AFTER HANDOVER FOR TRIGGERING HANDOVER

TECHNICAL FIELD

Embodiments herein relate to wireless communication systems, such as cellular radio communication systems. In particular, a method and a network node for managing a handover of a user equipment from a first radio network node to a second radio network node are disclosed. A corresponding computer program and a carrier therefor are also disclosed. BACKGROUND

In cellular radio communication systems, such as mobile networks, a process known as handover is used to provide e.g. mobility for a user equipment (UE) and load balancing between two base stations. The handover is typically signified by that the user equipment is served by one of the base stations, aka a source base station, before the handover and by the other base station, aka a target base station, after the handover.

For load balancing purposes, a load of a base station is measured. When the load reaches above a pre-configured threshold, a Mobility Load Balancing (MLB) procedure is triggered. The MLB procedure allows balancing, i.e. transfer of, the load to a neighbor cell of the same Radio Access Technology (RAT), a cell of another RAT or another frequency or the like.

Currently, Third Generation Partnership Project (3GPP), Technical Specification (TS) 36.423 X2AP, section 8.3.7 and section 8.2, specifies the following functions of the MLB procedure:

· a load reporting function,

• a load balancing function, and

• a handover function.

The load reporting function includes an exchange of cell specific load information between neighbor base stations, such as enhanced NodeBs (eNBs), e.g. as exemplified in TS 36.423, X2AP, section 8.3.7 and 9.1 .2.1 .

The load balancing function describes decisions, by the base station, about which user equipments to be handed over and to which neighbor cells. This is sometimes referred to as UE selection and cell selection, respectively. These decisions are typically taken based mainly on the load information, provided by the load reporting function, and any available radio measurements performed by one or more user equipment served by a source base station.

A known exemplifying manner of performing the load balancing function is to effectively change size of a source cell operated by the source base station. For example, if the cell is overloaded, the size of the source cell may be decreased. This implies that user equipments at the outskirts of the source cell will be handed over to a neighbor cell, which size can be effectively increased. It is known to the change of the size of the cell by applying a handover (HO) offset. The source base station negotiates with a target base station, operating a target cell, for the HO offset settings to avoid handover Ping-Pong between the source and target cells. The agreed offset will be signaled to the user equipments served by the source base station.

Another known exemplifying manner of performing the load balancing function is to select a specific set of user equipments which should be handed over to one or more target cells, operated by one or more target base stations. The selection of the specific set of user equipments can take the following information into account: load and capacity of source and target cell, UE radio measurement reports, UE traffic characteristics, UE bearers information, historical/current resource utilization of the UE, UE subscription, profile. The UE traffic characteristics can be given as heavy or light data usage. The UE bearer information can be given as guaranteed bit-rate bearer or default bearer. The UE subscription profile may be given as gold, silver or bronze subscriber.

The handover function performs the handover of the selected user equipment from the source base station to the target base station. During execution of the handover function, mobility settings are negotiated between the source and target base stations. In this manner, it may be avoided that the selected user equipment is for example handed over back to the source base station as a consequence of its' mobility, when it initially was handed over to the target base station due to load balancing.

For mobility purposes, a signal strength received at the user equipment from the source base station is measured. When the signal strength at the user equipment passes below a threshold value, the user equipment begins to search for one or more target base stations, to which it potentially can be handed over. In this case, the source base station decides to which target base station the user equipment is to be handed over.

In some cases, that the source base station may, by its decision regarding which user equipment to handover and to which target base station, cause a performance degradation of the overall system, and in particular for the user equipment selected to be handed over. A problem, related to handover, e.g. caused by mobility or load, may thus be how to enable the source base station to make an improved decision. SUMMARY

An object may be to eliminate, or at least reduce, the above mentioned problem.

According to a first aspect, the object is achieved by a method, performed by a network node, for managing a handover of a user equipment from a first radio network node to a second radio network node. Before the handover, the network node obtains a set of parameters, wherein the set of parameters comprises a radio measurement parameter relating to radio measurements, which radio measurements are performed by the user equipment towards the second radio network node, wherein the user equipment is served by the first radio network node. After the handover, the network node receives, from the second radio network node, a report including a quality of service value relating to quality of service for the user equipment when served by the second radio network node. Subsequently, the network node trains a set of models for prediction of the quality of service for the user equipment, wherein the training comprises adapting each model of the set of models using the quality of service value of the report as a ground truth to be predicted given a respective subset of the set of parameters, which respective subset is taken as input to said each model, wherein each respective subset ranges from including one parameter of the set of parameters to all parameters of the set of parameters. According to a second aspect, the object is achieved by a network node configured to manage a handover of a user equipment from a first radio network node to a second radio network node. The network node is configured to obtain, before the handover, a set of parameters. The set of parameters comprises a radio measurement parameter relating to radio measurements, which radio measurements are performable by the user equipment towards the second radio network node, wherein the user equipment is servable by the first radio network node. Furthermore, the network node is configured to receive, after the handover, a report including a quality of service value relating to quality of service for the user equipment when served by the second radio network node. The report is received from the second radio network node. Moreover, the network node is configured to train a set of models for prediction of the quality of service for the user equipment, wherein the network node is configured to train the set of models by adapting each model of the set of models using the quality of service value of the report as a ground truth to be predicted given a respective subset of the set of parameters, which respective subset is taken as input to said each model, wherein each respective subset ranges from including one parameter of the set of parameters to all parameters of the set of parameters.

According to further aspects, the object is achieved by a computer program and a carrier for the computer program corresponding to the aspects above.

By use of the set of parameters and the report including the quality of service value, the network node trains the set of models for prediction of the quality of service value for the user equipment after the handover. In this manner, the network node effectively builds the set of models, such as predictive models for predicting the quality of service value, using post-handover information, e.g. in the form of the report, as the ground truth to be predicted. These models enable the network node to select a user equipment and/or a target cell for handover based on a predicted quality of service value calculated by means of one of the models of the set of models. The selected user equipment may appear to benefit, or at least not degrade, in terms of quality of service from being handed over to the target cell. In this context, the quality of service value is a measure of performance for the selected user equipment. In this manner, the first radio network node is enabled to make an improved handover decision for selection of user equipment and/or target radio network node, such as the second radio network node.

