are capable of storing rules to be applied per radio cell, the matrix operation performed throughout all the modules 20,22,24,26 of the NMS 13 is chosen as best option in this example, as described in section 'KPI differentiation of ΚΡΓ above.

The network management system 13 will be provided with data from the O&M/OSS 3' from GSM, WCDMA, LTE and the IP transport network via interface 28 of Figure 1 B. From the GSM, WCDMA and LTE sub-networks data is provided per radio cell, comprising one or more of:

- Number of active connected terminals premium users,

- Number of active connected terminals economy users,

- Total number of bearer requests/minute economy,

- Total number of bearer requests/minute economy, premium content,

- Total number of bearer requests/minute premium,

- Total number of bearer requests/minute premium, premium content,

- Successful for economy users total/minute,

- Successful for economy users premium content/minute,

- Successful for premium users total/minute,

- Successful for premium users premium content/minute,

- Actual data rate total and occupancy (fraction of the max capacity used),

- Actual data rate sum economy users,

- Actual data rate sum premium users,

- Actual data rate sum premium content. From the IP transport O&M/OSS data is provided on the load of the Label Switched Path (LSP) to WIFI, GSM/WCDMA and LTE towards the core network. Provided data may comprise the actual rate and the occupancy of maximum rate.

From this data the KPI calculator 20 determines a KPI value for each of the categories and for each radio cell;

- premium content success rate for a radio cell is the sum of (economy and

premium users successful premium content for GSM + WCDMA + LTE)/ the sum of (economy and premium users request for premium content for GSM + WCDMA + LTE) premium user success rate for a radio cell is the sum of (premium user successful total - premium content for GSM + WCDMA + LTE)/ the sum of (premium user request total - premium content for GSM + WCDMA + LTE) The 7Mb/s limitation is provided for in the O&M/OSS monitoring that has pre-adjusted the values based on not granted and granted requests above the 7MB/s limit,

economy user success rate as for premium user for a radio cell is the sum of (economy user successful total - premium content for GSM + WCDMA + LTE)/ the sum of (economy user request total - premium content for GSM + WCDMA + LTE) The 7Mb/s limitation is provided for in the O&M/OSS monitoring that has pre-adjusted the values, based on not granted and granted requests above the 7MB/s limit.

The so calculated matrix is provided to the KPI comparator 22. Internally the KPI comparator 22 has the limits 100-95, 100-60 and 100-10 stored as target values in the KPI target storage 225. The PID regulator 223 is adjusted for filtering spikes and early action on strong increases. The PID factor is 60/20/20% initially but can be adjusted per cell. In the whole matrix approach all the values used in the KPI comparator are defined per cell of the matrix and can be adjusted accordingly. In this example however they are stated as initially the same for all matrix cells. The integrating time constant is one parameter that is set per radio cell and equals the time to pass the geographic area covered by the radio cell with 25 meter/second. The differentiating time constant is kept low as the ranges for premium content is quite narrow and it prevents unstable system behavior. With threshold circuit 227 and prioritizer 229 a result matrix is provided in which each matrix cell corresponds to a radio cell and has the following KPI identities: KPI 0 = no KPI exceeds in the radio cell

KPI 1 = lowest priority, economy user success rate is below 10% in the radio cell

KPI 2 = medium priority, premium user success rate is lower than 60% in the radio cell

KPI 3 = high priority, premium content success rate is lower the 95% in the radio cell.

The filled matrix is provided to the controller module 24. The selector 242 in the controller module 24 is state-machine organized and a state machine is present for each radio cell of the provided matrix. Hence the state can be different in each of the radio cells. For the state machine one normal operation and four exemption operation states are defined in this example. Each of them has a specific set of rules that shall be used by the respective O&M/OSS 3' for the specific radio cell. The state diagram is shown in Figure 7, the representing table 2 below. It is valid for each cell of the matrix.

table 2

Now the output 25 of the controller module 24 is again a matrix of cells, each cell now comprises an identifier of a rule set that must be applied for that radio cell. This output is provided to the rule set repository in 25. The rule set repository comprises 5 rules sets identified by RS 00 - 04. A subset approach is chosen here, meaning that a rule set is divided into subsets each subset applicable for a specific O&M/OSS 3'. Not all rule sets are listed in table 3 but only those for RS 00 and RS 03.

