Quality of experience aware methods for network load balancing

Abstract

Methods and systems for performing load balancing in a wireless communication network are provided. The method includes receiving quality of experience, QoE, data from a plurality of electronic devices connected to the wireless communication network. The QoE data is indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide a service using the wireless communication network. The method further includes inputting the received QoE data into a load balancing algorithm. The load balancing algorithm is configured to output a load balancing action based on the QoE data where the output load balancing action, when implemented, adjusts a balance of network traffic across a plurality of nodes in the wireless communication network to improve an overall QoE across the plurality of electronic devices using the wireless communication network. The method additionally includes performing the load balancing action output by the load balancing algorithm.

Description

QUALITY OF EXPERIENCE AWARE METHODS FOR NETWORK LOAD BALANCING

[0001] The present disclosure relates to methods and apparatus for load balancing in a wireless communication network. The method and apparatus may find particular application in generating and taking actions to balance network traffic in a wireless communication network.BACKGROUND

[0002] A wireless communication network, such as a mobile telecommunications network, provides a range of services to devices connected to the wireless communication network. For example, an electronic device connected to the network may be provided with voice, messaging and / or data services. Users are increasingly using their devices for high bit-rate real time services over the network, such as video streaming services or extended reality (XR) services. For these types of services, even intermittent quality degradation can have a large impact on the quality of a user's experience with the service. Many of these streaming services are a significant part of the commercial traffic growth rate, so network operators are focused on improving the end users' experiences when accessing these services.

[0003] It is in this context that the subject matter contained in the present application has been devised.SUMMARY

[0004] It has been realised that current load balancing techniques can be improved by considering quality of experience (QoE) data obtained from electronic devices using the wireless communication network. Providing a QoE aware load balancing technique can achieve efficient load or traffic balancing in the wireless communication network whilst also optimising or improving a user QoE. The QoE data can be used to refine the load balancing technique by enabling a load balancing action for a particular situation to be more optimally selected. This helps to improve or guarantee an overall QoE experienced by the plurality of users of the wireless communication network.

[0005] According to a first aspect of the present disclosure there is provided a method for performing load balancing in a wireless communication network. The method includes receiving quality of experience, QoE, data from a plurality of electronic devices connected to the wireless communication network. The QoE data is indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide aservice using the wireless communication network. The method further includes inputting the received QoE data into a load balancing algorithm. The load balancing algorithm is configured to output a load balancing action based on the QoE data where the output load balancing action, when implemented, adjusts (or improves) a balance of network traffic across a plurality of nodes in the wireless communication network to improve an overall QoE across the plurality of electronic devices using the wireless communication network. The method additionally includes performing the load balancing action output by the load balancing algorithm.

[0006] Advantageously, this helps to improve an overall QoE across the wireless communication network when load balancing is being performed.

[0007] Load balancing distributes a traffic load more evenly across cells or groups of cells in a network to improve network performance. Current load balancing decisions typically rely on the current or past state cell load status, which is insufficient for ensuring a quality of a user experience. This is because the traffic load and a resource status of the wireless communication network can change rapidly, especially in situations with high-mobility and a large number of connections. This can lead to ping-pong handover between different cells, cell overload and degradation of user service quality. It is presently difficult to guarantee an overall network and service performance when performing load balancing. For example, a user equipment (UE) may be offloaded from a congested cell to a target cell but, if the UEs with time-varying traffic requirements are offloaded to the target cell, the target cell may itself become overloaded with the newly arrived heavy traffic. It is difficult to predict whether a service performance of a network will meet desired performance targets after the offloading action has been performed.

[0008] Load balancing methods have been proposed which utilise Artificial Intelligence (Al) algorithms to generate load balancing actions. For example, machine learning (ML) techniques may be used to train a model and the trained model may be used to generate actions and / or predictions for performing load balancing in a wireless communication network. However, proposals for the use of Al and ML techniques for load balancing have to date only focused on the balancing of network traffic and without consideration of a user experience which might result from load balancing actions.

[0009] The use of QoE data in the present application advantageously enables predictive load balancing methods to be fine-tuned so that a load balancing action which is more optimal for both a distribution of network traffic and an overall QoE across the plurality of electronic devices can be performed. By considering the QoE data from a plurality of electronic devices using aservice over the network, a load balancing action which is detrimental to the overall QoE of the plurality of electronic devices may not be selected, even if the distribution of network traffic across the network would be greatly improved by said load balancing action. Moreover, a load balancing action which is detrimental to a QoE for one of the electronic devices from among the plurality of electronic devices may not be selected.

[0010] For example, a first load balancing action involving one or more electronic devices being offloaded from a first cell to a second cell may not be selected if it is determined that a QoE of one or more of the electronic devices in the network will be detrimentally affected by the offloading, even if the first load balancing action would improve the distribution of network traffic across the network. Instead, a second load balancing action different to the first load balancing action may be selected even if the distribution of network traffic across the network is not as improved as the distribution that the first load balancing action would achieve, if the second load balancing action is deemed to have a better effect on the overall QoE of the plurality of electronic devices using a service over the network.

[0011] Optionally, the load balancing algorithm may be configured to receive a plurality of potential load balancing actions, where each of the potential load balancing actions is an action which may be performed in the communication network to adjust the balance of network traffic across a plurality of nodes in the communication network. The load balancing algorithm may be further configured to select a load balancing action from the plurality of potential load balancing actions in dependence on the received QoE data where the selected load balancing action, when implemented, improves an overall QoE across the plurality of electronic devices using the wireless communication network. The load balancing algorithm may be further configured to output the selected load balancing action.

[0012] In some examples, the load balancing algorithm does not generate the plurality of potential load balancing actions itself. The generation of the plurality of potential load balancing actions may be performed by a different entity to the entity running the load balancing algorithm so computational resources may be more efficiently used. In some examples, the plurality of potential load balancing actions may be generated using Al and / or ML techniques. For example, an Al and / or ML model may be trained (e.g., using historical data collected in the wireless communication network) so as to generate a trained model configured to generate a plurality of potential load balancing actions for balancing a load of network traffic across the mobile telecommunications network.

[0013] The load balancing algorithm may select a load balancing algorithm from the plurality of potential load balancing actions based on an expected change in a QoE at the plurality of electronic devices. The method may further comprise determining an expected change in a QoE for a plurality of electronic devices for each of the plurality of potential load balancing actions. An expected change in a QoE for an electronic device which results from a potential load balancing action may be determined using a predictive model configured to predict a change in a QoE which results from a potential load balancing action. The predictive model may be configured to predict a change in QoE which results from a potential load balancing action in dependence on the received QoE data. For example, the predictive model may use a current state of QoE (as reflected by the QoE data) to predict a change in QoE for a particular load balancing action. Such a predictive model may be configured through training using historically measured data. For example, a predictive model for determining a change in QoE for a load balancing action may be trained using historically collected training data indicative of measured QoE data for configurations of devices in the network (and / or for past load balancing actions). The predictive model may be trained and / or implemented using Al and / or ML techniques and may therefore be referred to as an Al and / or ML model.

[0014] In other examples, the predictive model may be rules based. A rules based predictive model may use the QoE data and one or more pre-configured rules to determine an expected change in QoE which results from a given potential load balancing algorithm. To provide a simple illustrative example, a potential load balancing algorithm may comprise a handover of an electronic device from a first cell, carrier and / or radio access technology (RAT) to a second cell, carrier and / or RAT. The QoE data may include data indicative of a QoE currently measured by one or more devices using the first cell, carrier and / or RAT and data indicative of a QoE currently measured by one or more devices using the second cell, carrier and / or RAT. If the QoE measured by the one or more devices using the second cell, carrier and / or RAT is generally better than the QoE measured by the one or more devices using the first cell, carrier and / or RAT then a predictive model may determine that the QoE at an electronic device is expected to be improved by handover from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT. In such a scenario, the load balancing algorithm may select the potential load balancing algorithm comprising a handover of an electronic device from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT based on the expected improvement in QoE. Conversely if the QoE measured by the one or more devices using the second cell, carrier and / or RAT is generallyworse than the QoE measured by the one or more devices using the first cell, carrier and / or RAT then the predictive model may determine that the QoE at an electronic device is expected to be reduced by handover from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT. In such a scenario, the load balancing algorithm may not select the potential load balancing algorithm comprising a handover of an electronic device from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT based on the expected worsening in QoE.

[0015] Optionally, the load balancing algorithm may be further configured to receive data indicative of current traffic load distribution in the wireless communication network and generate the plurality of potential load balancing actions in dependence on the data indicative of current traffic load distribution in the wireless communication network. Further optionally, the method may also include receiving data indicative of current traffic load distribution in the wireless communication network, and inputting the received data indicative of current traffic load distribution in the wireless communication network into the load balancing algorithm to generate the plurality of potential load balancing actions.