An advantage is that the set of models thus obtained enable enhanced selection of target cell, such as the second radio network node, and the user equipment. The enhanced selection may take the predicted quality of service value into account. BRIEF DESCRIPTION OF THE DRAWINGS

The various aspects of embodiments disclosed herein, including particular features and advantages thereof, will be readily understood from the following detailed description and the accompanying drawings, in which:

Figure 1 is a schematic overview of an exemplifying wireless communication system in which embodiments herein may be implemented,

Figure 2 is a schematic combined signaling and flowchart illustrating

embodiments herein,

Figure 3 and 4 are flowcharts illustrating further details of the embodiments herein,

Figure 5 is a schematic flowchart illustrating an embodiment of the method when performed by a central network node,

Figure 6 is a schematic flowchart illustrating an embodiment of the method when performed by the first radio network node, and

Figure 7 is a block diagram illustrating embodiments of the network node.

DETAILED DESCRIPTION

In order to better appreciate the benefits and advantages of the embodiments herein, the following analysis is provided here.

In view of the problem stated in the background section, it has been realized that the MLB procedure, and in particular the load balancing function, cannot verify what impact, e.g. to quality of service for the selected user equipment, a handover may have. Therefore, an enhancement of the load balancing function is to enable selection of user equipment and target cell based on a predicted impact, e.g. in terms of a predicted quality of service after the handover.

Instead of using static models for the selection of the user equipment and/or the target cell, it is herein proposed to use a self-learning algorithm. The self-learning algorithm advantageously overcomes shortcoming of the static models, such as requirement to update the model, difficulty in finding the static model and low expected accuracy of static models. It has been demonstrated that it is possible to predict, by machine learning methods, a throughput value for a user equipment at a certain point in time based on historical data and a set of parameters currently applying to a connection towards the user equipment.

The machine learning methods, such as the above mentioned self-learning algorithm, include one or more of the following prediction models: a Nonlinear

Autoregressive Exogenous Model using a Wavelet Network as its nonlinearity estimator; a Nonlinear Autoregressive (NAR) model using a Neural Network Time Series, a feedforward Neural Network with 10 or 15 neurons and one hidden layer, Support Vector Machine (SVM) and the like.

In a first use case, throughput prediction was studied. A machine learning method was evaluated, using one of the prediction models for all UEs in the system. The model was trained and updated offline, e.g. during night.

In a second use case, prediction of time for download of a file, e.g. using File- Transfer-Protocol (FTP) was studied. Two Machine Learning methods have been evaluated; a prediction model per cell, valid for all UEs in that cell and a prediction model per UE, valid for that particular UE during the lifetime of the connection.

Using the models, described in the two use cases above, a prediction of how long time a download of a file will last may be obtained or a prediction of which throughput a user equipment will experience may be obtained.

Using system level simulations for an LTE system, the following features were evaluated as input to the Machine Learning methods:

1 . Number of active/schedulable users per Transmission Time Interval (TTI) (1 ms) 2. Number of scheduled users per TTI

3. Number of remaining physical resource blocks (PRBs) after scheduling per TTI (50 MHz = 50 PRBs)

4. Total number of bits in the downlink buffers for all users per TTI (bits)

5. Current average throughput for active users in the cell (Mbps) (active = TTIs

where user is scheduled but still have data left in the buffer)

6. Current total cell throughput (Mbps)

7. UE SINR (dBm), an estimation made by the eNB about the signal quality in the downlink, partly based on the channel quality indicator (CQI) reported by the UE

8. UE Rank (e.g. a value of 1 or 2), reported by the UE together with the CQI Two conclusions based on the above are that (1 ) good prediction can be provided for many users and at many different points in time, large deviances are rare and (2) small variations in throughput and/or time for downloading can be captured.

In the context of predicting the throughput of a UE using a minimum set of features, the following features were evaluated as essentials:

• CQI and UE Rank, which describe a point-to-point communication link and enable estimation of a rate at a physical layer of a UE. The CQI and Rank are mapped to modulation and coding scheme, which in turn define an achievable rate at the physical layer, e.g, a high order modulation may achieve a high rate, but is very sensitive to noise.

• Number of UEs in the cell, which enables the base station to know between how many UEs the available bandwidth of the base station is shared.

• History of CQI of a UE, since the channel quality is changing, knowing the past values of the CQI enables an accurate prediction of the evolution of the channel quality.

Other valid features include UE radio condition measures, such as Reference Signal Received Power (RSRP) and Reference Signal Received Quality (RSRQ) and measures relating to load of cell.

The present inventors have realized that these machine learning methods may advantageously be used to improve performance of a wireless communication system in connection with a handover.

Throughout the following description similar reference numerals have been used to denote similar features, such as nodes, actions, steps, modules, circuits, parts, items elements, units or the like, when applicable. In the Figures, features that appear in some embodiments are indicated by dashed lines.

Figure 1 depicts an exemplifying wireless communication system 100 in which embodiments herein may be implemented. In this example, the wireless communication system 100 is a Long Term Evolution (LTE) network. In other examples, the wireless communication system 100 may be any cellular or wireless communication system, such as a Global System for Mobile Communications (GSM), Universal Mobile

Telecommunication System (UMTS) and Worldwide Interoperability for Microwave Access (WiMAX), Wireless Local Area Network (WLAN), a wireless network according to 802.1 1 -standards suite or the like.

The wireless communication system 100 comprises a network node 110, which in this example includes a central network node 111 and a first radio network node 130, such as a source radio network node 130.

Additionally, the wireless communication system 100 comprises a second radio network node 140, such as a target radio network node 140.

