Scheduling resource to a user equipment

Abstract

The disclosure provides methods for scheduling radio resource to a user equipment A method (200) performed by a user equipment, UE, (101) for enabling scheduling of a radio resource to the UE (101) in a communication network (100). The UE (101) collecting (201) application information from an application (601) in the UE (101). The UE (101) receiving (202) Radio Access Network, RAN, performance data from a RAN node (102). The UE (101) obtaining (204) an inference result by inputting the application information and the RAN performance data to a trained machine learning, ML, framework (500). The UE (101) transmitting (205) the inference result to the RAN node. The UE (101) obtaining (206) the radio resource from the RAN node based on the inference result. Further, there are methods performed by the RAN node and the application boost service node for enabling the scheduling of the radio resource to the UE.

Description

[0001] SCHEDULING RESOURCE TO A USER EQUIPMENT

[0002] TECHNICAL FIELD

[0003] The disclosure relates to methods for scheduling a radio resource to a user equipment. The disclosure also relates to a user equipment, a radio access network node, an application boost service node and, computer programs and computer program products comprising the computer programs for scheduling a radio resource to the user equipment.

[0004] BACKGROUND

[0005] Today it is possible to deploy multiple performance characteristics in a 5G stand-alone (SA) network through a combination of network slicing, radio resource partitioning (RRP), quality-of-service profiles (QoS), and user equipment route selection policy (URSP). There could for instance be a “low-performance” offer where end-users get unlimited data at a low cost, but which might not always be available on-demand, or that might experience a large latency, or jitter. Alongside this, there might also be a “high-performance” offer, where end-users get a dedicated “on-demand” performance with low-latency and high throughput whenever the end-users require it, however at a higher cost. Naturally, there may also be other types offers with different characteristics.

[0006] While end-users can permanently select a certain performance characteristic, there is also a technical possibility to temporarily “boost”, or change, the performance characteristic of the end-user. For instance, by changing the performance characteristic from “low-performance” to “high-performance”.

[0007] The problem with the existing technology is that while there exists a possibility to change the performance characteristics today, for instance from a low-performance characteristic to a high-performance one, it does not necessarily always lead to an increased performance for the end-user.

[0008] One example would be that a “low-performance” user changes to a “high- performance” characteristic and in turn receives a higher scheduling priority in the radio access network. EP 1901442 B1 discloses a user prioritization scheme, wherein users are prioritized based on their average throughput. According to this document, a "score" is maintained for each active user to be scheduled for data transmission. A scheduling processor then uses the scores to prioritize users for the channel assignments. A set of active users, users with data to transmit, is prioritized such that the user with the lowest score is assigned the highest priority, and the user with the highest score is assigned the lowest priority. The score is approximately proportional to a normalized average throughput of the user.

[0009] Should there be a sufficiently poor signal quality for the user or should there be a sufficiently high contention in the network, the end-user will not see any improved performance characteristics in such a case. Of course, there might be some improvement, but it is not guaranteed, and it is not known beforehand. Finally, there is no possibility for the change from low performance to high performance to occur automatically in order to ensure a certain quality-of-experience for the end-user.

[0010] SUMMARY

[0011] The disclosed invention offers a mechanism to enable a boosting dynamically and automatically whenever it could lead to an increased quality-of-experience for an enduser. It is an aim of the invention to improve scheduling of a radio resource to a user equipment, UE, in a communication network and thus, to enable a better radio resource management in the communication network.

[0012] According to a first aspect of the invention, a method performed by a UE for enabling scheduling of a radio resource to the UE in a communication network is provided. The method comprises collecting application information from an application in the UE. The method comprises receiving Radio Access Network, RAN, performance data from a RAN node. The method comprises obtaining an inference result by inputting the application information and the RAN performance data to a trained machine learning, ML, framework. The method comprises transmitting the inference result to the RAN node. The method comprises obtaining the radio resource from the RAN node based on the inference result. Hereby, the invention provides radio resource to the UE based on the application information and the RAN performance data and thus, enabling an improvement in radio resource scheduling.

[0013] According to a second aspect of the invention, a method performed by a RAN node for scheduling a radio resource to the UE in a communication network is provided. The method comprises transmitting RAN performance data to the UE. The method comprises receiving an inference result from the UE wherein the inference result comprises an indication whether to prioritize the UE during scheduling of the radio resource in the communication network. The method comprises providing the radio resource to the UE based on the inference result.

[0014] According to a third aspect of the invention, a method performed by an application boost service node for enabling scheduling of a radio resource to the UE in a communication network is provided. The method comprises receiving RAN performance data from a RAN node. The method comprises receiving application information from the UE. The method comprises training a ML framework, based on the received RAN performance data and the received application information, to predict whether to prioritize the UE when scheduling the radio resource in the communication network.

[0015] According to a fourth aspect of the invention, there is presented a UE for enabling scheduling of a radio resource to the UE in a communication network. The UE comprises processing circuitry configured to cause the UE to collect application information from an application in the UE. The processing circuitry is configured to cause the UE to receive RAN performance data from a RAN node. The processing circuitry is configured to cause the UE to obtain an inference result by inputting the application information and the RAN performance data to a trained ML framework. The processing circuitry is configured to cause the UE to transmit the inference result to the RAN node. The processing circuitry is configured to cause the UE to obtain the radio resource from the RAN node based on the inference result.