Rule Subset Rule Rule

set index index

RS 00 GSM R001 Best effort Bearer request from economy user allowed if best effort traffic occupies less than 80% of total cell traffic rate - actual premium content traffic rate - actual premium user traffic rate

RS 00 WCDMA R001 Best effort Bearer request from economy user allowed if best effort traffic occupies less than 80% of total cell traffic rate - actual premium content traffic rate - actual premium user traffic rate

RS 00 LTE R001 Best effort Bearer request from economy user allowed if best effort traffic occupies less than 70% of total cell traffic rate - actual premium content traffic rate - actual premium user traffic rate

RS 03 GSM R001 Best effort Bearer request from economy user allowed if best effort traffic occupies less than 100% of total cell traffic rate - actual premium content traffic rate - actual premium user traffic rate

RS 03 GSM R002 IRAT handover to WCDMA of premium user best effort bearers if terminal capable

RS 03 GSM R003 IRAT hand over of premium content bearers to LTE when terminal is capable

RS 03 WCDMA R001 Best effort Bearer request from economy user allowed if best effort traffic occupies less than 20% of total cell traffic rate - actual premium content traffic rate - actual premium user traffic rate

RS 03 WCDMA R002 IRAT handover of best effort bearers of economy users to GSM or

WIFI when terminal capable.

RS 03 WCDMA R003 IRAT handover of premium content bearers to LTE when terminal is capable

RS 03 WCDMA R004 Neighboring cell handover criterion = value - 10 for economy user when neighboring cell actual traffic rate is < 60% of maximum traffic rate.

RS 03 WCDMA R005 Best effort bearer request of premium user only granted current

granted capacity is les then 7Mb/s limit

RS 03 LTE R001 Best effort Bearer request from economy user allowed if best effort traffic occupies less than 20% of total cell traffic rate - actual premium content traffic rate - actual premium user traffic rate

RS 03 LTE R002 Best effort Bearer request from premium user allowed if best effort traffic occupies less than 60% of total cell traffic rate - actual premium content traffic rate - actual premium user traffic rate RS 03 LTE R003 Neighboring cell handover criterion = value - 10 for economy user

when neighboring cell actual traffic rate is < 60% of maximum traffic rate.

RS 03 LTE R004 IRAT handover of best effort bearers of economy users to GSM or

WIFI when terminal capable. table 3

Under normal operation new best effort bearers may only be set up if some free capacity is left. This limitation is only for economy users. In RS 03 the success rate is too low of premium users without prior a too low success rate of economy users. The remedy applied is to shift economy users faster to other radio cells even when quality of connection is lower. Further economy best effort traffic is moved to preferably WIFI and GSM access to free WCDMA and LTE. In this example the premium users are now bound by their 7 Mb/s limit and best effort bearers will not be set-up when they exceed this. Premium content is moved to preferably LTE as this provides a higher efficiency for buffer filling in streaming situations.

The rule set repository 26 starts to send the rules to the GSM, WCDMA and LTE

O&M/OSS, which then store the rules locally in their repository 6B and apply them by their rule engine 6A. To optimize rule transport, some options may be available to avoid sending the same rules over and over. A first option is that prior to the rules the output of the controller module 24 transfers the matrix providing an overview of rules sets applicable per cell. Then the needed rule sets are transferred (only those required by the matrix and applicable for the specific O&M/OSS) to those O&M/OSS.

A further improvement is possible in that the rule set is fully stored in the repository 6B. Each of the O&M/OSS 3' uses the matrix to select the correct rules to use for a radio cell by the rule engine 6A. This means however that now the output must be capable of version control on what rules are stored. If a rule is updated in the rule set repository 26, it requires sending an update to the specific O&M/OSS 3' that might use the rule when the rule is activated, or when a rule is deactivated. In case of a restart of an O&M/OSS 3' it might request a full rule download in these cases.