The central network node 1 1 1 may be a Mobility Management Entity (MME), Operation and Support System (OSS) or a similar network node for handling

coordination between multiple radio network nodes, such as the first and second radio network nodes 130, 140. Moreover, the wireless communication system 100 comprises one or more user equipments 120, 121. One or more of said user equipments 120, 121 may be located in a first cell C1 or in a second cell C2. The first cell C1 may be operated by the first radio network node 130 and the second cell C2 may be operated by the second radio network node 140. The first radio network node 130 may serve a plurality of user equipments 120, 121 , including the user equipment 120.

The first radio network node 130 may communicate 141 , e.g. send information to and/or receive information from, the central network node 1 1 1 . The communication 141 may be performed via an S1 interface in case of LTE.

The first radio network node 130 may communicate 142, e.g. send information to and/or receive information from, the second radio network node 140. The

communication 142 may be performed via an X2 interface in case of LTE. In case of a multi-RAT scenario, or inter-RAT scenario, a Radio Access Network (RAN) Information Message (RIM) interface between e.g. a eNB, such as the first radio network node 130, and a Radio Network Controller (RNC), such as the second radio network node 140, may be employed.

The first radio network node 130 may communicate 143, e.g. send information to and/or receive information from, the one or more user equipments 120, 121 . The communication 143 may be performed via a so called uU interface in case of LTE.

The second radio network node 140 may communicate 144, e.g. send information to and/or receive information from, the central node 1 1 1 . The communication 144 may be performed via a S1 interface in case of LTE. The second radio network node 140 may communicate 145, e.g. send information to and/or receive information from, the one or more user equipments 120, 121 . The communication 145 may be performed via a uU interface in case of LTE. In various embodiments, the above illustrated scenario may be a inter-Radio

Access Technology (RAT) scenario, a intra-RAT scenario, a multi-RAT scenario or the like.

As used herein, the term "radio network node" may refer to an evolved Node B (eNB), a Radio Network Controller (RNC), a Radio Base Station (RBS), a base station, a base station controller, a control node controlling one or more Remote Radio Units (RRUs), an access point or the like.

As used herein, the term "user equipment" may refer to a wireless device, a machine-to-machine (M2M) device, a mobile phone, a cellular phone, a Personal Digital Assistant (PDA) equipped with radio communication capabilities, a smartphone, a laptop or personal computer (PC) equipped with an internal or external mobile broadband modem, a tablet PC with radio communication capabilities, a portable electronic radio communication device, a sensor device equipped with radio communication capabilities or the like. The sensor may be any kind of weather sensor, such as wind, temperature, air pressure, humidity etc. As further examples, the sensor may be a light sensor, an electronic or electric switch, a microphone, a loudspeaker, a camera sensor etc. The term "user" may indirectly refer to the user equipment.

The method disclosed herein includes a training phase. In the training phase, a set of models for prediction of a quality of service value are trained to predict the quality of service value based on a set of parameters. Actions of the training phase, as will be described in more detail below, may be performed by the central node 1 1 1 or the first radio network node 130. The training phase may thus include one or more of actions 203, 207, 210, 21 1 and 212, which are described below. The training phase may also include building of models, where different models take different parameters as input, e.g. different number of parameter and/or different types of parameters etc.

Furthermore, the method disclosed herein includes an execution phase. In the execution phase, the set of models, trained during the training phase or even trained during the execution phase, are used when selecting, by the first radio network node 130, a user equipment to be handed over and to which target radio network node, such as the second radio network node 140, the user equipment should be handed over. The execution phase may thus include one or more of actions 201 , 204-206, which are described below.

Hence, the network node 1 10 of Figure 1 may refer to the central node 1 1 1 or the first radio network node 130 unless otherwise indicated.

In some examples, the training phase may be performed a number of times in order to make the set of models accurate enough, before the execution phase may be performed.

However, in some examples, the set of models are provided with default values which are trained inline during the execution phase. The default values may be random values or some values that are assumed to give reasonably accurate predictions for many scenarios. Said some values may have been obtained during training in a cellular radio communication system 100 including a typical scenario. The scenario may be typical in terms on distances between radio network nodes of the cellular radio communication system 100, and/or in terms of number of user equipments served by a source radio network node, and/or in terms of load at potential target radio network nodes and/or load at the source radio network node etc.

With this in mind, Figure 2 will now be described in more detail. Figure 2 illustrates an exemplifying method according to embodiments herein when performed in the wireless communication system 100 of Figure 1 . Thus, the network node 1 10 performs a method for managing a handover of the user equipment 120 from the first radio network node 130 to the second radio network node 140. The following actions may be performed in any suitable order.

Action 201

In order to trigger, or initiate, a process of selecting a user equipment 120 and/or a target radio network node to be used when performing e.g. the handover function mentioned in the background section, the network node 1 10 may detect that the handover of at least one of the plurality of user equipments is evaluable. This may mean that the network node 1 10 may detect that the handover is to be evaluated, such as should be evaluated by performing the actions 204, 205 and 206. See also action 302 in Figure 3 and 401 in Figure 4. The detection of that the handover should be evaluated may be that a current quality of service value for at least one of the plurality of user equipments has passed, e.g. is below or above, a threshold value for evaluating handover from the first radio network node 130. When the current quality of service is below the threshold value, the handover may be evaluated in order to avoid performance degradation, e.g. in terms of dropped calls, interrupted service etc. When the current quality of service is above the threshold value, the handover may be evaluated in order to perform load balancing.

The detection of that the handover should be evaluated may be that a measurement report, received from at least one user equipment, indicates that the handover is evaluable.

The detection of that the handover should be evaluated may be that a current load at the first radio network node 130 is above a threshold value for indicating load at the first radio network node 130 at which handover is evaluable.

The detection of that the handover should be evaluated may be that a neighbour load at the second radio network node 140 is below a neighbour load threshold for indicating load at the second radio network node 140 at which handover is evaluable.