[0016] According to a fifth aspect of the invention, there is presented a RAN node for scheduling a radio resource to a UE in a communication network. The RAN node comprises processing circuitry configured to cause the RAN node to transmit RAN performance data to the UE. The processing circuitry is configured to cause the RAN node to receive an inference result from the UE wherein the inference result comprises an indication whether to prioritize the UE during scheduling of the radio resource in the communication network. The processing circuitry is configured to cause the RAN node to provide the radio resource to the UE based on the inference result. According to a sixth aspect of the invention, there is presented an application boost service node for enabling scheduling of a radio resource to a UE connected to a communication network. The application boost service node comprises processing circuitry configured to cause the application boost service node to receive RAN performance data from a RAN node. The processing circuitry is configured to cause the application boost service node to receive application information from the UE. The processing circuitry is configured to cause the application boost service node to train a ML framework, based on the received RAN performance data and the received application information, to predict whether to prioritize the UE when scheduling the radio resource in the communication network.

[0017] According to a seventh aspect of the invention, there is presented a computer program comprising instructions which when executed on a processor of a UE causes the UE to perform a method according to the first aspect of the invention.

[0018] According to an eighth aspect of the invention, there is presented a computer program product which comprises a computer readable storage medium on which a computer program according to the seventh aspect of the invention is stored.

[0019] According to a ninth aspect of the invention, there is presented a computer program comprising instructions which when executed on a processor of a RAN node causes the RAN node to perform a method according to the second aspect of the invention.

[0020] According to a tenth aspect of the invention, there is presented a computer program product which comprises a computer readable storage medium on which a computer program according to the ninth aspect of the invention is stored.

[0021] According to an eleventh aspect of the invention, there is presented a computer program comprising instructions which when executed on a processor of an application boost service node causes the application boost service node to perform a method according to the third aspect of the invention.

[0022] According to a twelfth aspect of the invention, there is presented a computer program product which comprises a computer readable storage medium on which a computer program according to the eleventh aspect of the invention is stored. Other objectives, features and advantages of the enclosed embodiments will be apparent from the following detailed disclosure, from the attached dependent claims as well as from the drawings.

[0023] BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Features and advantages of the invention, will be better understood through the following illustrative and non-limiting detailed description of embodiments of the invention, with reference to the appended drawings, in which:

[0025] Figure 1 illustrates a communication network.

[0026] Figure 2 is a flow chart illustrating a method for enabling scheduling of a radio resource to a UE in the communication network.

[0027] Figure 3 is a flow chart illustrating a method for scheduling a radio resource to the UE in the communication network.

[0028] Figure 4 is a flow chart illustrating a method for enabling scheduling of a radio resource to the UE in the communication network.

[0029] Figure 5 depicts a Machine Learning (ML) framework for scheduling a radio resource.

[0030] Figure 6 illustrates a signaling diagram depicting the interaction between the UE, a RAN node and an application boost service node.

[0031] Figure 7 is a block diagram illustrating an example UE.

[0032] Figure 8 is a block diagram illustrating an example RAN node.

[0033] Figure 9 is a block diagram illustrating an example application boost service node.

[0034] All the figures are schematic, not necessarily to scale, and generally only show parts which are necessary in order to elucidate the invention, wherein other parts may be omitted or merely suggested.

[0035] DETAILED DESCRIPTION

[0036] The invention proposes a dynamic identification of when Radio Access Network (RAN) performance of a user equipment (UE) might benefit from a boost and a dynamic learning and identification of when an application of the UE might benefit from the boost. The invention enables a possibility of automatic boost, or changes, to the performance characteristics offered by the RAN to the UE. The invention enables monitoring and learning of necessary radio network conditions for the boost to be successful. The invention enables monitoring and learning of the necessary UE application conditions for the boost to be successful. The invention combines necessary radio network conditions and necessary UE application conditions for the boost to be successful to suggest and perform an automatic boost, or change, to the performance characteristics offered by the RAN to the UE.

[0037] Boost refers to a UE being given a higher priority for a better QoS, compared to other UEs, in the RAN during scheduling of resources to UEs in RAN.

[0038] Scheduling a resource or resources to a UE in a communication network refers to a process of determining when and how a particular UE is allowed to access one or more available radio resources (e.g., time slots, frequency bands, and codes) for transmitting and / or receiving data. In the communication network, multiple UEs may need to share the same radio resources, and scheduling is necessary to avoid interference and ensure efficient use of the available radio resources.

[0039] In the communication network, scheduling radio resources to the UE is necessary to ensure efficient utilization of network and to provide fair access to all users. When multiple users are connected to the network, they compete for resources such as bandwidth and processing power. With poor scheduling, there is a risk of network congestion, where some users may experience slow data rates or dropped connections due to insufficient radio resources.

[0040] Moreover, scheduling can also ensure fair access to radio resources for all users, as it prevents any particular user from monopolizing the network to the detriment of others. It can also help to enforce quality of service (QoS) guarantees, where different types of traffic can be assigned different priority levels based on their requirements. Radio resource scheduling in the communication network is a critical function that helps to optimize resource utilization, improve network performance, and ensure fair access to radio resources for all users.

[0041] A scheduling algorithm takes into account various factors such as the QoS requirements of different UEs, the amount of data to be transmitted or received, channel conditions, and the priority of the UEs. Based on these factors, the scheduler assigns radio resources to different UEs at different times, ensuring that each UE gets its fair share of the radio resources and the network capacity is utilized optimally.