Example 2: using barring matrix for network with multiple radio technology coverage As provided in example 1 , users differentiated to service level may be handed over to a network using a different radio technology. This can be done in a radio cell but also selectively during handover. In that case for a cell a radio technology or a carrier frequency band is barred for a certain service level. This can be used to provide premium service for premium user. An example of such a situation is when general QoS levels for one radio technology are significant lower than another radio technology where these have cells that overlap. Another example is when the load is so high for a certain radio technology that one wishes to bar economy users from using that radio technology and hand them over to another radio technology of which the radio cell overlaps. For premium user one whishes not only the best QoS but also spare capacity so that loss of connection of moving terminals has a lower probability, and accepts for economy users a higher probability of loss with minimum spare capacity.

The actual barring is a rule that is executed by the O&M/OSS 3' by setting parameters for the handover into the cell. The rule itself is provided by the network management system 13. The barring can be unconditional (which will be applied in extreme overload conditions) and be conditional (for more moderate overload situations). Using the service level categories from example 1 a possible extreme condition is when LTE load by premium content is so high that all other data traffic of premium users and economy user is barred from LTE and is handed over to GSM or WCDMA (or WIFI) when the terminal is capable or when not connection is lost. An example of conditional barring is taking the capability of the terminal into account.

To enable the NMS 13 to control the barring it may use the following data per radio cell. From GSM, WCDMA and LTE per radio cell:

- Average QoS level premium content,

- Average QoS level premium users,

- Average QoS level economy users,

- Actual data rate total and occupancy (fraction of the max capacity used),

- Actual data rate sum economy users,

- Actual data rate sum premium users,

- Actual data rate sum premium content. The O&M/OSS 3' of each radio technology provides the data. For QoS level an estimated Bit Error Rate, EBER, may be calculated taking into account factors as interference (signal to noise level), latency and jitter. From this data the KPI calculator 20 determines KPI values for each radio cell. The threshold for premium content and the threshold for premium users are different but they use the same value of EBER for the radio technology. To solve this issue the KPI may be doubled and is indicated below with A and B to accommodate application of more than one threshold on the same variable:

KPI 1A QoS level LTE premium content,

KPI 1 B QoS level LTE premium users,

KPI 2A QoS level WCDMA premium content,

KPI 2B QoS level WCDMA premium users,

KPI 3A QoS level GSM premium content,

KPI 3B QoS level GSM premium users,

KPI 4 occupancy GSM.

The occupancy for LTE and WCDMA is not required as the capacity increases with a deeper code tree but at the same time, the Signal to noise ratio becomes worse. As a result of the 7 defined KPIs mentioned above, the output of the KPI calculator 20 will be a matrix of 7 KPI values per radio cell. The PID regulator may be adjusted for spike filtering in QoS to a larger extend but no differential regulating is used. The PID factor can for example 60/40/0%. The time constant for the integrating is rather short but adapted per category (2 minutes for premium content and 10 minutes for premium users). The threshold circuit uses EBER maximum threshold for premium content and premium user, and a max occupancy factor for GSM of 90%.

The prioritizer 229 is however slightly different to the one described in example 1 because the state machine would become very complex if a single KPI condition would be used. Instead now a more complex condition is used comprising more than one variable. To achieve this no priority is applied and notices of an exceeding of each of the KPI will go to the controller module24. The output of the KPI comparator 22 is therefore a matrix comparable to the input of the KPI comparator 22; 7 KPI indicators per radio cell. Each KPI is either 0 (not exceeded) or 1 (exceeded) which allows a Boolean condition in the state machine. The condition column of the state machine does now not hold a single KPI as selection but it might be a more complex condition. Example: KPI 1A and not KPI 2A and not KPI 3A.