In embodiments, in which actions 204, 205 and 206 are performed, the network node 1 10 may be the first radio network node 130.

Actions 204, 205 and 206 may preferably be performed before the handover. In some examples, the actions 204, 205 and 206 are performed only when action 201 implies that the handover should be evaluated. In other examples, these actions may be performed periodically, e.g. trigger by a timer. This may mean that actions 204, 205 and 206 are performed once during a specified time interval, such as 2 seconds or similar. This time scale is matching the working time scale of the MLB function mentioned in the background section.

To conclude, the handover may be caused by overload at the first radio network node 130, or the handover may be caused by mobility of the user equipment 120. In further examples, the handover may be periodically evaluated, i.e. the handover is caused at regular time intervals, e.g. by a timer event.

Action 202 The second radio network node 140 may send one or more parameters of a set of parameters.

These one or more parameters may include a load parameter relating to load in the second radio network node 140. See further parameters in list in connection with action 203 below.

Action 203

Before the handover, the user equipment 120 is served by the first radio network node 130. Also before the handover, the network node 1 10 obtains, such as receives, the set of parameters.

One or more parameters of the set may be received from the user equipment 120 (not shown in Figure 2). Alternatively or additionally, one or more parameters of the set may be received from the second radio network node 140.

The set of parameters comprises a radio measurement parameter relating to radio measurements. The radio measurements are performed by the user equipment 120 towards the second radio network node 140.

The set of parameters may comprise one or more of:

the load parameter relating to load in the second radio network node 140, a capability information relating to capability of the user equipment 120, a channel quality indicator parameter relating to one or more Channel Quality

Indicators (CQI), towards the second radio network node 140, which CQI is reported by the user equipment 120,

a rank parameter relating to rank of a connection, e.g. a potential connection towards the second radio network node 140, to the user equipment 120,

a signal strength parameter relating to received signal strength reported by the user equipment 120, where the signal strength is received from the second radio network node 140,

an amount parameter relating to an amount of user equipments served by the first and/or radio network node 130, 140,

a resource parameter relating to utilization of physical resource blocks in a downlink from the second radio network node 140,

a bearer parameter relating to a bearer type of a bearer associated to the user equipment 120, the bearer parameter may indicate a default bearer, a VoLTE bearer, etc., and the like.

Action 204

For each user equipment 120, 121 of the plurality of user equipments 120, 121 , the network node 1 10 may select a model out of the set of models according to the parameters in the obtained set of parameters, e.g. number of parameters and type of parameters in the set. See also action 403 in Figure 4.

As an example, the set of models may comprise a first model M1 = f(P1 , P2, P3), where P1 , P2, P3 are any three of the parameters of the set of parameters listed above, and a second model M2 = g(P1 , P2, P3, P4). f and g may be any two of the above mentioned self-learning algorithms. In this action the first model M1 would be selected if only P1 , P2 and P3 have been obtained in action 203. In case P1 , P2, P3 and P4 have been obtained, the first and/or second model would be selected.

In some cases, an accuracy of the models may be calculated if action 209 has been performed. Then, the first model may be selected if its accuracy is better than an accuracy of the second model.

The selection of a model may be performed at regular, or irregular, time intervals. When the first radio network node 130 makes a prediction for e.g. the user equipment 120, it may make several predictions using different models. Later on when a user equipment has to be handed over, the first radio network node 130 may use the currently most accurate model. After the user equipment has been handed over, the second radio network node 140 may report the performance of the user equipment to the first radio network node 130. The first radio network node 130 may then compare the ground truth, i.e. the reported performance, to all the predictions it did with the different models. Using this comparison the first radio network node 130 may calculate the accuracy of each of the models.

Action 205

For each user equipment 120, 121 of the plurality of user equipments 120, 121 , the network node 1 10 may determine a predicted quality of service value for said each user equipment 120 while using the selected model.

In case multiple models have been selected in action 204, the network node 1 10 may determine a respective predicted quality of service value for each selected model. Action 206

The network node 1 10 may select a user equipment 120 out of the plurality of user equipments 120, 121 for which the predicted quality of service value is above a threshold value for selecting one or more user equipments to be handed over from the first radio network node 130.

The threshold may indicate one or more of:

a current quality of service value for the user equipment 120, 121 ;

an average current quality of service value for the plurality of user equipments 120, 121 ;

an average predicted quality of service value for the plurality of user equipments

120, 121 ; and the like.

When the threshold indicates the current quality of service, it may be ensured that the quality of service for the selected user equipment, after handover, is improved, i.e. increased, or at least maintained.

When the threshold indicates the average current quality of service, it may be ensured the handed over user equipment obtains a QoS that is better than average current QoS for the user equipments served by the first radio network node, i.e. a source cell, when served in a target cell, e.g. operated by the second radio network node 140.

When the threshold indicates the average predicted quality of service value for the plurality of user equipments 120, 12, it may be ensured that the handed over user equipment obtains a QoS that is among the best, highest, as compared to if any other user equipment would be handed over instead.

Other criteria include for example the selection of the user equipment for which the increase, or improvement, of QoS is the highest, where the increase of QoS may be measured in relative or absolute values. E.g. a QoS value of 3 before the handover and a QoS value of 5 after the handover for one user equipment would be considered better than a QoS value of 1 before the handover and a QoS value of 2 after the handover if the increase is measured in absolute values. However, if the increase of QoS is measured in relative values, the latter case would imply a relative increase of 100%, thus better than, or higher than, a relative increase of 67% for the former case. For this other criteria, the selected user equipment is that for which an increase of the QoS value is the highest, or among the highest, of increases of QoS determined for all, or almost all, user equipments served by the first radio network node 130. Those user equipments for which the estimated increase, or potentially decrease, of QoS is determined may be candidates for handover. Candidates for handover may be selected in various manners, e.g. based on current QoS.

Action 207

This action only applies when the network node 1 10 is the first radio network node 130.