[0042] Prioritizing a UE ensures that the UE receives better QoS compared to other UEs. It means that the prioritized UE's data packets are given higher priority for transmission compared to other UEs with lower priority, resulting in reduced latency and improved throughput for the prioritized UE. This is particularly beneficial for applications that require real-time or low-latency communication, such as voice and video calls, online gaming, and critical loT applications. By prioritizing a UE, the network can deliver a better overall user experience. The prioritized UE will experience faster data rates, reduced packet loss, and improved responsiveness, leading to smoother and more reliable services. This can enhance user satisfaction, increase customer loyalty, and reduce churn rates for service providers.

[0043] Figure 1 illustrates a communication network 100. The communication network 100 enables connectivity between the UE and communication network nodes. In that sense, the communication network 100 is here a 3rdGeneration Partnership Project (3GPP) network configured to operate according to predefined rules or procedures, such as specific RAN standards that comprise, but are not limited to: Global System for Mobile Communications (GSM); Universal Mobile Telecommunications System (UMTS); Long Term Evolution (LTE), New Radio (NR) or any applicable future generation standard (e.g., 6G). The communication network 100 comprises one or more radio RAN nodes 102 and comprises one or more core network nodes. A RAN node 102 comprises a base station, eNodeB, gNodeB, or any other current or future implementation of functionality facilitating the exchange of radio network signals between nodes and / or UEs of the communication network 100. The communication network 100 enables communication between a UE 101 and communication network nodes- the RAN node 102 and the Application boost service node 103. The UE 101 refers to a device capable, configured, arranged and / or operable to communicate wirelessly with network nodes and / or other UEs in a communication network. Examples of a UE include, but are not limited to, a smart phone, mobile phone, cell phone, voice over IP (VoIP) phone, wireless local loop phone, personal digital assistant (PDA). Other examples include any UE identified by the 3rd Generation Partnership Project (3GPP), including a narrow band internet of things (NB-loT) UE, a machine type communication (MTC) UE, and / or an enhanced MTC (eMTC) UE. A UE in the form of an Internet of Things (loT) device may be a device for use in one or more application domains, these domains comprising, but not limited to, home, city, wearable technology, extended reality, industrial application, and healthcare. The communication network 100 comprises an application boost service node 103. The application boost service node 103 is implemented in a computing device, in a server apparatus and / or in a virtualized environment, for example in a cloud, or edge cloud. The application boost service node 103 may be comprised or be instantiated in any part of the communication network 100, for example in a core network, a network management center, and a network operations center. The functionality of the application boost service node 103 can be divided into several logical and / or physical functions and can be implemented in a computing device, in a server apparatus and / or in a virtualized environment, for example in a cloud, or edge cloud.

[0044] Figure 2 is a flow chart illustrating a method 200, performed by the UE 101 , for enabling scheduling of a radio resource to the UE 101 in the communication network 100 according to a first embodiment of the invention. The method 200 comprises collecting, 201 , application information from an application 601 in the UE 101. The method 200 comprises receiving, 202, RAN performance data from the RAN node 102. The method 200 comprises obtaining, 204, an inference result by inputting the application information and the RAN performance data to a trained machine learning, ML, framework 500. The method 200 comprises transmitting, 205, the inference result to the RAN node 102. The method 200 comprise obtaining, 206, radio resource from the RAN node 102 based on the inference result.

[0045] According to a second embodiment of the invention, the method 200 further comprises receiving, 203, the trained ML framework 500 (referring to Figure 5) from the application boost service node 103.

[0046] Figure 3 depicts a method 300 performed by the RAN node 102 for scheduling radio resource to the UE 101 according to the first and second embodiment of the invention. The method 300 comprises transmitting, 301 , RAN performance data to the UE 101. RAN performance data comprises value of one or more parameters of the RAN, such as bandwidth, latency, jitter, packet loss, throughput or a combination thereof, that the UE 101 may experience when the UE 101 uses the radio resource offered by the RAN node 102. The method 300 comprises receiving, 302, the inference result from the UE 101. The inference result may indicate whether UE 101 can be prioritized or not. The method comprises providing, 303, radio resource to the UE 101 based on the inference result. The inference result indicates whether the UE 101 can be prioritized for a boost in the RAN performance. The RAN node 102, based on the inference result, may prioritize the UE 101 when scheduling radio resource. When prioritized, the UE 101 may experience improvement in bandwidth, latency, jitter, packet loss, throughput or a combination thereof and thus, the UE may experience a boost in its RAN performance.

[0047] Figure 4 depicts a method 400 performed by the application boost service node 103 for enabling scheduling of radio resource to the UE 101 according to the first embodiment of the invention. The method 400 comprises receiving, 401 , RAN performance data from the RAN node 102. The method 300 comprises receiving, 402, application information from the UE 101. Application information comprises the information associated with state of the application. Application information comprises the information associated with performance of the application. The method 400 comprises training, 403, the ML framework 500 based on the received RAN performance data and the received application information to predict whether to prioritize the UE 101 when scheduling radio resources in the communication network

[0048] 100. Training the ML framework 500 is performed according to Figure 5. The ML training framework is implemented in application boost service node 103.

[0049] According to the second embodiment of the invention, the method 400 further comprises transmitting, 404, the trained ML framework 500 to the UE 101 .