Current Next state Condition Rule set state

* SO * RS 00

SO S1 KPI 1A . not KPI 2A . not KPI 3A RS 01

SO S2 Not KPI 1A . KPI 2A . not KPI 3A RS 02 so S3 KPI 1A . KPI 2A . not KPI3A RS 03 so S4 Not KPI 1A . not KPI 2A . KPI 3A RS 04

S1 S3 Not KPI 1A . not KPI 2A . KPI 3A RS 03

S1 S4 KPI 1A . KPI 2A RS 04

S1 S5 KPI 1A . KPI 1 B . not KPI 2A RS 05

S1 so Not KPI 1A . not KPI 2A . not KPI3A RS 00

S2 S3 Not KPI 1A . not KPI 2A . KPI 3A RS 03

S2 S4 KPI 1A . KPI 2A RS 04

S2 S6 Not KPI 1A . KPI 2A . KPI 2B RS 06

S2 so Not KPI 1A . not KPI 2A . not KPI3A RS 00

S3 SO Not KPI 1A . not KPI 2A . not KPI3A RS 00

S3 S8 Not KPI 1A . not KPI 2A . KPI3A . KPI 3B RS 08

S3 S1 KPI 1A . not KPI 2A RS 01

S3 S2 Not KPI 1A . KPI 2A RS 02

S3 S4 KPI 1A . KPI 2A RS 04

S4 SO Not KPI 1A . not KPI 2A . not KPI3A RS 00

S4 S1 KPI 1A . not KPI 2A RS 01

S4 S2 Not KPI 1A . KPI 2A RS 02

S4 S3 Not KPI 1A . not KPI 2A . KPI3A RS 03

S4 S9 KPI 4 RS 09

S5 S1 KPI 1A . not KPI 1 B . not KPI 2A RS 01

S5 S4 KPI 1A . KPI 2A RS 04

S5 S7 KPI 1 B . KPI 2B RS 07

S6 S2 Not KPI 1A . KPI 2A . not KPI 2B RS 02

S6 S7 KPI 1 B . KPI 2B RS 07

S6 S4 KPI 1A . KPI 2A RS 04

S7 S4 KPI 1A . KPI 2A . not KPI 1 B . not KPI 2B RS 04

S7 S5 KPI 1A . KPI 1 B . not KPI 2A RS 05

S7 S6 Not KPI 1A . KPI 2A . KPI 2B RS 06 S8 S7 KPI 1A + KPI 2A RS 07

S8 S3 Not KPI 1A . not KPI 2A . KPI 3A . not KPI 3B RS 03

S9 S4 Not KPI 4 . KPI 1A . KPI 2A RS 04

S9 S8 Not KPI 4 . not KPI 1A . not KPI 2A . KPI 3A . KPI RS 08

3B

S9 S3 Not KPI 4 . not KPI 1A . not KPI 2A . KPI 3A . not RS 03

KPI 3B

S9 SO Not KPI 1A . not KPI 2A . not KPI3A RS 00

^* start up meaning any current state unconditionally

table 4

The threshold for the premium contents is reached always before the threshold of premium users due to the difference in threshold value. Offload of premium content to GSM when QoS in LTE and WCDMA are both below the Premium content threshold is no option as the GSM system is not capable to maintain the data rates for several premium content streams. Therefore no barring is any longer used and users have to live with less quality of service as long as they are in this cell. The O&M/OSS might delay handover till the moment the current cell QoS has substantial lower value then the target cell by means of soft handover parameter setting. This cell overload protection mechanism is standard behavior of the radio access network and used here. If handover can take place to more than one cell then the standard operation (soft handover) of the radio access network will take care that the handover takes place to the best radio cell not having barred traffic.

The rule sets define the barring condition per service level and radio technology. This is visualized in the small matrices below. 0 means no barring, 1 is conditional barring and 2 unconditional barring.

RS00 RS01

LTE WCDMA GSM LTE WCDMA GSM

Premium 0 0 0 Premium 2 0 0 content content

Premium 0 0 0 Premium 0 0 0 user user Economy 0 0 0 Economy 1 0 0 user user

RS02 RS03

LTE WCDMA GSM LTE WCDMA GSM

Premium 0 2 0 Premium 0 0 2 content content

Premium 0 0 0 Premium 0 0 0 user user

Economy 0 1 0 Economy 0 0 0 user user

RS04 RS05

LTE WCDMA GSM LTE WCDMA GSM

Premium 0 0 0 Premium 1 0 0 content content

Premium 1 1 0 Premium 1 0 0 user user

Economy 2 2 0 Economy 2 1 0 user user

RS06 RS07

LTE WCDMA GSM LTE WCDMA GSM

Premium 0 1 0 Premium 0 0 0 content content

Premium 0 1 0 Premium 1 1 0 user user

Economy 1 2 0 Economy 2 2 2 user user

RS08 RS09

LTE WCDMA GSM LTE WCDMA GSM

Premium 0 0 2 Premium 0 0 0 content content

Premium 0 0 2 Premium 1 1 2 user user Economy 0 0 0 Economy 2 2 2 user user table 5