The set of parameters may comprise capability information relating to capability of the user equipment 120, often referred to as "UE capability". More generally, the set of parameters may comprise a so called UE Context, which is known from 3GPP terminology. Accordingly, the UE Context may comprise the UE capability.

In contrast to prior art, the first radio network node 130 may keep the capability information available for, or e.g. to be used by, action 205 below. This may mean that the capability information is stored such that it can be retrieved also after the handover is completed.

At this stage, the handover is completed, see e.g. also action 307 of Figure 3 and action 408 of Figure 4.

Action 208

The second radio network node 140 may measure a quality of service value relating to quality of service for the user equipment 120 when served by the second radio network node 140.

The quality of service, which is a measure of performance, may thus have been measured by the second radio network node 140, e.g. during a first time period. In this manner, the quality of service value may be obtained. The first time period may be a pre- configured time period.

The quality of service value may comprise one or more of:

a throughput value relating to throughput of transmission between the second radio network node 140 and the user equipment 120;

a delay value relating to delay of transmission between the second radio network node 140 and the user equipment 120;

a packet loss rate relating to loss of packets in transmission between the second radio network node 140 and the user equipment 120; a packet error rate relating to error in packets of transmission between the second radio network node 140 and the user equipment 120;

a reliability value relating to reliability of a connection carrying transmission between the second radio network node 140 and the user equipment 120; and

the like.

The throughput value may comprise an uplink throughput value relating to throughput from the user equipment 120 to the second radio network node 140 and/or a downlink throughput value relating to throughput from the second radio network node 140 to the user equipment 120.

The measured performance, e.g. the quality of service value, for Non-GBR bearers may be the distribution of an achieved bit rate measured during a pre-configured time T1 after the handover. The measured performance for GBR bearers may also be the bit rate distribution or the binary distribution of fulfilment, e.g. 1 's, or non-fulfilment, e.g. 0's, of the requested bit rate. In the case of LTE, the performance measurement period starts when the second radio network node 140 receives from the user equipment an RRC CONNECTION RECONFIGURATION MESSAGE (Handover complete), i.e. part of the handover function. At this point, UE context has already been transferred to the second radio network node 140 so it is possible to associate the measurement to that specific user equipment. This is needed for further correlation at the first radio network node 130, e.g. in action 21 1 and/or 212.

Action 209

The second radio network node 140 may send a report including the quality of service value. Accordingly, the quality of service value is a true value of what the quality of service became for the user equipment 120 after the handover.

Action 210

After the handover, the network node 1 10 receives, from the second radio network node 140, a report including a quality of service value relating to quality of service for the user equipment 120 when served by the second radio network node 140.

Action 211

Before the action of training, but after action 210, the network node 1 10 may correlate the report to the obtained set of parameters by use of an identity of the user equipment 120. The report and the obtained set of parameters are identifiable by means of the identity. The identity may be a International Mobile Subscriber Identity (IMSI), a Temporary Mobile Subscriber Identity (TMSI), a Cell Radio Network Temporary Identifier (C-RNTI) or the like. See also action 309 of Figure 3.

The correlation may be per user equipment or per data bearer associated to the user equipment.

Action 212

The network node 1 10 trains a set of models for prediction of the quality of service for the user equipment 120, wherein the training 205 comprises adapting each model of the set of models using the quality of service value of the report as a ground truth to be predicted, e.g. using said each model, given a respective subset of the set of parameters, which respective subset is taken as input to said each model, wherein each respective subset ranges from including one parameter of the set of parameters to all parameters of the set of parameters.

Figure 3 illustrates a schematic combined signaling and flow chart according to embodiments herein. In this embodiment, the first radio network node 130 performs the training phase in order to build the set of models. In this example, the handover is triggered by that the first radio network node 130 becomes overloaded.

The following exemplifying actions are performed.

Action 301

The user equipment 120 sends a measurement report to the first radio network node 130. The measurement report may include one or more of RSRP, RSRQ, and the like, for the second radio network node 140.

Action 302

The first radio network node 130 detects overload. This may mean that the load on the first radio network node 130 is above a load threshold value. The load may be measured in terms of number of dropped packets, number of served user equipments, total throughput for a cell operated by the first radio network node 130, etc.

Action 302 may be performed before action 301 . Action 303

The first radio network node 130 sends a X2AP RESOURCE STATUS

REQUEST, e.g. in response to action 302, i.e. when overload is detected.

Action 304

The second radio network node 140 sends a X2AP RESOURCE STATUS update. Action 305

Part of the handover function is executed.

Action 306

The first radio network node 130 saves values of the set of parameters. The first radio network node 130 may store the values of the set of parameters during a number of seconds, e.g. up until action 310 has been performed. Thereafter, UE context in case of LTE may be released. This action may be performed before action 305.

Action 307

The user equipment 120 sends a RRC RECONFIGURATION COMPLETE message to the second radio network node 140, whereby it is signaled that the handover is completed.

Action 308

The second radio network node 140 sends a report about performance measurements. The report includes a quality of service value, being an actual QoS obtained for the user equipment after the handover. This action is performed after action 307.

It may here be mentioned that, in 2010, the Next Generation of Network

Management (NGNM) forum issued a recommendation concerning the monitoring of throughput before/after handovers. In addition to that, 3GPP working group (WG) RAN3 proposed to discuss solutions to enable post-handover throughput monitoring for non- GBR bearers [R3-141 1 13]. With the embodiments herein, the report generally includes QoS parameters, possibly excluding the throughput, which is mentioned in the above documents.

Action 309

The first radio network node 130 correlates the received performance measurements with the saved values of the set of parameters by use of an identity, which uniquely identifies the user equipment 120.

Action 310

At least one model that uses at least a sub-set of the available parameters of the set is trained to predict the actual QoS.

Figure 4 illustrates a schematic combined signaling and flow chart according to embodiments herein. In this embodiment, the first radio network node 130 performs the execution phase, thereby enhancing the handover procedure. In this example, the handover is trigger by that the first radio network node 130 becomes overloaded.