[0050] Figure 6 depicts the interaction between the UE 101 , the RAN node 102 and the application boost service node 103 according to the first and second embodiments of the present disclosure.

[0051] The method 200 comprises collecting, 201 , application information from the application 601 in the UE 101. Referring to Figure 6, the application 601 is executed in the UE 101 . Application 601 may refer to a computer program or a set of computer programs designed to perform specific tasks or functions for end-user(s) of the UE

[0052] 101. Application 601 is a software entity that is typically developed to address a particular need or provide a specific service. Application 601 may comprise web applications such as, online banking systems, email clients, project management tools, social media platforms, etc. Application 601 may comprise mobile applications such as messaging, gaming, navigation, weather updates, and social networking, etc. Application 601 may comprise enterprise applications such as customer relationship management (CRM), enterprise resource planning (ERP), human resources management (HRM), and supply chain management (SCM), etc. Application 601 may comprise embedded applications such as applications designed for home appliances, automobiles, medical equipment, or industrial machinery, etc.

[0053] The UE 101 comprises a UE Boost controller 603. The UE Boost controller 603 controls when to propose a boost to improve the performance characteristics of the application 601 , when the application 601 is executed in the UE 101. The UE Boost controller 603 is a software application. The UE Boost controller 603 may be a computer program or a set of computer programs. The UE Boost controller 603 is designed to perform controlling when to propose a boost to improve the performance characteristics of the application 601 . The UE 101 is equipped to execute the UE Boost controller 603. The UE 101 may comprise one or more processors. The UE 101 may execute the UE Boost controller 603 software application in the one or more processors.

[0054] The UE Boost controller 603 controls when to propose a boost to improve the performance characteristics of the application 601 by collecting, 201 , application information from the application 601 in the UE 101 according to the method 200. The application information may be collected via an interface 600c. Application information may comprise information associated with state of the application. Application information may comprise information associated with performance of the application. The interface 600c may comprise two interfaces-first interface and second interface. The first interface may be used for communicating information associated with state of the application, that is, any data that the application 601 regards as important to determine its state. State of the application 601 may refer to a condition or configuration of the application 601 at a particular point in time. Information associated with state of the application 601 may refer to a collection of data and variables that describe a condition or configuration of the application 601 at a particular point in time. Information associated with state of the application 601 represents values of all variables, objects, and resources that the application 601 is actively using or managing. For example, information associated with state of the application may comprise data related to a current status and progress of any ongoing tasks or operations performed by the application 601 , such as file uploads, calculations, or network requests. State of the application may comprise data related to settings and preferences, which may affect the performance of the application 601 . State of the application may also comprise location settings of the UE 101 . Application information may comprise the location information of the UE 101 .

[0055] The second interface may be used for communicating information associated with performance of the application 601 . Information associated with performance of the application 601 may refer to any data that quantifies the performance of the application 601 when the application 601 is executed in the UE 101 and uses the communication network 100 for one or more aspects of its execution. For example, information associated with performance of the application 601 may refer to data related to latency of the application. For example, information associated with performance of the application 601 may refer to frame rate of the application 601 .

[0056] The UE Boost controller 603 may request the application 601 to provide the application information from the application 601 using the interface 600a. The application 601 transmits the application information via interface 600a either upon receiving a request from the UE Boost controller 603 or automatically, that is, without receiving a request from the UE Boost controller 603. The UE Boost controller 603 may collect the application information during an execution time of the application 601 . The UE Boost controller 603 may collect the application information after the application 601 is executed.

[0057] The method 200 comprises receiving, 202, RAN performance data from the RAN node 102. RAN node 102 may comprise a base station with which the UE 101 is in communication with. RAN performance data may comprise value of one or more parameters such as bandwidth, latency, jitter, packet loss, throughput or a combination thereof, that the UE 101 may experience when the UE 101 uses the radio resources offered by the RAN node 102. RAN node 102 may transmit RAN performance data to the UE 101 using an interface 600e. Interface 600e may be an interface that may be capable of transmitting RAN performance data from the RAN node 102 to the UE 101 .

[0058] Application information may provide an indication regarding a performance of RAN required for the performance of the application 601 . For example, the application information may comprise a frame rate of the application 601 . The frame rate of an application refers to the number of frames (or images) that are displayed per second. A high frame rate generally results in smoother and more fluid motion in the application 601 . However, achieving a high frame rate requires rendering and transmitting a larger amount of data.

[0059] On the other hand, the bandwidth of the RAN refers to the capacity of the RAN connection to transmit data. Bandwidth determines how much data can be transferred over the RAN within a given time. Higher bandwidth allows for faster and more reliable data transmission between devices.

[0060] The frame rate of the application 601 can have implications for the required network bandwidth. A higher frame rate means more frames need to be transmitted over the network per second, which results in increased data transfer. Consequently, a higher frame rate generally requires a higher bandwidth to accommodate the increased data load.

[0061] For example, in video streaming applications, a higher frame rate (e.g., 60 frames per second) requires more data to be transmitted compared to a lower frame rate (e.g., 30 frames per second). This means that streaming a video at 60 frames per second will require a higher bandwidth to ensure smooth playback without buffering or interruptions.

[0062] It is important to consider both the frame rate (application information) of the application 601 and the available network bandwidth (RAN performance data) to provide a seamless and optimal user experience. Insufficient network bandwidth may result in dropped frames, reduced video quality, or increased latency, leading to a less smooth and enjoyable experience for the users.