In addition to the barring, priority is set to maximum for premium content in RS04, RS07 and RS09. The output of the rule set repository 26 are rules to set barring and where applicable priority to each radio access network O&M/OSS 3'.

A further embodiment relates to the control of handover and cell selection in a not homogeneous network, meaning that certain areas might not be covered by each radio technology an operator has. The O&M/OSS 3' may be configured to execute two different types of rule sets. A first type of rule set will influence the cell selection and reselection decisions while the other type comprises rules that modify the handover decisions in multiple networks. Preferably, the cell selection/reselection decisions in the cellular networks are determined by periodic radio environment measurements in idle mode, as well as by cell selection/reselection policies which are implemented and set in the cellular network(s). Please note that cell reselection is possible only between those cells that are configured as neighbors. This embodiment implies that different neighbor matrixes are maintained for cell selection and handover, as well as for different services types and user groups.

First a state of the art network situation is described with reference to Figure 8 wherein cell selection and handover procedures are performed. Figure 8 schematically shows three radio cells 41 , 42, 43 of a cellular network working in the same frequency band according to the state of the art. In this example, an economy user 51 , an average user 52 and a premium user 53 travel the same route which is serviced by the three cells 41 , 42, 43. E.g. they get on a train in the same railway station and start using mobile internet. When connecting to the network, see arrow 44, the three users 51 , 52, 53 will connect to the same cell 41 but with different service level (i.e. economy, average, premium). The parallel horizontal arrows in Figure 8 indicate the travel of three users. Arrows 45 and 46 indicate a hand over to a next cell. During their travel, the three users share the same cells and the same radio resources. During the cell selection and handover events, there is no differentiation possible with regards to QoS level or GoS level.

Figure 9 shows another example according to the state of the art. In this example, the economy user 51 , the average user 52 and the premium user 53 travel in the same way as in Figure 8. The route is serviced by the three cells 41 , 42, 43. As shown in Figure 9 two further cells 54, 55 are available operating in a higher frequency band. Also two more cells 56, 57 are present working in a lower frequency band. Since the cell selection priority is the same for all users, they are connected to the same cell 41 with the same probability. While they are traveling by train they handover to the same cells 42, 43.

Since there is no service level differentiation, the other frequency bands (cells 54, 55, 56, 57) will not be exploited.

Figure 10 shows an example for the operation in a multicarrier network. In the

multicarrier network three frequency bands are available. The cell selection parameters are managed by the network management system 13 according to an embodiment as described above with reference to Figure 1 -5. The O&M/OSS 3' of each of the carriers may have the following data for controlling the single access system for that radio technology access network.

• number of idle and active connections

• actual threshold values for radio environment data for cell selection and

handover

• actual neighbor relations for cell selection and handover

• neighbor relation barring matrix

• actual priorities for cell selection and handover

The rule engine 6A may be configured to execute two different types of rule sets. A first type of rule set will influence the cell selection and reselection decisions while the other type comprises rules that modify the handover decisions in multiple networks.

Preferably, the cell selection/reselection decisions in the cellular networks are

determined by periodic radio environment measurements in idle mode, as well as by cell selection/reselection policies which are implemented and set in the cellular network(s) 2. Please note that cell reselection is possible only between those cells that are configured as neighbors. This embodiment implies that different neighbor matrixes are maintained for cell selection and handover, as well as for different services types and user groups. In this embodiment the network management system 13 creates out of the data obtained from each radio access network O&M/OSS 3' one or more of the following KPIs:

• KPI1 : Number of active connection per cell

• KPI2: Total throughput per cell

• KPI3: user throughput

• KPI4: Speech quality

· KPI5: jitter

• KPI6: average packet delay

• KPI7: calls setup success rate

• KPI8: Call drop rate

• KPI9: Handover success rate

· KPI10: signal level

• KPI1 1 : interference level

Now the rule set repository 26 of the NMS 13 could comprise at least the following rules: Rule 1 : The cell selection priority for premium users is set 1 for the best radio environment neighbor. The cell selection priority for average users is set 1 for the second best radio environment neighbor. The cell selection priority for economy users is set 1 for the third best radio environment neighbor.