The following exemplifying actions are performed. Action 401

The first radio network node 130 detects overload. This may mean that the load on the first radio network node 130 is above a load threshold value.

Action 402

The user equipment 120 sends a measurement report to the first radio network node 130. The measurement report may include one or more of RSRP, RSRQ, and the like, for the second radio network node 140.

Action 402 may be performed before action 401 . Action 403

The first radio network node 140 selects a model.

Action 404 The first radio network node 130 sends a X2AP RESOURCE STATUS

REQUEST. The RESOURCE STATUS REQUEST may be adapted to the selected model, i.e. only those parameters used by the model are requested. In this manner, valuable bandwidth may be saved, since parameters not used by the model will not be send in the subsequent RESOURCE STATUS UPDATE in action 405.

Action 405

The second radio network node 140 sends one or more X2AP RESOURCE STATUS UPDATES.

Action 406

The first radio network node 130 selects the user equipment to be handed over and to which target radio network node to hand it over. Action 407

Part of the handover function is executed.

Action 408

The user equipment 120 sends a RRC RECONFIGURATION COMPLETE message to the second radio network node 140, whereby it is signaled that the handover is completed.

In Figure 5, a schematic flowchart of exemplifying methods in the central network node 1 1 1 is shown. Again, the same reference numerals as above have been used to denote the same or similar features, in particular the same reference numerals have been used to denote the same or similar actions. In these embodiments, the network node 1 10 is the central network node 1 1 1 . Accordingly, the central network node 1 1 1 performs a method for managing a handover of a user equipment 120 from a first radio network node 130 to a second radio network node 140.

As mentioned, the handover may be caused by overload at the first radio network node 130, or the handover may be caused by mobility of the user equipment 120.

The following actions may be performed in any suitable order. Only those actions that may be performed by the central network node 1 1 1 are illustrated here. Action 203

Before the handover, the central network node 1 1 1 obtains a set of parameters, wherein the set of parameters comprises a radio measurement parameter relating to radio measurements, which radio measurements are performed by the user equipment 120 towards the second radio network node 140, wherein the user equipment 120 is served by the first radio network node 130.

The set of parameters may comprise one or more of:

a load parameter relating to load in the second radio network node 140, a capability information relating to capability of the user equipment 120, a channel quality indicator parameter relating to one or more channel quality indicators reported by the user equipment 120,

a rank parameter relating to rank a connection to the user equipment 120, a signal strength parameter relating to received signal strength reported by the user equipment 120,

an amount parameter relating to an amount of user equipments served by the first radio network node 130,

a resource parameter relating to utilization of physical resource blocks, a bearer parameter relating to a bearer type of a bearer associated to the user equipment 120, and the like.

Action 210

After the handover, the central network node 1 1 1 receives, from the second radio network node 140, a report including a quality of service value relating to quality of service for the user equipment 120 when served by the second radio network node 140.

The quality of service value may comprise one or more of:

a throughput value relating to throughput of transmission between the second radio network node 140 and the user equipment 120;

a delay value relating to delay of transmission between the second radio network node 140 and the user equipment 120;

a packet loss rate relating to loss of packets in transmission between the second radio network node 140 and the user equipment 120;

a packet error rate relating to error in packets of transmission between the second radio network node 140 and the user equipment 120; a reliability value relating to reliability of a connection carrying transmission between the second radio network node 140 and the user equipment 120, and the like.

The throughput value may comprise an uplink throughput value relating to throughput from the user equipment 120 to the second radio network node 140 and/or a downlink throughput value relating to throughput from the second radio network node 140 to the user equipment 120.

Action 211

Before the action 212, the central network node 1 1 1 may correlate the report to the obtained set of parameters by use of an identity of the user equipment 120, wherein the report and the obtained set of parameters are identifiable by means of the identity.

The handover may be caused by overload at the first radio network node 130, or the handover may be caused by mobility of the user equipment 120.

Action 212

The central network node 1 1 1 trains a set of models for prediction of the quality of service for the user equipment 120, wherein the training 205 comprises adapting each model of the set of models using the quality of service value of the report as a ground truth to be predicted given a respective subset of the set of parameters, which respective subset is taken as input to said each model, wherein each respective subset ranges from including one parameter of the set of parameters to all parameters of the set of parameters. In Figure 6, a schematic flowchart of exemplifying methods in the first radio network node 130 is shown. Again, the same reference numerals as above have been used to denote the same or similar features, in particular the same reference numerals have been used to denote the same or similar actions. In these embodiments, the network node 1 10 is the first radio network node 130. Accordingly, the first radio network node 130 performs a method for managing a handover of a user equipment 120 from a first radio network node 130 to a second radio network node 140.

As mentioned, the handover may be caused by overload at the first radio network node 130, or the handover may be caused by mobility of the user equipment 120. The first radio network node 130 may serve a plurality of user equipments 120, 121 , including the user equipment 120.

The following actions may be performed in any suitable order.

Action 201

The first radio network node 130 may detect that the handover of at least one of the plurality of user equipments is evaluable. This may mean that the first radio network node 130 may detect that the handover is to be evaluated, such as should be evaluated by performing the actions 208, 209 and 210.

The detecting 201 may be performed due to that a current quality of service value for at least one of the plurality of user equipments is below a threshold value for evaluating handover from the first radio network node 130.

The detecting 201 may be performed due to that a measurement report, received from at least one user equipment, indicates that the handover is evaluable.

The detecting 201 may be performed due to that a current load at the first radio network node 130 is above a threshold value for indicating load at the first radio network node 130 at which handover is evaluable.

The detecting 201 may be performed due to that a neighbour load at the second radio network node 140 is below a neighbour load threshold for indicating load at the second radio network node 140 at which handover is evaluable.

Actions 203, 204-207 may be performed before the handover.