[0063] Referring to Figure 2, the method 200 comprises receiving, 203, a trained ML framework 500 (referring to Figure 5) from the application boost service node 103 according to the second embodiment of the invention. The application boost service node 103 is responsible for training the ML framework 500. The trained ML framework 500 is executed by the UE Boost controller 603. The ML framework 500 is used by the UE 101 to decide whether to prioritize the UE 101 for a boost in RAN performance to the UE 101. The ML framework 500 uses the application information and the RAN performance data as inputs for training and inference. In the second embodiment, the method 200 comprises, transmitting the application information to the application boost service node 103. The application information transmitted to application boost service node 103 is used for training the ML framework 500. The application information transmitted to application boost service node 103 is transmitted using an interface 600h. The ML framework 500 is trained in the application boost service node 103. According to the second embodiment, once trained, the ML framework 500 is transmitted to the UE 101.

[0064] The method 200 comprises obtaining, 204, an inference result by inputting the application information and the RAN performance data to the trained ML framework 500. The inference result may indicate whether UE 101 can be prioritized or not during scheduling of radio resources in the communication network 100.

[0065] According to the first embodiment, once trained, the trained ML framework 500 is not be transmitted to the UE 101 but rather comprised in the application boost service node 103. In the first embodiment, obtaining, 204, the inference result may comprise obtaining, 204, the inference result from the application boost service node 103 by inputting the application information and the RAN performance data to the trained ML framework 500 comprised in the application boost service node 103. The UE 101 may obtain the inference result from the application boost service node. The UE 101 may obtain the inference result from the application boost service node using an interface 600g. In order to obtain the interference result, the UE 101 the application information and the RAN performance data are provided as inputs to the application boost service node 103 using the interface 600h.

[0066] The UE 101 communicates the inference result to a boost interface 602 of the application 601. The UE 101 may communicate the inference result to the boost interface 602 of the application 601 using an interface 600d. The boost interface 602 may be in communication with application 601 using an interface 600a.

[0067] The method 200 comprises transmitting, 205, the inference result to the RAN node 102. The method comprises obtaining, 206, radio resources from the RAN node 102 based on the inference result. The inference result may indicate that the UE 101 can be prioritized for a boost in the RAN performance to the UE 101 . The RAN node 102, based on the inference result, may prioritize the UE 101 when scheduling radio resources. When prioritized, the UE 101 may experience improvement in bandwidth, latency, jitter, packet loss, throughput or a combination thereof and thus, boost in its RAN performance.

[0068] The method 300 comprises transmitting RAN performance data to the application boost service node 103. The RAN performance data may be transmitted to the application boost service node 103 using an interface 600f. The RAN performance data is transmitted to the application boost service node 103 for training the ML framework 500. The RAN performance data is transmitted to the application boost service node 103 for inference using the trained ML framework 500.

[0069] Referring to Figure 6, RAN node 102 may comprise a RAN Boost performance center 604. The RAN Boost Performance Center 604 may compute the inference result. The RAN Boost Performance Center 604 may offers prioritization of the UE 101 when it may lead to a boost in the RAN performance for the UE 101. Boosting the RAN performance for the UE 101 may comprise improving bandwidth, latency, jitter, packet loss, throughput that the UE 101 may experience when the UE 101 uses the radio resources offered by the RAN node 102, so that the UE 101 may receive a better QoS compared to other UEs in the RAN.

[0070] An example when prioritization would not lead to an improved or increased or boosted RAN performance may be when the UE 101 is alone in the network 100 but has a poor signal quality (for instance, poor signal quality from being located at an edge of cell area covered by the RAN node 102). In such a case, prioritizing the UE 101 may not lead to a boosted RAN performance and, thus, it may not be suitable to prioritize the UE 101.

[0071] On the other hand, if the UE 101 is in a situation with good signal quality, but the network 100 may be overloaded (for example, there may be too many other users in the network 100), it might be a good opportunity to prioritize the UE 101. In such a scenario, the UE 101 might benefit from being given a higher priority compared to other UEs in the network 100, leading to an increased throughput and a decreased latency and thus, boosted RAN performance for the UE 101 .

[0072] RAN boost performance center 604 may ensure that the proposed invention only offers prioritization of the UE 101 when it may lead to a boost in the RAN performance for the UE 101 using the ML framework 500. The ML framework 500 may receive the application information from the UE 101. The ML framework 500 may receive the application information from the UE 101 using an interface 600b. The ML framework 500 that may be used by the RAN boost performance center 604 may be trained and deployed at the RAN node 102. The ML framework 500 that may be used by the RAN boost performance center 604 may be trained in a cloud environment and deployed at the RAN node 102. The ML framework 500 that may be used by the RAN boost performance center 604 may be trained in the application boost service node 103 and received using an interface 600i and deployed at the RAN node 102 for inference. The ML framework 500 may be trained according to Figure 5.

[0073] Referring to Figure 5, the ML training framework 500 monitors the UE 101 and RAN node 102. The ML training framework 500 receives the application information from the UE 101 . The ML training framework 500 receives the RAN performance data from the RAN node 102. The application information and the RAN performance data may represent measured states of the UE 101 and RAN node 102.

[0074] The ML training framework 500 receives the measured states in a format that the ML training framework 500 is arranged to understand and to predict a future state based on the received measured state. However, such arrangements, are complicated to set up as UE 101 and the RAN 102 may transmit the measured states in different data formats, which then need to be transformed and understood in order for an accurate prediction of a future state. Thus, the measured states may be normalized, for example to a value between 0 and 1 . The normalization may be performed by the ML framework 500.