Rule 1 is triggered e.g. when QoS degradation (speech quality, throughput) is experienced for premium users, namely KPI3 or KPI4 is below target.

Effect: Different cell priorities are applied to different user groups: Premium users are directed to cells with better radio environment (higher signal levels, lower

interference) than average and economy users. Rule 2: The handover thresholds are decreased by 1 DB for average user and 2 DB economy users.

Rule 2 is triggered for overloaded cells, when traffic load in the cell is too high, KPI1 or KPI2 exceeds target values. Effect: The cells are preempted by forcing lower priority users to handover.

Rule 3: If call establishment success KPI (KPI7) of a cell is below targets for premium users, the cell selection priority is decreased to minimum.

Effect: Higher priority users are directed to cells where call establishment success is high.

Rule 4: The cell selection priorities are decreased by 1 for each user groups if the cell coverage is limited, i.e. KPI10 is low.

Rule 5: If KPI9 does not meet the target, the corresponding cell relations with are added to the barring matrix

Effect: the related transitions will be denied). Rule 6: The cell load for premium users should always be kept optimum. In a cell that serves premium users, there should be enough margin to new and handover premium connections. If KPI1 or KPI2 is high, the cell selection priority for premium users is decreased. Rule 7: The cell load for premium users should always be kept optimum. In a cell that serves premium users, there should be enough margin to new and handover premium connections. If KPI1 or KPI2 is high, the handover selection priority for premium users is decreased. Rule 8: The handover priorities are decreased by 1 for each user groups if the cell coverage is limited, i.e. KPI10 is low.

Rule 9 If KPI4 does not meet in WCDMA, the handover threshold to GSM network for voice traffic is decreased.

Rule 10 If KPI3 does not meet in WCDMA, the handover threshold to LTE network for PS traffic is decreased. Network selection preferences can be set by service types. As an example, voice traffic is directed to GSM, smart phone traffic (using multiRABs) may be directed to WCDMA, pure PS sessions are directed to LTE, see rule 9 and 10. As a result of the rules 1 , 3, 4 and 6 cell selection rules the premium user 53 is served with cells with good coverage. The average user 52 and the economy user 51 are attached to different cells with a higher probability of unfavorable KPIs. The economy user 51 and the average user 52 are attached to the highest and central carrier frequency. The three users are connected to different cells: user 51 is connected to cell 71 , user 52 is connected to cell 66 and user 53 is connected to cell 61 . This means that the premium user 53 is connected to the cell 61 which will provide the best service as a result of applying rule 1 , 3, 4 and 6 affecting the cell selection priorities, the average user will be connected to the cell 66 which will provide the second best service level, and the economy user will be connected to the cell 51 providing the least best service. This differentiation in service levels is also referred to as 'service differentiation'.

As can be seen for the Figure 10, these two users 51 , 52 are enforced to more frequent inter-frequency handover than the premium user 53. Cells serving the premium user 53 may have lower handover priorities for other (i.e. non-premium) users in order to moderate load. The economy user 51 during his travel will need to perform a handover, see arrow 75, from the highest carrier frequency to the lower frequency. The economy user 51 will perform a second handover once he is about to leave the cell 67. This is also true for the average user 52. The economy user 51 and the average user 52 will join the premium user 53 in cell 62 because this cell is the only cell covering the area indicated by 76.

Let's assume that the rules affecting the handover priorities (Rules 2, 5, 7, 8) are not applied and the handover priority is the same for each user group. Since the handover priority and neighbor definition is the same, all three users will hand over into the same cells, i.e. cells 63, 64, 65. Once arrived in cell 63 there is no service differentiation and the other carrier frequencies (i.e. cell 68, 69, 70, 72, 73) will not be exploited. Another example for the operation in a multicarrier network is shown in Figure 11 , where all rules, affecting both cell selection and handover selection are applied. Figure 1 1 shows the same cells as were shown in Figure 10. In this example the cell selection parameters and the handover parameters of the cellular network 2 are managed by a network management system 13 as described above with reference to Figure 1 -5.