[0016] In other words, the method may further comprise receiving data indicative of current load distribution in the wireless communication network; and further inputting the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data and the data indicative of current load distribution in the wireless communication network, such that the output load balancing action, when implemented, improves a balance of network traffic across the plurality of nodes in the wireless communication network and the overall QoE across the plurality of electronic devices using the wireless communication network.

[0017] In examples, in which the load balancing algorithm is configured to generate the plurality of potential load balancing actions, at least part of the load balancing algorithm (e.g., a first stage of the load balancing algorithm which is configured to generate the plurality of potential load balancing actions in dependence on the data indicative of current traffic load distribution) may comprise an Al and / or machine learning algorithm. For example, at least part of the load balancing algorithm may comprise a model configured to generate predictions and / or optimal load balancing actions, where the model has been configured through training. The training used to configure the model may utilise any suitable Al and / or machine learning techniques. The training may be based on historically collected data. The historically collecteddata may, for example, comprise measurements made by a plurality of electronic devices connected to the wireless communication network and / or measurements indicative of traffic load distribution in the network at a plurality of time points in the past.

[0018] Optionally, the data indicative of current load distribution in the wireless communication network may comprise one or more measurements indicative of a quality of a network signal received by an electronic device from among the plurality of electronic devices.

[0019] The data indicative of current load distribution in the wireless communication network may, for example, comprise measurements made by electronic devices connected to the wireless communication network. The measurements may be indicative of a location and / or movement of an electronic device. The measurements may, for example, comprise Minimization of Drive Test (MDT) measurements. The measurements may, for example, comprise measurements indicative of one or more of a power, quality and / or a signal to noise ratio of radio signal received at an electronic device. For example, the measurements may comprise one or more of a Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ) and / or Signal to Interference & Noise Ratio (SINR). Measurements indicative of received radio signal (e.g. RSRP, RSRQ and / or SINR) may comprise measurements corresponding to a serving cell and / or may comprise measurements corresponding to one or more neighbouring cells.

[0020] The data indicative of current load distribution in the wireless communication network may, for example, comprise measurements indicative of a traffic load in the wireless communication network. The measurements may be made by one or more base stations in the wireless communication network. The measurements may, for example, comprise one or more of a total transmit power, a total received power, interference in a cell throughput (e.g., in downlink and / or uplink), an increase in blocking and / or a handover failure rate.

[0021] Optionally, the load balancing algorithm uses the input QoE data to select the output load balancing action, from the plurality of potential load balancing actions, based on an expected change in the overall QoE of the plurality of electronic devices in the wireless communication network which results from each of the plurality of potential load balancing actions.

[0022] As was explained above, an expected change in the overall QoE of the plurality of electronic devices in the wireless communication network which results from each of the plurality of potential load balancing actions may be determined using a predictive model. The predictive model may, for example, comprise a model configured through training and usinghistorically collected training data (where the training may comprise Al and / ML techniques). In some examples, the predictive model may be rules based.

[0023] Optionally, the load balancing algorithm is further configured to receive data indicative of a current load distribution in the wireless communication network and the QoE data. The load balancing algorithm may determine, in dependence on both the data indicative of a current load distribution and the QoE data, the load balancing action which when implemented, improves both a balance of network traffic across the plurality of nodes in the wireless communication network and an overall QoE across the plurality of electronic devices using the wireless communication network. Further optionally, the method may further comprise receiving data indicative of current load distribution in the wireless communication network. The method may further comprise inputting the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm such that the load balancing algorithm outputs the load balancing action which when implemented, improves both a balance of network traffic across the plurality of nodes in the wireless communication network.

[0024] Optionally, the load balancing algorithm jointly optimizes the balance of traffic operating across the plurality of nodes in the communication system and the overall QoE of the plurality of electronic devices, and the output load balancing action represents a result of the optimization.

[0025] In other words, the method may further comprise receiving data indicative of current load distribution in the wireless communication network, and further inputting the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data and the data indicative of current load distribution in the wireless communication network, such that the output load balancing action, when implemented, jointly optimizes the balance of network traffic across the plurality of nodes in the wireless communication system and the overall QoE of the plurality of electronic devices.

[0026] Advantageously, this provides a better compromise for improving both the distribution of the load across the network and the QoE received by users of the electronic devices. The output load balancing action may not be an action which fully optimizes the load balancing itself if such an action would likely have a detrimental impact or only a limited benefit on the QoE received by users. Similarly, the output load balancing action may not be an action which fully optimizes the QoE received by users if such an action would likely have a detrimental impact or only alimited improvement on balancing a load operating across the network. Instead, the load balancing algorithm outputs a load balancing action which provides the greatest synergistic benefit across the wireless communication network. That is, the output load balancing action may make a trade off to sacrifice some of the efficiency in the load balancing of the traffic load and / or some of the overall QoE experienced by the users of the electronic devices using the wireless communication system in order to reach a solution which has the greatest benefit on the wireless communication system as a whole and accounting for user QoE.

[0027] The load balancing algorithm may comprise an Al and / or ML algorithm. For example, the load balancing algorithm may be configured through training using Al and / or ML techniques. The training may be based on training data comprising historically collected data. The skilled person would be aware that there are many potential Al and / or machine learning algorithms which could be used as the load balancing algorithm. Some examples of suitable algorithms include heuristic, meta-heuristic, hybrid algorithms, and Content-aware Machine Learning based Load Balancing. However, this is not an exhaustive list and the invention is not intended to be limited to the listed algorithms.

[0028] Optionally, the method further comprises receiving feedback information from the plurality of electronic devices after the load balancing action is performed.

[0029] Advantageously, this enables an analysis to be performed to determine whether the output load balancing action has achieved the desired effect. The result of the analysis may be used to perform further load balancing actions across the wireless communication network. The result of the analysis may be used as input data for training the load balancing algorithm through artificial intelligence and / or machine learning. In some examples, the feedback information may be provided to a predictive model for determining an expected change in QoE which results from a load balancing action. The feedback information may, for example, be used to evaluate and / or further train the predictive model.

[0030] Optionally, the load balancing action comprises one or more of: retaining an existing connection between an electronic device from the plurality of electronic devices and a connected first node of the wireless communication network; performing a cell handover to transfer an electronic device attached to a first cell of the wireless communication network to a second cell of the wireless communication network; and performing an inter-radio access technology, RAT, handover on an electronic device to transfer the electronic device from a first RAT to a secondRAT.

[0031] Optionally, the load balancing action comprises one or more of: performing idle mode load balancing to adjust a cell reselection parameter for electronic devices in an idle mode; modifying a power of a pilot signal transmitted in the wireless communication network; adjusting an angle of an antenna of a node in the wireless communication network; or modifying a handover parameter condition for cell handover of electronic devices connected to the wireless communication network.

[0032] Optionally, the QoE data comprises one or more of: an average throughput, an initial playout delay, a buffer level or a playout delay for a media start-up, as measured whilst the electronic device provides the service using the communication network.

[0033] Optionally, the load balancing algorithm is configured through training to select the output load balancing action. The training of the load balancing algorithm may use historical data received from the plurality of electronic devices. The historical data may comprise one or more of data relating to previous traffic load distribution in the wireless communication network, data relating to previous outcomes of load balancing measures performed by the wireless communication network, and data relating to previous measurements made by one or more of the plurality of electronic devices.

[0034] Advantageously, this enables the load balancing algorithm to improve the selecting of the output load balancing action. Training the load balancing algorithm using historical data enables the load balancing algorithm to better predict a likely outcome of a particular load balancing action. This in turn enables the algorithm to better determine a suitable load balancing action to be applied to the network according to a current situation of the network.

[0035] Optionally, the method may comprise training the load balancing algorithm to select the output load balancing action. The training may be based on received historical data.

[0036] Training of the load balancing algorithm may comprise using Al and / or ML learning algorithms. The skilled person would be aware that there are many potential Al and / or machine learning algorithms which could be used to train the load balancing algorithm. Some examples of suitable algorithms include heuristic, meta-heuristic, hybrid algorithms, and Content-aware Machine Learning based Load Balancing. However, this is not an exhaustive list and the invention is not intended to be limited to the listed algorithms.

[0037] According to a second aspect of the present disclosure, there is provided a system for performing load balancing in a wireless communication network. The system is configured to receive quality of experience, QoE, data from a plurality of electronic devices connected to thewireless communication network. The QoE data is indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide a service using the wireless communication network. The system is further configured to input the received QoE data into a load balancing algorithm, where the load balancing algorithm is configured to output a load balancing action based on the QoE data. The output load balancing action, when implemented, adjusts a balance of network traffic across a plurality of nodes in the wireless communication network to improve an overall QoE across the plurality of electronic devices using the wireless communication network. The system is further configured to perform the load balancing action output by the load balancing algorithm.