[0075] The measured states received from the UE 101 and RAN node 103 is inputted to a state predictor 501 , to a data storage 502 and to a mis-prediction detector 503.

[0076] The measured states received from the UE 101 and the RAN node 103 is inputted to the state predictor 501 to obtain predicted states of the UE 101 and the RAN node 102. The state predictor 501 is configured to operate based on machine learning. The state predictor 501 is configured to determine predicted states of the UE 101 and the RAN node 102.

[0077] The measured state refers to the actual or observed state of a system at a given point in time. In the proposed invention, UE 101 and RAN node 102 comprise a system. The system, comprising UE 101 and RAN node 102, is described using state variables, which represent the system's internal state and its relevant parameters. These state variables may comprise the application information and RAN performance data or any other variables that characterizes the behavior of the UE 101 and RAN node 102. Application information that may be obtained at time t may represent the measured state of the UE 101. RAN performance data that may be obtained at time t may represent the measured state of the RAN node 102. The measured state may represent the real-time information about the system's (UE 101 and RAN node 102) current state, typically used as feedback for control or estimation purposes.

[0078] The predicted state, on the other hand, is an estimation or forecast of the system's (UE 101 and RAN node 102) state at a future time (t+1) based on measured state and a mathematical model. In many applications, for example in telecommunication, it may not be possible to directly measure or observe the system's state in real-time, or there might be a time delay in obtaining measurements. In such cases, mathematical models and algorithms are used to predict the system's state based on previous measurements (measured states), system dynamics, and assumptions about its behavior. The predicted state provides an estimate of the system's future behavior, which can be used for planning, control, or decision-making purposes.

[0079] The measured state (application information and RAN performance data at time t) represents the actual observed state of the system (UE 101 and RAN node 102) at a given time t, while the predicted state (application information and RAN performance data at time t+1) refers to the estimated or forecasted state of a system (UE 101 and RAN node 102) at a future time t+1 based on the measured state (application information and RAN performance data at time t) and a mathematical model.

[0080] Times t and t+1 may represent a time instant or a time interval.

[0081] Mathematical models used for predicting state of a system may comprise predictive models. Predictive models can be simple or complex, ranging from basic linear equations to advanced machine learning algorithms and can be used for predicting the future states of the system. Predictive models may comprise supervised or unsupervised or semi-supervised machine learning models or neural networks or reinforcement learning models. The state predictor 501 may comprise a predictive model capable of predicting the future states (future application information and future RAN performance data at time t+1) of the system (UE 101 and RAN node 102) using the measured states (current application information and future RAN performance data at time f).

[0082] The measured states received from the UE 101 and RAN node 103 may be stored in the data storage 502. The predicted states obtained from the state predictor 501 may be stored in the data storage 502.

[0083] The ML framework 500 may comprise the mis-prediction detector 503. The misprediction detector 503 stores the predicted states at any given time and determine whether the stored predicted state is accurate or not. The mis-prediction detector 503 may obtain stored predicted state at any given time from the data storage module 502 and determine whether the stored predicted state is accurate or not. For example, the mis-prediction detector 503 stores the predicted states predicted for a time instant or interval t+1 at a time instant or interval t. The mis-prediction detector 503 obtains actual states or measured states of the system for the time instant or interval t+1 at time instant or interval t+1. The mis-prediction detector 503 compares predicted states, predicted for the time t+1 with the measured states of the system, measured for the time t+1 and obtained at time t+1. By comparing, the mis-prediction detector 503 determines if the predicted states are mis-predictions or not. The mis-prediction detector 503 provides data on mis-predictions to a model performance estimator 504. The model performance estimator 504 quantifies the performance of the state predictor 501. The model performance estimator 504 computes one or more performance metrics. The model performance estimator 504 computes one or more performance metrics of the state predictor 501 , based on the data on mis-predictions provided by the mis-prediction detector 503, to quantify the performance of the state predictor 501. The one or more performance metrics may comprise accuracy, precision, recall, F1- score and / or Mean Squared Error (MSE). The model performance estimator 504 may monitor the computed performance metrics. Each performance metric may have a performance metric threshold associated with it and if the performance metric is below the performance metric threshold associated with it, then the model performance estimator 504 may send a notification to a model trainer 505. If the state predictor 501 has a poor prediction performance, that is, performance metrics of the state predictor 501 being below performance metric thresholds, the model trainer 505 obtains a notification from the model performance estimator 504 and re-train the predictive model comprised in the state predictor 501 . In such a case, the model trainer 505 gathers relevant training data, re-train the predictive model and transmit the re-trained predictive model to the state predictor 501. The model trainer 505 gathers training data from the data storage 502.

[0084] The predicted RAN performance data indicates future characteristics of the RAN. The predicted RAN performance data comprises a future (predicted) value of one or more parameters such as bandwidth, latency, jitter, packet loss, throughput or a combination thereof that the UE 101 may experience when the UE 101 uses the radio resources offered by the RAN node 102. The predicted application information indicates future (predicted) characteristics of the application 601 executed in the UE 101 . For example, the predicted application information may comprise a future (predicted) value of frame rate of the application 601 executed in the UE 101 or a future (predicted) value of latency of the application 601 executed in the UE 101. For example, the predicted application information may comprise a predicted location of the UE 101 in which the application 601 is executed.