Action 203

Before the handover, the first radio network node 130 obtains a set of

parameters, wherein the set of parameters comprises a radio measurement parameter relating to radio measurements, which radio measurements are performed by the user equipment 120 towards the second radio network node 140, wherein the user equipment 120 is served by the first radio network node 130.

The set of parameters may comprise one or more of:

a load parameter relating to load in the second radio network node 140, a capability information relating to capability of the user equipment 120, a channel quality indicator parameter relating to one or more channel quality indicators reported by the user equipment 120, a rank parameter relating to rank a connection to the user equipment 120, a signal strength parameter relating to received signal strength reported by the user equipment 120,

an amount parameter relating to an amount of user equipments served by the first radio network node 130,

a resource parameter relating to utilization of physical resource blocks, a bearer parameter relating to a bearer type of a bearer associated to the user equipment 120, and the like. Action 204

For each user equipment 120, 121 of the plurality of user equipments 120, 121 , the first radio network node 130 may select a model out of the set of models according to number of parameters and type of parameters in the obtained set of parameters. Action 205

For each user equipment 120, 121 of the plurality of user equipments 120, 121 , the first radio network node 130 may determine a predicted quality of service value for said each user equipment 120 while using the selected model. Action 206

The first radio network node 130 may select a user equipment 120 out of the plurality of user equipments 120, 121 for which the predicted quality of service value is above a threshold value for selecting one or more user equipments to be handed over from the first radio network node 130.

The threshold may indicate one or more of:

a current quality of service value for the user equipment 120, 121 ;

an average current quality of service value for the plurality of user equipments 120, 121 ;

an average predicted quality of service value for the plurality of user equipments 120, 121 ; and the like.

Action 207 When the set of parameters comprises the capability information relating to capability of the user equipment 120, the first radio network node 130 may keep the capability information available for the training of the set of models. Action 210

After the handover, the first radio network node 130 receives, from the second radio network node 140, a report including a quality of service value relating to quality of service for the user equipment 120 when served by the second radio network node 140. The quality of service value may comprise one or more of:

a throughput value relating to throughput of transmission between the second radio network node 140 and the user equipment 120;

a delay value relating to delay of transmission between the second radio network node 140 and the user equipment 120;

a packet loss rate relating to loss of packets in transmission between the second radio network node 140 and the user equipment 120;

a packet error rate relating to error in packets of transmission between the second radio network node 140 and the user equipment 120;

a reliability value relating to reliability of a connection carrying transmission between the second radio network node 140 and the user equipment 120; and the like.

The throughput value may comprise an uplink throughput value relating to throughput from the user equipment 120 to the second radio network node 140 and/or a downlink throughput value relating to throughput from the second radio network node 140 to the user equipment 120.

Action 211

Before action 212, the first radio network node 130 may correlate the report to the obtained set of parameters by use of an identity of the user equipment 120. The report and the obtained set of parameters may be identifiable by means of the identity.

Action 212

The first radio network node 130 trains a set of models for prediction of the quality of service for the user equipment 120, wherein the training 205 comprises adapting each model of the set of models using the quality of service value of the report as a ground truth to be predicted given a respective subset of the set of parameters, which respective subset is taken as input to said each model, wherein each respective subset ranges from including one parameter of the set of parameters to all parameters of the set of parameters.

With reference to Figure 7, a schematic block diagram of embodiments of the network node 1 10 of Figure 1 is shown. The network node 1 10 is thus configured to manage a handover of a user equipment 120 from the first radio network node 130 to the second radio network node 140.

As mentioned, the handover may be caused by overload at the first radio network node 130, or the handover may be caused by mobility of the user equipment 120.

The network node 1 10 may comprise a processing module 701 , such as a means, one or more hardware modules and/or one or more software modules for performing the methods described herein.

The network node 1 10 may further comprise a memory 702. The memory may comprise, such as contain or store, a computer program 703.

According to some embodiments herein, the processing module 701 comprises, e.g. 'is embodied in the form of or 'realized by', a processing circuit 704 as an exemplifying hardware module. In these embodiments, the memory 702 may comprise the computer program 703, comprising computer readable code units executable by the processing circuit 704, whereby the network node 1 10 is operative to perform the methods of Figure 2, 5 and/or 6.

In some other embodiments, the computer readable code units may cause the network node 1 10 to perform the method according to Figure 2, 5 and/or 6 when the computer readable code units are executed by the network node 1 10.

Figure 7 further illustrates a carrier 705, comprising the computer program 703 as described directly above. The carrier 705 may be one of an electronic signal, an optical signal, a radio signal, and a computer readable medium. In some embodiments, the processing module 701 comprises an Input/Output module 706, which may be exemplified by a receiving module and/or a sending module as described below when applicable. In further embodiments, the processing module 701 may comprise one or more of an obtaining module 710, a receiving module 720, a training module 730, a correlating module 740, a keeping module 750, a detecting module 760, a first selecting module 770, a determining module 780 and a second selecting module

790 as exemplifying hardware modules. In other examples, one or more of the aforementioned exemplifying hardware modules may be implemented as one or more software modules.

Therefore, according to the various embodiments described above, the network node 1 10 is operative to and/or the network node 1 10, the processing module 701 and/or the obtaining module 710 is configured to obtain, before the handover, a set of parameters, wherein the set of parameters comprises a radio measurement parameter relating to radio measurements, which radio measurements are performable by the user equipment 120 towards the second radio network node 140, wherein the user equipment 120 is servable by the first radio network node 130.

The set of parameters may comprise one or more of:

a load parameter relating to load in the second radio network node 140, a capability information relating to capability of the user equipment 120, a channel quality indicator parameter relating to one or more channel quality indicators reported by the user equipment 120,

a rank parameter relating to rank a connection to the user equipment 120, a signal strength parameter relating to received signal strength reported by the user equipment 120,

an amount parameter relating to an amount of user equipments served by the first radio network node 130,

a resource parameter relating to utilization of physical resource blocks, a bearer parameter relating to a bearer type of a bearer associated to the user equipment 120, and the like.