[0038] According to a third aspect of the present disclosure there is provided a system for performing load balancing in a wireless communication network. The system comprises at least one processing unit, and an input / output interface. The processing unit is configured to receive quality of experience, QoE, data from a plurality of electronic devices connected to the wireless communication network via the input / output interface, the QoE data indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide a service using the wireless communication network. The processor is further configured to input the received QoE data into a load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data wherein the output load balancing action, when implemented, adjusts a balance of network traffic across a plurality of nodes in the wireless communication network to improve an overall QoE across the plurality of electronic devices using the wireless communication network. The processing unit is further configured to output instructions via the input / output interface to perform or implement in the wireless communication network the load balancing action output by the load balancing algorithm.

[0039] Optionally, the system is a wireless communication network, and may further comprise the plurality of electronic devices and the plurality of nodes.

[0040] According to a fourth aspect of the present disclosure there is provided a computer- readable medium comprising instructions which, when executed by a computing apparatus, cause the computing apparatus to carry out a method according to the first aspect.

[0041] Within the scope of this application it is expressly intended that the various aspects, embodiments, examples and alternatives set out in the preceding paragraphs, in the claims and / or in the following description and drawings, and in particular the individual features thereof, maybe taken independently or in any combination. That is, all examples and / or features of any example can be combined in any way and / or combination, unless such features are incompatible. The applicant reserves the right to change any originally filed claim or file any new claim accordingly, including the right to amend any originally filed claim to depend from and / or incorporate any feature of any other claim although not originally claimed in that manner.The following numbered clauses show further illustrative examples only:1. A method for performing load balancing in a wireless communication network, the method comprising: receiving quality of experience, QoE, data from a plurality of electronic devices connected to the wireless communication network, the QoE data indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide a service using the wireless communication network; inputting the received QoE data into a load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data wherein the output load balancing action, when implemented, improves a balance of network traffic across a plurality of nodes in the wireless communication network and an overall QoE across the plurality of electronic devices using the wireless communication network; and performing the load balancing action output by the load balancing algorithm.2. The method of clause 1, wherein the load balancing algorithm is configured to: receive a plurality of potential load balancing actions, wherein each of the potential load balancing actions is an action which may be performed in the communication network to balance the load across a plurality of nodes in the communication network; select from the plurality of potential load balancing actions and in dependence on the received QoE data, a load balancing action which, when implemented, improves an overall QoE across the plurality of electronic devices using the wireless communication network; and output the selected load balancing action.3. The method of clause 2, wherein the load balancing algorithm is further configured to receive data indicative of current traffic load distribution in the wireless communication network and generate the plurality of potential load balancing actions in dependence on the data indicative of current traffic load distribution in the wireless communication network, and wherein the method further comprises:receiving data indicative of current traffic load distribution in the wireless communication network; and inputting the received data indicative of current traffic load distribution in the wireless communication network into the load balancing algorithm to generate the plurality of potential load balancing actions.4. The method of clause 3, wherein the data indicative of current load distribution in the wireless communication network comprises one or more measurements indicative of a quality of a network signal received by an electronic device from among the plurality of electronic devices.5. The method of any of clause 2 to clause 4, wherein the load balancing algorithm uses the input QoE data to select the output load balancing action, from the plurality of potential load balancing actions, based on an expected change in the overall QoE of the plurality of electronic devices in the wireless communication network, which results from each of the plurality of potential load balancing actions.6. The method of clause 1, wherein the load balancing algorithm is further configured to receive data indicative of a current load distribution in the wireless communication network and the QoE data and determine, in dependence on both the data indicative of a current load distribution and the QoE data, the load balancing action which when implemented, improves both a balance of network traffic across the plurality of nodes in the wireless communication network and an overall QoE across the plurality of electronic devices using the wireless communication network, wherein the method further comprises: receiving data indicative of current load distribution in the wireless communication network; and; further inputting the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm such that the load balancing algorithm outputs the load balancing action which when implemented, improves both a balance of network traffic across the plurality of nodes in the wireless communication network.7. The method of any of the preceding clauses, wherein the load balancing algorithm concurrently optimizes both the balance of traffic operating across the plurality of nodes in the communication system and the overall QoE of the plurality of electronic devices, and the output load balancing action represents a result of the optimization.8. The method of any of the preceding clauses, further comprising receiving feedback information from the plurality of electronic devices after the load balancing action is performed.9. The method of any of the preceding clauses, wherein the load balancing action comprises one or more of: retaining an existing connection between an electronic device from the plurality of electronic devices and a connected first node of the wireless communication network, performing a cell handover to transfer an electronic device attached to a first cell of the wireless communication network to a second cell of the wireless communication network; or performing an inter-radio access technology, RAT, handover of an electronic device to transfer the electronic device from a first RAT to a second RAT.10. The method of any of the preceding clauses, wherein the load balancing action comprises one or more of: performing idle mode load balancing to adjust a cell reselection parameter for electronic devices in an idle mode; modifying a power of a pilot signal transmitted in the wireless communication network; adjusting an angle of an antenna of a node in the wireless communication network; or modifying a handover parameter condition for cell handover of electronic devices connected to the wireless communication network.11. The method of any of the preceding clauses, wherein the QoE data comprises one or more of: an average throughput, an initial playout delay, a buffer level or a playout delay for a media start-up, as measured whilst the electronic device provides the service using the communication network.12. The method of any of the preceding clauses, wherein the load balancing algorithm is configured through training to select the output load balancing action; wherein the training of the load balancing algorithm uses historical data received from a plurality of electronic devices; and wherein the historical data comprises one or more of data relating to previous traffic load distribution in the wireless communication network, data relating to previous outcomes of loadbalancing measures performed by the wireless communication network, and data relating to previous measurements made by one or more of the plurality of electronic devices.13. The method of clause 12, further comprising training, using the received historical data, the load balancing algorithm to select the output load balancing action.14. A system for a wireless communication network, the system configured to: receive quality of experience, QoE, data from a plurality of electronic devices connected to the wireless communication network, the QoE data indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide a service using the wireless communication network; input the received QoE data into a load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data wherein the output load balancing action, when implemented, improves a balance of network traffic across a plurality of nodes in the wireless communication network and an overall QoE across the plurality of electronic devices using the wireless communication network; and perform the load balancing action output by the load balancing algorithm.15. The system of clause 14, wherein the load balancing algorithm is configured to: receive a plurality of potential load balancing actions, wherein each of the potential load balancing actions is an action which may be performed in the communication network to balance the load across a plurality of nodes in the communication network; select from the plurality of potential load balancing actions and in dependence on the received QoE data, a load balancing action which, when implemented, improves an overall QoE across the plurality of electronic devices using the wireless communication network; and output the selected load balancing action.16. The system of clause 15, wherein, the load balancing algorithm is further configured to receive data indicative of current traffic load distribution in the communication network and generate the plurality of potential load balancing actions in dependence on the data indicative of current traffic load distribution in the communication network, and wherein the system is further configured to: receive data indicative of current traffic load distribution in the wireless communication network; andinput the received data indicative of current traffic load distribution in the wireless communication network into the load balancing algorithm to generate the plurality of potential load balancing actions.17. The system of clause 16, wherein the data indicative of current load distribution in the wireless communication network comprises one or more measurements indicative of a quality of a network signal received by an electronic device from among the plurality of electronic devices.18. The system of any one of clause 15 to clause 17, wherein the load balancing algorithm uses the input QoE data to select the output load balancing action, from the plurality of potential load balancing actions, based on an expected change in the overall QoE of the plurality of electronic devices in the wireless communication network, which results from each of the plurality of potential load balancing actions.19. The system of clause 14, wherein the load balancing algorithm is further configured to receive data indicative of a current load distribution in the wireless communication network and the QoE data and determine, in dependence on both the data indicative of a current load distribution and the QoE data, the load balancing action which when implemented, improves both a balance of network traffic across the plurality of nodes in the wireless communication network and an overall QoE across the plurality of electronic devices using the wireless communication network, wherein the system is further configured to: receive data indicative of current load distribution in the wireless communication network; and; further input the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm such that the load balancing algorithm outputs the load balancing action which when implemented, improves both a balance of network traffic across the plurality of nodes in the wireless communication network.20. The system of any one of clause 14 to clause 19, wherein the load balancing algorithm concurrently optimizes both the balance of traffic operating across the plurality of nodes in the communication system and the overall QoE of the plurality of electronic devices, and the output load balancing action represents a result of the optimization.21. The system of any one of clause 14 to clause 20, wherein the system is further configured to:receive feedback information from the plurality of electronic devices after the load balancing action is performed.22. The system of any one of clause 14 to clause 21, wherein the load balancing action comprises one or more of: retaining an existing connection between an electronic device from the plurality of electronic devices and a connected first node of the wireless communication network, performing a cell handover to transfer an electronic device attached to a first cell of the wireless communication network to a second cell of the wireless communication network; or performing an inter-radio access technology, RAT, handover on an electronic device to transfer the electronic device from a first RAT to a second RAT.23. The system of any one of clause 14 to clause 22, wherein the load balancing action comprises one or more of: performing idle mode load balancing to adjust a cell reselection parameter for electronic devices in an idle mode; modifying a power of a pilot signal transmitted in the wireless communication network; adjusting an angle of an antenna of a node in the wireless communication network; or modifying a handover parameter condition for cell handover of electronic devices connected to the wireless communication network.24. The system of any one of clause 14 to clause 23, wherein the QoE data comprises one or more of: an average throughput, an initial playout delay, a buffer level or a playout delay for a media start-up, as measured whilst the electronic device provides the service using the communication network.25. The system of any one of clause 14 to clause 24, wherein the load balancing algorithm is configured through training to select the output load balancing action; wherein the training of the load balancing algorithm uses historical data received from the plurality of electronic devices; and wherein the historical data comprises one or more of data relating to previous traffic load distribution in the wireless communication network, data relating to previous outcomes of load balancing measures performed by the wireless communication network, and data relating to previous measurements made by one or more of the plurality of electronic devices.26. The system of clause 25, wherein the system is further configured to train, using the received historical data, the load balancing algorithm to select the output load balancing action.27. A computer-readable medium comprising instructions which, when executed by a computing apparatus, cause the computing apparatus to carry out the method of any of clauses 1 to 13.BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