[0085] The predicted states may be transmitted to an automatic boost controller 506. The automatic boost controller 506 may combine both the predicted states to determine if the predicted characteristic of the RAN is capable of handling the predicted characteristics of the application 601 . If the predicted characteristic of the RAN is capable of handling the predicted requirements of the application 601 , then the automatic boost controller 506 determines that the UE 101 requires no prioritization during scheduling of radio resources in the RAN. If the predicted characteristic of the RAN is not capable of handling the predicted requirements of the application 601 , then the automatic boost controller 506 determines that the UE 101 requires prioritization during scheduling of radio resources in the RAN. The inference result comprises the determination by the automatic boost controller 506.

[0086] The method 400 comprises transmitting, 404, the trained ML framework 500 to the UE 101. The trained ML framework 500 is executed by the UE boost controller 603. The UE 101 inputs the application information and the RAN performance data (measured states) to the trained ML framework 500 and obtains the inference result using the inputted application information and RAN performance data. The inference result may be provided by the automatic boost controller 506. The inference result may comprise the determination by the automatic boost controller 506. The ML framework 500 may continuously updated by the application boost service node 103. The updated ML framework 500 may be downloaded by the UE boost controller 603 at the start-up of the application 601 .

[0087] To prioritize the UE 101 , there are two necessary conditions to be fulfilled: i) the application 601 must benefit from the prioritization, and ii) it must be possible for the RAN to increase the RAN performance.

[0088] In essence, if prioritization does not lead to increased RAN performance or if the application 601 does not require it, then prioritization of the UE 101 may not be necessary. For instance, if a “video conversation call” application of the UE 101 uses 2 megabits per second, Mbps, bandwidth then prioritizing the UE 101 in order to boost the RAN performance to UE 101 in such a way that the UE 101 receives 100 Mbps bandwidth may not be necessary. Similarly, prioritizing the UE 101 may not be necessary if the application information of the UE 101 indicates that the UE 101 is in a location with poor radio connection.

[0089] For example, a user may start a gaming application (application 601 ) in the UE 101 which is connected to the RAN node 102. Initially, the user may be browsing through the gaming menus of the gaming application for some time. The predicted state of the UE 101 may comprise application information such as data transmission, frame rate and a location of the UE 101 executing the gaming application. Since the user is browsing through the gaming menus and browsing through gaming menus is a less resource and data intensive task in comparison to other resource and data intensive tasks (for example, live streaming) the application information may comprise low values for data transmission and / or for frame rate. As predicted state comprises low values of application information, the predicted state may indicate a low throughput and a low latency requirement for the UE 101 , thus, a reduced demand for the RAN node 102 from the UE 101. The predicted state of the RAN node 102, that is, the predicted RAN performance data (for example, latency and throughput) are sufficient to handle the requirements of the gaming application executed in the UE 101 . So far, there may not be a very high demand neither on latency nor on throughput for this gaming application executed on the UE 101 , so the UE 101 may not be prioritized during the radio resource scheduling. The automatic boost controller 506 may determine that the predicted state of the RAN is capable of satisfying the predicted state of the gaming application, then the automatic boost controller 506 may provide an inference result indicating that UE 101 requires no prioritization during scheduling of radio resources in the RAN.

[0090] Next, the user may decide to launch a “live multiplayer event” in the gaming application, which has a higher demand on both throughput and latency requirements compared to browsing gaming menu. The predicted state for the UE 101 executing the gaming application within which a “live multiplayer event” has been launched comprises high values for application information such as data transmission or frame rate in comparison to values of application information in relation to browsing through gaming menus. In this scenario, the automatic boost controller 506 combines the predicted states of the UE 101 and the RAN node 102 to determine if the predicted state of the RAN node 102 is capable of satisfying the predicted state of the gaming application. If the predicted state of the RAN node 102 is capable of satisfying the predicted state of the UE 101 , then the automatic boost controller 506 provides an inference result indicating that UE 101 requires no prioritization during scheduling of radio resource in the RAN. If the predicted state of the RAN node 102 is not capable of satisfying the predicted requirements of the UE 101 , then the automatic boost controller 506 provides an inference result indicating that UE 101 requires prioritization during scheduling of radio resource in the RAN.

[0091] After a while the user moves around in a city and enters a crowded area. The predicted state for the UE 101 executing the gaming application may comprise application information indicating a location of the UE 101 executing the gaming application. The application information may indicate that the location of the UE 101 is crowded, and the predicted values of data transmission or frame rate are high due the live multiplayer event. The predicted state of the RAN node 102 may not be capable of satisfying the predicted state of the UE 101 , as the predicted state of the RAN node may comprise low values of bandwidth or throughput or latency due to the crowded location of the UE 101. In this scenario, the automatic boost controller 506 provides an inference result indicating that UE 101 requires prioritization during scheduling of radio resources in the RAN. A while after this, the user may still be in the crowded area (and thereby still obtains boosted RAN performance to the UE 101 due to prioritization the UE 101 ), but the “live multiplayer event” may be over and the need to be boosted may no longer be there for UE 101. The predicted state of the UE 101 comprises low values (due to the end of live multiplayer event) of application information such data transmission or frame rate and the predicted state of the RAN node 102 comprises low values (due to the crowded location) of RAN performance data such as bandwidth or latency or throughput. The predicted state of the RAN node 102 is capable of satisfying the predicted state of the UE 101 , then the automatic boost controller 506 provides an inference result indicating that UE 101 requires no prioritization during scheduling of radio resources in the RAN.