As a result of the cell selection and handover rules the premium user 53 is served with cells with good coverage. The average user 52 and the economy user 51 are attached to different cells with high probability. The economy user 51 will be attached to carriers with partial coverage, therefore, he is enforced to more frequent inter-frequency handover than the other users. Cells serving premium users may have lower handover priorities for non-premium users in order to moderate load. At some point, see area 76, there is coverage only at one of the frequency band, so all three users are handover to the same cells 62, 63. Since in this embodiment the handover priority and neighbor definition are different for economy, average and premium users, they will handover to different cells with different service levels whenever possible. In this example, the economy user 51 will be handed over to call 72 and the average user 52 will be handed over to cell 68. The economy user 51 will need to perform another inter frequency handover to cell 69, see Figure 1 1 . In Figure 1 1 , there is service differentiation in the full network.

The solutions described above enable operators to continuously monitor the KPIs for different user groups: premium, average, economy. It provides QoS and GoS monitoring for different user groups and services. It ensures QoS and GoS differentiation for the different user groups. The network management system 13 checks and indicates if the KPIs meet their target values for different user groups. Since the network parameters are modified each time that at least one monitored KPI does not meet its target value, the measured KPIs are automatically improved during operation. The neighbor relation barring matrix can be used to license certain cell selection or IRAT handover, as well as QoS differentiation. I.e. one can define a rule that adds all IRAT relation to the barring matrix if the required license for IRAT handover is missing. Figure 12 is a flow chart of a method of configuring at least one cellular network according to an embodiment of the invention. The method comprises the step 121 in which performance data is received from the one or more sub-network management systems 3'. Then in a step 122 values are calculated for one or more KPIs using the performance data. In a step 123 the calculated KPI values are compared with target values, to render off-target KPIs. If one or more off-target KPIs occur, see test 124, a step 125 is performed in which a set of rules is selected associated with the off-target KPIs and said set of rules is provide to each sub-network management system 3'. Those skilled in the art will also appreciate that the various features and functionalities described above may be performed by a combination of analogue and digital circuits, and/or one or more processors configured with software and/or firmware (e.g., stored in memory) that, when executed by the one or more processors, perform as described above. One or more of these processors, as well as the other digital hardware, may be included in a single application-specific integrated circuit (ASIC), or several processors and various digital hardware may be distributed among several separate components, whether individually packaged or assembled into a system-on-a-chip (SoC).

It is noted that the scope of the invention is not limited to the embodiments described above. The above described embodiments clarify the aspects of the invention the person skilled in the art can use to implement the invention for a specific communication network having a specific mix of sub networks. Other variants from the above will be apparent for the skilled person and will be part of the invention as well. The scope of the present invention is only limited to the features mentioned in accompanying claims.

Abbreviations:

ATM Asynchronous Transfer Mode

BE Best effort

CS Circuit Switched

DSL Digital Subscriber Line

EBER Estimated Bit Error Rate

GBR Guaranteed bit rate

GoS Grade of Service

GSM Global System for Mobile communication

HO Handover(s)

I RAT Inter radio technology

ISDN Integrated Services Digital Network

KPI Key Performance Indicator

LTE Long Term Evolution

O&M Operation and maintenance

OSS Operational support subsystems

PS Package Switched

PDN Public data network

PID proportional-integral-derivative

QoS Quality of Service

RAB Radio Access Bearer

SNMS sub-network management system

TDM Time-division multiplexing

WCDMA Wireless Code Division Multiple Access

WIFI Wireless Fidelity

WLAN Wide local area network

Claims

Claims

1 . Network management system (13) for a communication network (1 ), said

communication network (1 ) having one or more sub-networks (2) each being managed by a sub-network management system (3'), wherein said network management system (13) comprises;