[0042] One or more embodiments of the invention are shown schematically, by way of example only, in the accompanying drawings, in which:- FIG. 1 is a schematic illustration of a network architecture for a wireless communication network;- FIG. 2 is a flow chart of an example method for performing load balancing in a wireless communication network;- FIG. 3 is a signal flow diagram illustrating an example method for performing load balancing in a wireless communication network; and- FIG. 4 is a schematic illustration of an electronic device which may be used to implement all or part of any method disclosed herein.DETAILED DESCRIPTION

[0043] Hereinafter, embodiments of the disclosure are described with reference to the accompanying drawings. However, it should be appreciated that the disclosure is not limited to the embodiments, and all changes and / or equivalents or replacements thereto also belong to the scope of the disclosure. The same or similar reference denotations may be used to refer to the same or similar elements throughout the specification and the drawings.

[0044] As used herein, the terms “have,” “may have,” “include,” or “may include” a feature (e.g., a number, function, operation, or a component such as a part) indicate the existence of the feature and do not exclude the existence of other features.

[0045] As used herein, the terms “A or B,” “at least one of A and / or B,” or “one or more of A and / or B” may include all possible combinations of A and B. For example, “A or B,” “at leastone of A and B,” “at least one of A or B” may indicate all of (1) including at least one A, (2) including at least one B, or (3) including at least one A and at least one B.

[0046] As used herein, the terms “first” and “second” may modify various components regardless of importance and do not limit the components. These terms are only used to distinguish one component from another. For example, a first user device and a second user device may indicate different user devices from each other regardless of the order or importance of the devices. For example, a first component may be denoted a second component, and vice versa without departing from the scope of the disclosure.

[0047] It will be understood that when an element (e.g., a first element) is referred to as being (operatively or communicatively) “coupled with / to,” or “connected with / to” another element (e.g., a second element), it can be coupled or connected with / to the other element directly or via a third element. In contrast, it will be understood that when an element (e.g., a first element) is referred to as being “directly coupled with / to” or “directly connected with / to” another element (e.g., a second element), no other element (e.g., a third element) intervenes between the element and the other element.

[0048] As used herein, the terms “configured (or set) to” may be interchangeably used with the terms “suitable for,” “having the capacity to,” “designed to,” “adapted to,” “made to,” or “capable of’ depending on circumstances. The term “configured (or set) to” does not essentially mean “specifically designed in hardware to.” Rather, the term “configured to” may mean that a device can perform an operation together with another device or parts.

[0049] For example, the term “processor configured (or set) to perform A, B, and C” may mean a generic-purpose processor (e.g., a CPU or application processor) that may perform the operations by executing one or more software programs stored in a memory device or a dedicated processor (e.g., an embedded processor) for performing the operations.

[0050] The terms as used herein are provided merely to describe some embodiments thereof, but not to limit the scope of other embodiments of the disclosure. It is to be understood that the singular forms “a,” “'an,” and “the” include plural references unless the context clearly dictates otherwise. All terms including technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which the embodiments of the disclosure belong. It will be further understood that terms, such as those defined in commonly used dictionaries, should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal senseunless expressly so defined herein. In some cases, the terms defined herein may be interpreted to exclude embodiments of the disclosure.

[0051] As used herein, "base station" refers to a network node comprised in a wireless cellular network (which can also be referred to as a Radio Access Network (RAN)) for providing, via one or more antennae, cells of radio coverage in which a wireless radio connection to user equipment may be established. The Base Station may comprise a base station (BS), radio base station, base transceiver station (BTS), base station controller (BSC), radio network controller (RNC), evolved Node B (eNB or eNodeB), Node B, next generation eNodeB (ng-eNodeB), gNodeB (gNB), multistandard radio (MSR) radio node such as MSR BS, multi-cell / multicast coordination entity (MCE), relay node, donor node controlling relay, radio access point (AP), and suchlike. Each base station provides one or more cells of a radio communications system within which wireless radio connections can be established between the base stations and user equipment. Each base station may support voice and data communication by one or more Radio Access Technologies (RATs), such as UMTS, LTE and 5G NR. The cells provided by each base station are roughly designated as land areas of coverage arranged as sectors emanating from radio antennas arranged at the base station. User equipment located in the cells in which coverage is provided can gain network access by establishing a radio connection with the base station providing the cell. The Radio Access Network can provide access through a Core Network to other networks such as the public Internet, and can allow voice and data communication with other user equipment in the Radio Access Network.

[0052] As used herein, "wireless cellular network" refers to a network of base stations arranged to support radio communications within one or more cells emanating therefrom implementing one or more Radio Access Technologies to establish wireless radio connections to user equipment to enable them to communicate with the core network and other connected networks via the base stations.

[0053] As used herein, "cell" refers to the area of signal coverage a base station provides, within which it is intended that user equipment can establish a wireless radio connection to the base station. The cell is typically formed as a sector emanating from the antenna at the base station, with the nominal radius of the coverage area depending on a range of intrinsic factors including the frequency band, the Radio Access Technology, and antenna and base station configuration and power, and on extrinsic factors such as the topology, man-made structures and atmospheric conditions. The nominal coverage provided by a cell may differ from the location and distancedistribution of user equipment actually establishing a connection with the base station. For example, in an urban environment the density of users may be such that a large number of user equipment are densely located near the base station (e.g. within 1-2 km) such that the majority of the connections are well within the nominal coverage area (which may be e.g. around 12 km for LTE 2100 MHz band). However, connections with user equipment beyond this nominal distance may be established if the user equipment receives a sufficiently strong signal from the base station. Generally, the further away from a base station, the lower the signal strength, the more unstable the signal and the lower the bandwidth. As such the “coverage” a base station can provide within different locations throughout a cell in terms of the number and stability of wireless connections and the data and voice throughput that can be supported thereby depends on a number of local factors including user demand, topology, the built environment, weather, etc, and is thus subject to change. This can lead to variable user experience and unfulfilled user demand, falling below desired or expected quality of experience.

[0054] As used herein, "user equipment" refers to any suitable mobile or stationary device that can transmit and receive wireless communication signals and can form wireless connections with a base station of a wireless cellular network, to thereby conduct voice and data communications through a core network of the wireless cellular network with other user equipment and nodes within the network, or with external networks such as the Internet. UEs can be embodied by any of a number of types of devices including but not limited to printed circuit (PC) cards, compact flash devices, external or internal modems, wireless phones, smartphones, tablets, tracking devices, asset tags, and so on.

[0055] Figure 1 is a schematic illustration of an example of a wireless communication network 100. The network 100 includes a first base station 102, a second base station 104, a core network 106 and an electronic device 108. The electronic device 108 may also be referred to as a user equipment (UE).