When the set of parameters comprises the capability information relating to capability of the user equipment 120, and the network node 1 10 is the first radio network node 130, the network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the keeping module 750 may be configured to keep the capability information available for the training of the set of models. Furthermore, the network node 1 10 is operative to and/or the network node 1 10, the processing module 701 and/or the receiving module 720 is configured to, after the handover, receive, from the second radio network node 140, a report including a quality of service value relating to quality of service for the user equipment 120 when served by the second radio network node 140.

The quality of service value may comprise one or more of:

a throughput value relating to throughput of transmission between the second radio network node 140 and the user equipment 120;

a delay value relating to delay of transmission between the second radio network node 140 and the user equipment 120;

a packet loss rate relating to loss of packets in transmission between the second radio network node 140 and the user equipment 120;

a packet error rate relating to error in packets of transmission between the second radio network node 140 and the user equipment 120;

a reliability value relating to reliability of a connection carrying transmission between the second radio network node 140 and the user equipment 120; and the like.

Moreover, the network node 1 10 is operative to and/or the network node 1 10, the processing module 701 and/or the training module 730 is configured to train a set of models for prediction of the quality of service for the user equipment 120, wherein the network node 1 10 is configured to train the set of models by adapting each model of the set of models using the quality of service value of the report as a ground truth to be predicted given a respective subset of the set of parameters, which respective subset is taken as input to said each model, wherein each respective subset ranges from including one parameter of the set of parameters to all parameters of the set of parameters.

The network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the correlating module 740 may be configured to correlate the report to the obtained set of parameters by use of an identity of the user equipment 120, wherein the report and the obtained set of parameters are identifiable by means of the identity. The network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the detecting module 760 may be configured to detect that the handover of at least one of the plurality of user equipments is evaluable.

The network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the detecting module 760 may be configured to detect that the handover is evaluable when a current quality of service value for at least one of the plurality of user equipments is below a threshold value for evaluating handover from the first radio network node 130.

The network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the detecting module 760 may be configured to detect that the handover is evaluable when a measurement report, received from at least one user equipment, indicates that the handover is evaluable.

The network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the detecting module 760 may be configured to detect that the handover is evaluable when a current load at the first radio network node 130 is above a threshold value for indicating load at the first radio network node 130 at which handover is evaluable.

The threshold may indicate one or more of:

a current quality of service value for the user equipment 120, 121 ;

an average current quality of service value for the plurality of user equipments 120, 121 ;

an average predicted quality of service value for the plurality of user equipments 120, 121 and the like. The network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the detecting module 760 may be configured to detect that the handover is evaluable when a neighbour load at the second radio network node 140 is below a neighbour load threshold for indicating load at the second radio network node 140 at which handover is evaluable.

The first radio network node 130 may be configured to serve a plurality of user equipments 120, 121 , including the user equipment 120. In embodiments when the network node 1 10 is the first radio network node 130, the network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the first selecting module 770 may be configured to, before the handover, for each user equipment 120, 121 of the plurality of user equipments 120, 121 , select a model out of the set of models according to number of parameters and type of parameters in the obtained set of parameters.

Furthermore, in these embodiments, the network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the determining module 780 may be configured to determine a predicted quality of service value for said each user equipment 120 while using the selected model.

Moreover, in these embodiments, the network node 1 10 may be operative to and/or the network node 1 10, the processing module 701 and/or the second selecting module 790 may be configured to select a user equipment 120 out of the plurality of user equipments 120, 121 for which the predicted quality of service value is above a threshold value for selecting one or more user equipments to be handed over from the first radio network node 130.

As used herein, the term "processing module" may include one or more hardware modules, one or more software modules or a combination thereof. Any such module, be it a hardware, software or a combined hardware-software module, may be a determining means, estimating means, capturing means, associating means, comparing means, identification means, selecting means, receiving means, sending means or the like as disclosed herein. As an example, the expression "means" may be a module

corresponding to the modules listed above in conjunction with the Figures.

As used herein, the term "processing circuit" may refer to a processing unit, a processor, an Application Specific integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA) or the like. The processing circuit or the like may comprise one or more processor kernels.

As used herein, the expression "configured to" may mean that a processing circuit is configured to, or adapted to, by means of software configuration and/or hardware configuration, perform one or more of the actions described herein.

As used herein, the term "memory" may refer to a hard disk, a magnetic storage medium, a portable computer diskette or disc, flash memory, random access memory (RAM) or the like. Furthermore, the term "memory" may refer to an internal register memory of a processor or the like.

As used herein, the term "computer readable medium" may be a Universal Serial Bus (USB) memory, a DVD-disc, a Blu-ray disc, a software module that is received as a stream of data, a Flash memory, a hard drive, a memory card, such as a MemoryStick, a Multimedia Card (MMC), etc.

As used herein, the term "computer readable code units" may be text of a computer program, parts of or an entire binary file representing a computer program in a compiled format or anything there between.

As used herein, the term "radio resource" may refer to a certain coding of a signal and/or a time frame and/or a frequency range in which the signal is transmitted. In some examples, a resource may refer to one or more Physical Resource Blocks (PRB) which is used when transmitting the signal. In more detail, a PRB may be in the form of Orthogonal Frequency Division Multiplexing (OFDM) PHY resource blocks (PRB). The term "physical resource block" is known from 3GPP terminology relating to e.g. Long Term Evolution Systems.

As used herein, the terms "number" and/or "value" may be any kind of digit, such as binary, real, imaginary or rational number or the like. Moreover, "number" and/or "value" may be one or more characters, such as a letter or a string of letters. "Number" and/or "value" may also be represented by a bit string.

As used herein, the expression "in some embodiments" has been used to indicate that the features of the embodiment described may be combined with any other embodiment disclosed herein. Even though embodiments of the various aspects have been described, many different alterations, modifications and the like thereof will become apparent for those skilled in the art. The described embodiments are therefore not intended to limit the scope of the present disclosure.