- a KPI calculator (20) configured for receiving performance data from said subnetwork management system (3'), and for determining KPI values on basis of said performance data;

- a KPI comparator module (22) configured for receiving said KPI values from said calculator (20), comparing said KPI values with KPI target values and outputting identifiers of off-target KPIs not meeting their KPI target values,

- a controller module (24) configured for receiving said identifiers of said off-target KPIs, and selecting a selected rule set based on the off-target KPIs, and outputting a set indicator of said selected rule set,

- a rule set repository (26) configured for storing a plurality of rule sets, each of said rule sets being associated with a rule set identifier, and for providing an output rule set out of said plurality of rule sets for which the rule set identifier matches said set indicator of said selected rule set, to each of said sub-network

management systems (3').

2. Network management system (13) according to claim 1 , wherein said controller module (24) comprises a state machine, a state of which is defined by a state table comprising in each row an indicator for the current state, the new state, the condition for changing the state, and an indicator of a rule set.

3. Network management system (13) according claim 2, wherein the condition field of said table contains a KPI identifier.

4. Network management system (13) according to claim 2, wherein the condition field contains a Boolean expression of KPI identifiers.

5. Network management system (13) according to any of the preceding claims, wherein said KPI comparator (22) comprises: - a PID module (223) for performing PID processing on said KPI values,

- a threshold comparator (227) for comparing an output of said PID module (223) with KPI target values.

6. Network management system (13) according to claim 5, wherein said KPI comparator (22) further comprises a prioritizer (229) configured for selecting one off-target KPI out of a number of off-target KPIs having the highest priority.

7. Network management system (13) according to any of the preceding claims, wherein said KPI calculator (20) is configured for calculating KPI values for each item in one or more categories.

8. Network management system (13) according to claim 7, wherein a category is one of service types, service levels or geographic areas (61 ;62) of said sub-networks (2).

9. Network management system (13) according to claim 7 or 8, wherein said KPI comparator (22), said controller module (24) and said rule set repository (26) are configured to process one item of one category at a time provided by the KPI calculator (20).

10. Network management system (13) according to claim 7 or 8, wherein said controller module (24) and said rule set repository (26) are configured to process one item of one category at a time provided by the KPI comparator (20).

1 1 . Network management system (13) according to claim 7 or 8, wherein said KPI comparator (20), said controller module (24) and said rule set repository (20) are configured to process each item in each category in parallel.

12. Network management system (13) according to any of the preceding claims, wherein rules comprised in said rule sets are supplemented with sub-network indicators indicating for which sub-network management system (3') the rules are valid.

13. Network management system (13) according to any of the claims 1 -1 1 , wherein said rule sets each comprise several subsets of rules, said subsets each being associated with a sub-network indicator so as to be sent to the management system (3') of the indicated sub-network.

14. Network management system (13) according to any of the preceding claims, wherein said rule set repository is configured to store for each rule set a version number indicating which version of a rule set is stored by each specific sub-network

management system (3'), and configured to send rules set identifiers to the specific subnetwork management system (3') to indicate which rule set shall be executed.

15. Network management system (13) according to any of claims 12-14, wherein a rule comprises a place holder that is substituted by said rule set repository before the rule is send to the sub-network (3').

16. Network management system (13) according any of claims 12-14, wherein a rule comprises a place holder that is to be substituted by the receiving O&M/OSS.

17. Network management system (13) according to any of the preceding claims, wherein a rule in the output rule set provides one or more of the following:

- a criterion for I RAT HO,

- a dividing algorithm between GBR and BE traffic of available bandwidth,

- a selection rule for traffic diversion on MPLS,

- a intra cell HO criterion,

- a communication service grant limiting algorithm.

18. Method of managing a communication network on basis of Key Performance

Indicators (KPIs), said communication network comprising one or more sub-networks each having a sub-network management system (3'), wherein said method comprises:

- receiving performance data from said one or more sub-network management systems,

- calculating values for one or more KPIs from said performance data,

- comparing calculated KPI values with target values, to render off-target KPIs,

- if one or more off-target KPIs occur:

o selecting a set of rules associated with said off-target KPIs,

o providing said set of rules to each sub-network management system (3').