[0056] The first base station 102 and the second base station 104 provide network coverage using a Radio Access Technology (RAT). Well known examples of RATs which may be employed in a mobile telecommunications network include, the Global System for Mobile Communications (GSM), the Universal Mobile Telecommunications System (UMTS), Long- Term Evolution (LTE), 5G New Radio (NR) and 6G radio networks. Each of these RATs are well-documented and subject to industry standardisation. Typically, each RAT is assigned one or more frequency bands over which radio communications may be transmitted and receivedusing the respective RAT. Generally, different RATs are assigned different frequency bands so as to avoid frequency overlaps between different RATs. Communications using a given RAT therefore generally use one or more frequency bands assigned to the given RAT and use communication protocols adhering to the relevant standards for the given RAT. The first base station 102 and the second base station 104 may use the same RAT, or different RATs. Each of the first and second base stations may comprise a plurality of nodes.

[0057] The core network 106 may comprise any suitable network infrastructure for providing access to one or more network services, as is commonly known in the art. For example, the core network 106 may comprise an Evolved Packet Core (EPC). Additionally or alternatively, the core network 106 may comprise a Next Generation Core (NextGen Core) or 5G Core (5GC) or future core networks (for example, 6G core networks).

[0058] In the example depicted in Figure 1, the first base station 102 and the second base station 104 are connected to a common core network 106. That is, the first base station 102 and the second base station 104 may provide connectivity to the same core network 106. In the example shown in Figure 1, a connection 110 is provided between the first base station 102 and the core network 106. Furthermore, a connection 112 is provided between the second base station 104 and the core network 106. In addition, a connection 114 may also be provided between the first base station 102 and the second base station 104.

[0059] The connections 110-114 may be provided using one or more nodes provided at each end of the connection. A node is a network switch for connecting different parts of the network 100 together, and is used to route network traffic between the different parts of the network 100.

[0060] As is shown in Figure 1, a geographical coverage area 116 over which the first base station 102 provides network coverage and a geographical coverage area 118 over which the second base station 104 provides network coverage may at least partially overlap. The geographical coverage areas 116, 118 may be referred to as cells.

[0061] The at least partial overlap between the coverage areas 116, 118 provided by the first 102 and second 104 base stations may enable an electronic device 108 situated in a region of coverage overlap to connect to the core network 106 through either or both of the first 102 and the second 104 base stations. The first base station 102 may exchange communications, such as radio frequency signals with the UE 108 over a first air interface 120. The second base station 104 may exchange communications with the UE 108 over a second air interface 122. Exchanging communications may comprise transmitting and / or receiving communications in one or morefrequency bands assigned to a RAT used by the base station and using communication protocols specified for the RAT.

[0062] If network traffic through one or more nodes in the network 100 is uneven, load balancing can be used to redistribute the network traffic in a more equitable manner. The network traffic may be referred to as a load. Load balancing can help to improve the performance and reliability of the network 100, as well as prevent any single node from becoming overloaded.

[0063] Load balancing may comprise distributing the load evenly among cells, and among areas of cells. Load balancing may further comprise transferring part of the load from congested cells, or areas of cells. Load balancing may further comprise offloading one or more electronic devices 108 from one cell, cell area, carrier or RAT to improve network performance.

[0064] Figure 2 illustrates a method 200 for performing a load balancing operation in a wireless communication network 100. The method of FIG. 2 may be implemented by any suitable electronic device or combination of a plurality of electronic devices. For example, the method of FIG. 2 may be implemented by one or more devices which may form one or more network nodes or functions in a mobile telecommunications network. The method of FIG. 2 may, for example, be implemented by one or more devices, network nodes or functions situated in or communicatively coupled to the wireless communication network 100.

[0065] In step 202, quality of experience, QoE, data is received from a plurality of electronic devices 108 connected to the wireless communication network 100. The electronic devices 108 from which the QoE data is received may comprise user electronic devices, UEs, operated by users of the wireless communication network 100. The QoE data comprises data indicative of a quality of a user experience provided by each of the plurality of electronic devices 108 whilst the plurality of electronic devices 108 provide a service using the wireless communication network 100. QoE data as referred to herein directly relates to a quality of a user experience may be linked to a specific end user service or end user service type (e.g., playout and / or streaming of media content). QoE data may, for example, relate to factors which are directly observable by a user whilst receiving an end user service (e.g., delay, quality, experienced when receiving a service such as playout of media content). QoE data is typically measured and / or collected at an application layer (e.g., by an application providing an end user service). QoE data differs from other measures of network performance such as, for example, radio signal measurements and / or Quality of Service (QoS) measures which are not typically measured and directly experienced at an application layer but are determined by a radio layer.

[0066] The QoE data may be collected by the electronic device 108 and, in particular, by an end user application in the electronic device 108. The collected QoE data may not be deducible directly from performance measurements in the network 100 (e.g., radio layer measurements).

[0067] The QoE data may comprise one or more metrics such as an average throughput, an initial playout delay, a buffer level, a playout delay for media start-up and device information. The QoE data may relate to a quality of a user's experience with a streaming of media content over the network 100. More information regarding QoE metrics is provided, for example, in the 3GPP Standards Specification TS 26.247 which is incorporated herein by reference in its entirety.

[0068] In step 204 of the method 200, the received QoE data is input into a load balancing algorithm. The load balancing algorithm is configured to output a load balancing action based on the QoE data wherein the output load balancing action, when implemented, improves a balance of network traffic across a plurality of nodes in the wireless communication network 100. The output load balancing action, when implemented, also improves an overall QoE across the plurality of electronic devices 108 using the wireless communication network 100.

[0069] In step 206 of the method 200, the load balancing action output by the load balancing algorithm is performed. Performing the load balancing action may comprise one or more of: retaining an existing connection between an electronic device from the plurality of electronic devices and a connected first node of the wireless communication network; performing a cell handover to transfer an electronic device attached to a first cell of the wireless communication network to a second cell of the wireless communication network; and performing an inter-radio access technology, RAT, handover on an electronic device to transfer the electronic device from a first RAT to a second RAT.

[0070] For example, as a result of the load balancing action, one or more electronic devices connected to the network may not be transferred to an alternative node, cell and / or RAT. One or more electronic devices may be transferred from a first cell on a first node to a second cell on the first node (intra-node handover). One or more electronic devices may be transferred from a first node to a second node (inter-node handover). One or more electronic devices may be transferred from a first RAT to a second RAT (inter-RAT handover). One or more electronic devices may be transferred from a first cell on a first RAT to a second cell on the first RAT (intra- RAT handover).

[0071] Performing the load balancing action may comprise one or more of: performing idle mode load balancing to adjust a cell reselection parameter for electronic devices in an idle mode;modifying a power of a pilot signal transmitted in the wireless communication network; adjusting an angle of an antenna of a node in the wireless communication network; and modifying a handover parameter condition for cell handover of electronic devices connected to the wireless communication network.

[0072] Figure 3 shows a signal flow diagram illustrating an example method for performing load balancing in a wireless communication network, such as the network 100. The method in Figure 3 is performed by a first node 302, a second node 304, and a UE 108 connected to the first node 302. However, this is merely exemplary and the skilled person would understand that many of the steps described herein could be performed by any entity (or group of entities). The skilled person would also understand that some of the steps described herein could be performed by one or more entities provided outside of the wireless communication network 100.

[0073] The first node 302 and the second node 304 may be Next Generation Radio Access Network (NG-RAN) nodes. The first node 302 and the second node 304 may be part of the same base stations (such as the first base station 102), or part of separate base stations (such as the first base station 102 and the second base station 104).

[0074] In step 306, the second node 304 optionally implements an artificial intelligence (Al) or machine learning (ML) model. The AI / ML model may enable the second node 304 to provide the first node 302 with input information. For example, the input information may comprise a predicted resource status of the second node 304.

[0075] In step 308, the first node 302 configures the UE 108 to provide measurements relating to the UE 108 and / or location information of the UE 108. The measurements may comprise one or more of radio resource management (RRM) measurements, minimization of drive test (MDT) measurements, a velocity of the UE 108, and a position of the UE 108.

[0076] In step 310, the UE 108 collects measurements related to a quality of a network signal received by the UE 108. The collected measurements may comprise Reference Signal Received Power (RSRP), Reference Signal Received Quality (RSRQ), and Signal to Interference plus Noise Ratio (SINR) measurements in relation to a cell servicing the UE 108 and / or one or more neighbouring cells. These measurements provide an indication of properties of network coverage provided to the UE 108 by the network 100.

[0077] In step 312, the UE 108 transmits the measurements collected in step 310 to the first node 302.

[0078] It will be appreciated that steps 310 and 312 may be performed for a plurality of UEs 108 attached to the first node 302. The first node 302 may therefore be provided a plurality of sets of measurements performed by different UEs 108 attached to the first node 302.

[0079] In step 314, the second node 304 transmits input information for load balancing model training to the first node 302. The input information sent to the first node 302 in step 314 may comprise measurements sent to the second node 304 by UEs attached to the second node 304. For example, similarly to steps 308 and 312 performed for the first node 302, the second node 304 may also request UEs attached to the second node 304 to provide measurements made by the UE. For example, UEs attached to the second node 304 may transmit measurements made by the UEs, which comprise properties such as RRM measurements, MDT measurements, velocity of a UE, position of a UE, RSRRP, RSRQ and / or SINR. The input information sent to the first node 302 in step 314 may comprise the measurements made by UEs attached to the second node 304.

[0080] It will be appreciated that, whilst not shown in FIG. 3 further input information (e.g., relating to UE measurements) may be provided to the first node 302 by one or more other nodes in the network 100. In general, the first node 302 may receive input information from one or more neighbouring nodes, where the input information received from each node comprises measurements made by UEs attached to the one or more neighbouring nodes.

[0081] Input information provided by other nodes (e.g., the input information sent by the second node 304 in step 314) may further comprise one or more measurements made by a RAN node. For example, each RAN node may measure or otherwise determine properties indicative of traffic load handled by the RAN node. Such measured or otherwise determined properties may, for example, include properties such as a total transmit power, a total received power, interference in a cell, cell throughput in downlink and / or uplink, increase in blocking and / or a handover failure rate.

[0082] In step 316, the first node 302 trains an AI / ML model located at the first node 302. The skilled person would be aware that there are many potential AI / ML algorithms which could be used as the load balancing algorithm. Some examples of suitable algorithms include heuristic, meta-heuristic, hybrid algorithms, and Content-aware Machine Learning based Load Balancing. However, this is not an exhaustive list and the invention is not intended to be limited to the listed algorithms.

[0083] The AI / ML learning model is trained using training data. The training data may comprise one or more measurements provided by UEs attached to the first node 302,measurements provided by UEs attached to one or more neighbouring nodes (such as the second node 304), measurements or otherwise determined properties indicative of network traffic at the first node and / or measurements or otherwise determined properties indicative of network traffic at one or more neighbouring nodes (e.g., the second node 304). The training data may comprise historical data related to measurements and properties made and determined over a period of time in the past. The historical data may, for example, relate to electronic devices which are, or have been, connected to the network 100. The historical data may comprise one or more of data relating to previous traffic load distribution in the wireless communication network, data relating to previous outcomes of load balancing measures performed by the wireless communication network, and data relating to previous measurements made by one or more of the plurality of electronic devices.

[0084] The AI / ML model may be configured through training (based on the training data) to generate predictions and / or actions which can be used to take load balancing actions based on a set of input data. For example, the AI / ML model may be configured through training to generate a prediction of network traffic load which will result from a given load balancing action. Additionally or alternatively, the AI / ML model may be configured through training to generate one or more load balancing actions which will improve a balance of network traffic load in the network. Any suitable Al and / or ML techniques may be used to train the AI / ML model.

[0085] Steps 308-316 shown in Figure 3 may be considered to form a training stage in which training data is collected (which may be referred to as historical data) and used to train an AI / ML model. Steps 318-330 below may be considered to form a load balancing stage in which one or more load balancing actions are generated and taken based on current measurements related to the state of device and / or network traffic in the network 100.

[0086] In step 318, the UE 108 transmits measurements made by the UE 108 to the first node 302. The measurements sent to the first node 302 at step 318 may correspond to the measurements described above with reference to step 312. For example, the measurements may comprise one or more of properties such as RRM measurements, MDT measurements, velocity of a UE, position of a UE, RSRP, RSRQ and / or SINR. The measurements provided in step 318 comprise data representing a current state of a UE for use in an inference stage to generate one or more load balancing actions (in contract to the measurements provided in step 312 which were used for training of a model). It will be appreciated that step 318 may be carried out for a pluralityof UEs attached to the first node 302 such that measurements corresponding to a plurality of UEs are collected by the first node 302.

[0087] In step 320, the second node 304 transmits input data for load balancing model inference to the first node 302. The input data transmitted at step 320 may correspond to the input data provided by the second node 304 at step 314. For example, the input data transmitted at step 320 may comprise measurements made by UEs attached to the second node 304 and / or measurements and / or properties indicative of network traffic at the second node 304. Corresponding input data may be provided by one or more other neighbouring nodes.

[0088] In step 322, the first node 302 uses a load balancing model to generate a plurality of possible load balancing actions in dependence on the data received in steps 318 and 320. The plurality of possible load balancing actions may generated in further dependence on data measured or determined at the first node 302. For example, the plurality of possible load balancing actions may generated in further dependence on one or more properties indicative of network traffic at the first node 302.

[0089] One or more of the data received at step 318, the data received at step 320 and / or the data measured or determined at the first node 302 may form input data for model inference. The input data may be provided to the trained AI / ML model. The trained AI / ML model may generate (based on the input data) predictions and / or actions which can be used to generate load balancing actions. For example, the AI / ML model may generate a prediction of network traffic load which will result from one or more possible load balancing actions. Such predictions may be used to generate a plurality of load balancing actions which improve a balancing of network traffic. Additionally or alternatively, the AI / ML model may directly generate a plurality of potential load balancing actions which will improve a balance of network traffic load in the network.

[0090] The load balancing actions generated in step 322 may form a plurality of potential load balancing actions which can be selected between.

[0091] In step 324, the first node 302 requests QoE data from the UE 108, and the UE 108 transmits the requested QoE data to the first node 302. The UE 108 may have been recording or logging QoE data prior to receiving the request for the QoE data, and the UE 108 may transmit this logged data to the first node 302. Additionally or alternatively, the UE 108 may begin recording or logging QoE data on request and may transmit the QoE data to the first node 302 once recorded or logged. As was explained in detail above, QoE data may relate to factors which are directly observable by a user whilst receiving an end user service (e.g., delay and / or quality,experienced when receiving a service such as playout of media content). QoE data is typically measured and / or collected at an application layer (e.g., by an application providing an end user service).

[0092] Step 324 may be carried out for a plurality of UEs attached to the first node 302 such that QoE data corresponding to a plurality of UEs is collected by the first node 302.

[0093] In step 326, the first node 302 requests QoE data from the second node 304 and the second node 304 transmits the requested QoE data to the first node 302. The QoE data transmitted from the second node 304 may relate to QoE data of one or more electronic devices attached to the second node 304. The second node 304 may receive and log QoE data from the one or more electronic devices attached to the second node 304 prior to receiving the request for QoE data from the first node 302. Alternatively, upon receiving the request for QoE data from the first node 302, the second node 304 may transmit its own request to its connected electronic devices for QoE data relating to the connected electronic devices.

[0094] Step 326 may be carried out for a plurality of neighbouring nodes (which may include the second node 304) such that QoE data corresponding to QoE data collected by UEs connected to a plurality of nodes is provided to the first node 302. The QoE data provided to the first node 302 (which may include QoE data provided directly by one or more UEs attached to the first node 302 and / or QoE data provided from one or more neighbouring nodes) is representative of an overall QoE being experienced in the network (or at least in the network in the vicinity of the first node 302) and may be used to determine a load balancing action (e.g., selected from the plurality of potential load balancing actions), which improves an overall QoE.

[0095] In step 328, the QoE data received from the UE 108 and the second node 304 (and / or one or more further neighbouring nodes) is input into a load balancing algorithm. The plurality of load balancing predictions generated in step 322 are also input into the load balancing algorithm. The load balancing algorithm uses the received QoE data to refine the plurality of load balancing actions and determine a load balancing action to be output. The determined output load balancing action is a load balancing action which, when implemented, is predicted to improve a distribution of network traffic across the network 100 and improve an overall QoE received by users using a service over the network 100.

[0096] The load balancing algorithm may, for example, select a load balancing action from the plurality of potential load balancing actions. The load balancing algorithm may select a load balancing algorithm from the plurality of potential load balancing actions based on an expectedchange in a QoE as a result of each of the plurality of potential load balancing actions. For example, the load balancing algorithm may be configured to determine an expected change in a QoE for a plurality of electronic devices for each of the plurality of potential load balancing actions. The load balancing algorithm may then select a load balancing algorithm of the potential load balancing algorithms which is expect to lead to an improvement in an overall QoE received by users using a service over the network 100 (based on the determined expected change in a QoE for a plurality of electronic devices for each of the plurality of potential load balancing actions).

[0097] An expected change in a QoE for an electronic device which results from a potential load balancing action may be determined using a predictive model configured to predict a change in a QoE which results from a potential load balancing action. The predictive model may be configured to predict a change in QoE which results from a potential load balancing action in dependence on the QoE data received at the first node 302. For example, the predictive model may use the current state of QoE in the network (as reflected by the QoE data received by the first node 302) to predict a change in QoE for a particular load balancing action.

[0098] Such a predictive model may be configured through training using historically measured data. For example, a predictive model for determining a change in QoE for a load balancing action may be trained using historically collected training data indicative of measured QoE data for configurations of devices in the network (and / or for past load balancing actions). The predictive model may be trained and / or implemented using Al and / or ML techniques and may therefore be referred to as an Al and / or ML model.

[0099] In other examples, the predictive model may be rules based. A rules based predictive model may use the QoE data received by the first node 302 and one or more pre-configured rules to determine an expected change in QoE which results from a given potential load balancing algorithm.

[0100] To provide a simple illustrative example, a potential load balancing algorithm may comprise a handover of an electronic device from a first cell, carrier and / or RAT to a second cell, carrier and / or RAT. The QoE data received by the first node 302 may include data indicative of a QoE currently measured by one or more devices using the first cell, carrier and / or RAT and data indicative of a QoE currently measured by one or more devices using the second cell, carrier and / or RAT. If the QoE measured by the one or more devices using the second cell, carrier and / or RAT is generally better than the QoE measured by the one or more devices using the firstcell, carrier and / or RAT then the predictive model may determine that the QoE at an electronic device is expected to be improved by handover from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT. In such a scenario, the load balancing algorithm may select the potential load balancing algorithm comprising a handover of an electronic device from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT based on the expected improvement in QoE. Conversely if the QoE measured by the one or more devices using the second cell, carrier and / or RAT is generally worse than the QoE measured by the one or more devices using the first cell, carrier and / or RAT, then the predictive model may determine that the QoE at an electronic device is expected to be reduced by handover from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT. In such a scenario, the load balancing algorithm may not select the potential load balancing algorithm comprising a handover of an electronic device from the first cell, carrier and / or RAT to the second cell, carrier and / or RAT based on the expected worsening in QoE.

[0101] Whilst a simple illustrative example has been described above in which a load balancing action is selected based on an expected change of a QoE for a single device, in some examples an expected change of a QoE for a plurality of devices may be considered when selecting a load balancing action. For example, a load balancing action may be selected which is expected to improve an overall QoE for a plurality of devices.

[0102] In step 330, the output load balancing action is implemented. One or more electronic devices may retain an existing connection between the one or more electronic devices and their respective connected nodes. One or more electronic devices be transferred from a first cell of the wireless communication network to a second cell of the wireless communication network. One or more electronic devices may be transferred from a first RAT to a second RAT.

[0103] In addition, the output load balancing action may involve one or more of performing idle mode load balancing to adjust a cell re-selection parameter for electronic devices in an idle mode; modifying a power of a pilot signal transmitted in the wireless communication network; adjusting an angle of an antenna of a node in the wireless communication network; and modifying a handover parameter condition for cell handover of electronic devices connected to the wireless communication network.

[0104] In step 332, the first node 302 receives feedback information from the second node 304. The feedback information may comprise information regarding a new load operating across the second node 304 after the output load balancing action has been performed. The feedbackinformation may comprise information relating to a new overall QoE received by users of electronic devices connected to the second node 304 after the output load balancing action has been performed.

[0105] In some examples, the first node 302 may also receive feedback information from each of one or more electronic devices 108 connected to the first node 302 after the output load balancing action has been performed. The feedback information received from the one or more electronic devices 108 may comprise data relating to a QoE experienced by a user of the UE 108 as a result of the load balancing action.

[0106] In some examples, feedback information may be used to update or further train one or more models or algorithms described above. For example, feedback information may be used to update or further train all or part of the load balancing algorithm.

[0107] The load balancing algorithm for selecting a load balancing action from a plurality of potential load balancing actions (and based on QoE data) has been described above as being separate from an algorithm or model for generating potential load balancing actions. For example, an AI / ML learning algorithm may be used to generate a plurality of potential load balancing actions for improving a balance of traffic load in the network, and the load balancing algorithm may select a load balancing action from the plurality of potential load balancing actions (and in dependence on the QoE data). However, in some examples the load balancing algorithm may be further configured to generate the plurality of potential load balancing actions.

[0108] In some examples, a load balancing algorithm may be configured to optimise, jointly, for both a balance of network traffic in the network and an overall QoE across a plurality of electronic devices using the network. This could also be considered concurrent optimisation of a balance of network traffic in the network and an overall QoE across a plurality of electronic devices using the network. For example, the load balancing algorithm may be configured to execute an AI / ML model which has been configured through training to generate predictions and / or optimal actions for both balancing of network traffic and improving QoE. The AI / ML model may be trained using training data which includes measured QoE data.

[0109] In the examples presented with reference to Figure 3, both the training of a model, model inference and the load balancing algorithm are all executed at the first node 302. However, in other examples, one or more of these actions may be implemented by one or more other devices, network nodes and / or network functions. For example, model training may be implemented, for example, by an Operations, Administration and Maintenance (0AM) function. A trained modelmay be provided to the first node 302 (or other node) for model inference. In some examples, the functionality of a RAN node (e.g., a gNB) may be split between a Centralized Unit (CU) or gNB-CU and a Distributed Unit (DU) of gNB-DU. In such examples, model inference may be implemented by a gNB-CU and model training may be implemented by an 0AM or a gNB-CU. In some examples, the model inference may be implemented by a node provided outside the RAN (for example, the model inference may be implemented in a Cloud network), and model training may be implemented by a RAN node such as a gNB. In some examples, both the model inference and the model training may be implemented outside the RAN (for example, in a Cloud network), and the load balancing action output by the model may be provided to a gNB.

[0110] Various methods have been described herein in which some of the method steps may be implemented on any suitable electronic device (such as a computing device) and / or combination of electronic devices (e.g. computing devices). Figure 4 is a schematic illustration of an example electronic device 402 which may be used to implement all or part of any method described herein.

[0111] The electronic device 402 may include at least one processing unit 404, memory 408 and an input / output interface 406. The processing unit 404 may include any suitable processor and / or combination of processors. For example, the processing unit 404 may include one or more of a Central Processing Unit (CPU) and a Graphical Processing Unit (GPU). The memory 408 may include volatile memory and / or non-volatile / persistent memory. The memory 408 may, for example, be used to store data such as an operating system, instructions to be executed by the processing unit (e.g. in the form of software to be executed by the processing unit), configuration information related to the electronic device 402, session information and / or configuration or registration information associated with any other device, node or module in the network. In some examples, the memory 408 may be used to store instructions for executing any of the methods disclosed herein.

[0112] At least the processing unit 404 is connected to the input / output interface 406. The input / output interface 406 may facilitate communication with one or more other devices. For example, the input / output interface 406 may be operable to transmit and / or receive communications to / from other devices in a network.

[0113] Optionally, the electronic device 402 may further include a display (not shown). The display may comprise any suitable electronic display such as a touch sensitive display. The display may be connected to at least to the processing unit 404. The processing unit 404 may generate display signals which are sent to the display in order to cause the display information.

[0114] In the interest of conciseness not all possible alternatives which fall within the scope of the present disclosure have been explicitly discussed herein. As the skilled person will appreciate, in the present disclosure any aspect discussed from the perspective of an element being operable to do an action also discloses the same feature from the perspective of a method including a method step corresponding to the action. Similarly, any discussion presented from the perspective of a method step also discloses the same features from the perspective of any one or more suitable elements being operable or configured to carry out some or all of the method step. It is also considered within the present disclosure that for any method step(s), there can be a computer program configured to carry out, when executed, the method step(s).

[0115] Examples of the present disclosure can be realised in the form of hardware, software or a combination of hardware and software. Any such software may be stored in the form of volatile or non-volatile storage such as, for example, a storage device like a ROM, whether erasable or rewritable or not, or in the form of memory such as, for example, RAM, memory chips, device or integrated circuits or on an optically or magnetically readable medium such as, for example, a CD, DVD, magnetic disk or magnetic tape. It will be appreciated that the storage devices and storage media are embodiments of machine-readable storage that are suitable for storing a program or programs that, when executed, implement examples of the present disclosure. Accordingly, examples provide a program comprising code for implementing a system or method as claimed in any preceding claim and a machine readable storage storing such a program. Still further, examples of the present disclosure may be conveyed electronically via any medium such as a communication signal carried over a wired or wireless connection and examples suitably encompass the same.

[0116] Features, integers, characteristics, or groups described in conjunction with a particular aspect, embodiment or example of the invention or disclosure are to be understood to be applicable to any other aspect, embodiment or example described herein unless incompatible therewith. All of the features disclosed in this specification (including any accompanying claims, abstract and drawings), and / or all of the steps of any method or process so disclosed, may be combined in any combination, except combinations where at least some of such features and / or steps are mutually exclusive. The invention is not restricted to the details of any foregoing examples.

Claims

CLAIMS1. A method for performing load balancing in a wireless communication network, the method comprising: receiving quality of experience, QoE, data from a plurality of electronic devices connected to the wireless communication network, the QoE data indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide a service using the wireless communication network; inputting the received QoE data into a load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data wherein the output load balancing action, when implemented, adjusts a balance of network traffic across a plurality of nodes in the wireless communication network to improve an overall QoE across the plurality of electronic devices using the wireless communication network; and performing the load balancing action output by the load balancing algorithm.

2. The method of claim 1, wherein the load balancing algorithm is configured to: receive a plurality of potential load balancing actions, wherein each of the potential load balancing actions is an action which may be performed in the communication network to adjust the balance of network traffic across a plurality of nodes in the communication network; select from the plurality of potential load balancing actions and in dependence on the received QoE data, a load balancing action which, when implemented, improves an overall QoE across the plurality of electronic devices using the wireless communication network; and output the selected load balancing action.

3. The method of claim 2, wherein, the load balancing algorithm is further configured to receive data indicative of current traffic load distribution in the wireless communication network and generate the plurality of potential load balancing actions in dependence on the data indicative of current traffic load distribution in the wireless communication network, and wherein the method further comprises: receiving data indicative of current traffic load distribution in the wireless communication network; and inputting the received data indicative of current traffic load distribution in the wireless communication network into the load balancing algorithm to generate the plurality of potential load balancing actions.

4. The method of claim 3, wherein the data indicative of current load distribution in the wireless communication network comprises one or more measurements indicative of a quality of a network signal received by an electronic device from among the plurality of electronic devices.

5. The method of any of claim 2 to claim 4, wherein the load balancing algorithm uses the input QoE data to select the output load balancing action, from the plurality of potential load balancing actions, based on an expected change in the overall QoE of the plurality of electronic devices in the wireless communication network, which results from each of the plurality of potential load balancing actions.

6. The method of claim 1, wherein the method further comprises: receiving data indicative of current load distribution in the wireless communication network; and further inputting the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data and the data indicative of current load distribution in the wireless communication network, such that the output load balancing action, when implemented, improves a balance of network traffic across the plurality of nodes in the wireless communication network and the overall QoE across the plurality of electronic devices using the wireless communication network.

7. The method of any of the preceding claims, wherein the method further comprises: receiving data indicative of current load distribution in the wireless communication network; and further inputting the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data and the data indicative of current load distribution in the wireless communication network, such that the output load balancing action, when implemented, jointly optimizes the balance of network traffic across the plurality of nodes in the wireless communication system and the overall QoE of the plurality of electronic devices.

8. The method of any of the preceding claims, further comprisingreceiving feedback information from the plurality of electronic devices after the load balancing action is performed.

9. The method of any of the preceding claims, wherein the load balancing action comprises one or more of: retaining an existing connection between an electronic device from the plurality of electronic devices and a connected first node of the wireless communication network, performing a cell handover to transfer an electronic device attached to a first cell of the wireless communication network to a second cell of the wireless communication network; or performing an inter-radio access technology, RAT, handover of an electronic device to transfer the electronic device from a first RAT to a second RAT.

10. The method of any of the preceding claims, wherein the load balancing action comprises one or more of: performing idle mode load balancing to adjust a cell reselection parameter for electronic devices in an idle mode; modifying a power of a pilot signal transmitted in the wireless communication network; adjusting an angle of an antenna of a node in the wireless communication network; or modifying a handover parameter condition for cell handover of electronic devices connected to the wireless communication network.

11. The method of any of the preceding claims, wherein the QoE data comprises one or more of: an average throughput, an initial playout delay, a buffer level or a playout delay for a media startup, as measured whilst the electronic device provides the service using the communication network.

12. The method of any of the preceding claims, wherein the load balancing algorithm is configured through training to select the output load balancing action; wherein the training of the load balancing algorithm uses historical data received from a plurality of electronic devices; and wherein the historical data comprises one or more of data relating to previous traffic load distribution in the wireless communication network, data relating to previous outcomes of load balancing measures performed by the wireless communication network, and data relating to previous measurements made by one or more of the plurality of electronic devices.

13. The method of claim 12, further comprising training, using the received historical data, the load balancing algorithm to select the output load balancing action.

14. A system for performing load balancing in a wireless communication network, the system configured to: receive quality of experience, QoE, data from a plurality of electronic devices connected to the wireless communication network, the QoE data indicative of a quality of a user experience provided by each of the plurality of electronic devices whilst the plurality of electronic devices provide a service using the wireless communication network; input the received QoE data into a load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data wherein the output load balancing action, when implemented, adjusts a balance of network traffic across a plurality of nodes in the wireless communication network to improve an overall QoE across the plurality of electronic devices using the wireless communication network; and perform the load balancing action output by the load balancing algorithm.

15. The system of claim 14, wherein the load balancing algorithm is configured to: receive a plurality of potential load balancing actions, wherein each of the potential load balancing actions is an action which may be performed in the communication network to adjust the balance of network traffic across a plurality of nodes in the communication network; select from the plurality of potential load balancing actions and in dependence on the received QoE data, a load balancing action which, when implemented, improves an overall QoE across the plurality of electronic devices using the wireless communication network; and output the selected load balancing action.

16. The system of claim 15, wherein, the load balancing algorithm is further configured to receive data indicative of current traffic load distribution in the communication network and generate the plurality of potential load balancing actions in dependence on the data indicative of current traffic load distribution in the communication network, and wherein the system is further configured to: receive data indicative of current traffic load distribution in the wireless communication network; andinput the received data indicative of current traffic load distribution in the wireless communication network into the load balancing algorithm to generate the plurality of potential load balancing actions.

17. The system of claim 16, wherein the data indicative of current load distribution in the wireless communication network comprises one or more measurements indicative of a quality of a network signal received by an electronic device from among the plurality of electronic devices.

18. The system of any one of claim 15 to claim 17, wherein the load balancing algorithm uses the input QoE data to select the output load balancing action, from the plurality of potential load balancing actions, based on an expected change in the overall QoE of the plurality of electronic devices in the wireless communication network, which results from each of the plurality of potential load balancing actions.

19. The system of claim 14, wherein the system is further configured to: receive data indicative of current load distribution in the wireless communication network; and; further input the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data and the data indicative of current load distribution in the wireless communication network, such that the output the load balancing action, when implemented, improves a balance of network traffic across the plurality of nodes in the wireless communication network and the overall QoE across the plurality of electronic devices using the wireless communication network.

20. The system of any one of claim 14 to claim 19, wherein the system is further configured to: receive data indicative of current load distribution in the wireless communication network; and further input the received data indicative of current load distribution in the wireless communication network into the load balancing algorithm, the load balancing algorithm configured to output a load balancing action based on the QoE data and the data indicative of current load distribution in the wireless communication network, such that the output load balancing action, when implemented, jointly optimizes the balance of network traffic across theplurality of nodes in the wireless communication system and the overall QoE of the plurality of electronic devices.

21. The system of any one of claim 14 to claim 20, wherein the system is further configured to: receive feedback information from the plurality of electronic devices after the load balancing action is performed.

22. The system of any one of claim 14 to claim 21, wherein the load balancing action comprises one or more of: retaining an existing connection between an electronic device from the plurality of electronic devices and a connected first node of the wireless communication network, performing a cell handover to transfer an electronic device attached to a first cell of the wireless communication network to a second cell of the wireless communication network; or performing an inter-radio access technology, RAT, handover on an electronic device to transfer the electronic device from a first RAT to a second RAT.

23. The system of any one of claim 14 to claim 22, wherein the load balancing action comprises one or more of: performing idle mode load balancing to adjust a cell reselection parameter for electronic devices in an idle mode; modifying a power of a pilot signal transmitted in the wireless communication network; adjusting an angle of an antenna of a node in the wireless communication network; or modifying a handover parameter condition for cell handover of electronic devices connected to the wireless communication network.

24. The system of any one of claim 14 to claim 23, wherein the QoE data comprises one or more of: an average throughput, an initial playout delay, a buffer level or a playout delay for a media start-up, as measured whilst the electronic device provides the service using the communication network.

25. The system of any one of claim 14 to claim 24, wherein the load balancing algorithm is configured through training to select the output load balancing action; wherein the training of the load balancing algorithm uses historical data received from the plurality of electronic devices; andwherein the historical data comprises one or more of data relating to previous traffic load distribution in the wireless communication network, data relating to previous outcomes of load balancing measures performed by the wireless communication network, and data relating to previous measurements made by one or more of the plurality of electronic devices.

26. The system of claim 25, wherein the system is further configured to train, using the received historical data, the load balancing algorithm to select the output load balancing action.

27. A computer-readable medium comprising instructions which, when executed by a computing apparatus, cause the computing apparatus to carry out the method of any of claims 1 to 13.

Images