[0092] Figure 7 depicts the UE 101. The UE 101 comprises a processing circuitry 702 and a memory 703 coupled with the processing circuitry 702, wherein the memory includes instructions 701 that when executed by the processing circuitry 702 causes the UE 101 to perform operations according to method 200 and embodiments thereof. The processing circuitry is exemplified as a processor 702 in Figure 7. The UE 101 may comprise one or more processors. The instructions are exemplified as a computer program 701 comprising computer-executable instructions in Figure 7. The memory 703 stores the computer program 701 comprising computer-executable instructions. The computer program 701 comprising the computer-executable instructions is executed on the processor 702 causing the UE 101 to perform operations according to method 200 and embodiments thereof. The computer program 701 comprising the computer-executable instructions may be loaded from the memory 703 and executed by the processor 702 causing the UE 101 to perform operations according to method 200 and embodiments thereof. Figure 7 shows a computer program product 704 which comprises a computer readable storage medium on which the computer program 701 is stored. A UE in the form of an Internet of Things (loT) device may be a device for use in one or more application domains, these domains comprising, but not limited to, home, city, wearable technology, extended reality, industrial application, and healthcare. By way of example, the loT device for a home, an office, a building or an infrastructure may be a baking scale, a coffee machine, a grill, a fridge, a refrigerator, a freezer, a microwave oven, an oven, a toaster, a water tap, a water heater, a water geyser, a sauna, a vacuum cleaner, a washer, a dryer, a dishwasher, a door, a window, a curtain, a blind, a furniture, a light bulb, a fan, an air-conditioner, a cooler, an air purifier, a humidifier, a speaker, a television, a laptop, a personal computer, a gaming console, a remote control, a vent, an iron, a steamer, a pressure cooker, a stove, an electric stove, a hair dryer, a hair styler, a mirror, a printer, a scanner, a photocopier, a projector, a hologram projector, a 3D printer, a drill, a hand-dryer, an alarm clock, a clock, a security camera, a smoke alarm, a fire alarm, a connected doorbell, an electronic door lock, a lawnmower, a thermostat, a plug, an irrigation control device, a flood sensor, a moisture sensor, a motion detector, a weather station, an electricity meter, a water meter, and a gas meter.

[0093] By further ways of example, the loT device for use in a city, urban, or rural areas may be connected street lighting, a connected traffic light, a traffic camera, a connected road sign, an air control / monitor, a noise level detector, a transport congestion monitoring device, a transport controlling device, an automated toll payment device, a parking payment device, a sensor for monitoring parking usage, a traffic management device, a digital kiosk, a bin, an air quality monitoring sensor, a bridge condition monitoring sensor, a fire hydrant, a manhole sensor, a tarmac sensor, a water fountain sensor, a connected closed circuit television, a scooter, a hoverboard, a ticketing machine, a ticket barrier, a metro rail, a metro station device, a passenger information panel, an onboard camera, and other connected device on a public transport vehicle.

[0094] As further way of example, the communication loT device may be a wearable device, or a device related to extended reality, wherein the device related to extended reality may be a device related to augmented reality, virtual reality, merged reality, or mixed reality. Examples of such loT devices may be a smart-band, a tracker, a haptic glove, a haptic suit, a smartwatch, clothes, eyeglasses, a head mounted display, an ear pod, an activity monitor, a fitness monitor, a heart rate monitor, a ring, a key tracker, a blood glucose meter, and a pressure meter.

[0095] As further ways of example, the loT device may be an industrial application device wherein an industrial application device may be an industrial unmanned aerial vehicle, an intelligent industrial robot, a vehicle assembly robot, and an automated guided vehicle.

[0096] As further ways of example, the loT device may be a transportation vehicle, wherein a transportation vehicle may be a bicycle, a motor bike, a scooter, a moped, an auto rickshaw, a rail transport, a train, a tram, a bus, a car, a truck, an airplane, a boat, a ship, a ski board, a snowboard, a snow mobile, a hoverboard, a skateboard, rollerskates, a vehicle for freight transportation, a drone, a robot, a stratospheric aircraft, an aircraft, a helicopter and a hovercraft. Figure 8 depicts the RAN node 102. The RAN node 102 comprises a processing circuitry 802; and a memory 803 coupled with the processing circuitry 802, wherein the memory includes instructions 801 that when executed by the processing circuitry 802 causes the RAN node 102 to perform operations according to method 300 and embodiments thereof. The processing circuitry is exemplified as a processor 802 in Figure 8. The RAN node 102 may comprise one or more processors. The instructions are exemplified as a computer program 801 comprising computer-executable instructions in Figure 8. The memory 803 stores the computer program 801 comprising computer-executable instructions. The computer program 801 comprising the computer-executable instructions is executed on the processor 802 causing the RAN node 102 to perform operations according to method 300 and embodiments thereof. The computer program 801 comprising the computer-executable instructions may be loaded from the memory 803 and executed by the processor 802 causing the RAN node 102 to perform operations according to method 300 and embodiments thereof. Figure 8 shows a computer program product 804 which comprises a computer readable storage medium on which the computer program 801 is stored.

[0097] Figure 9 depicts the application boost service node 103. The application boost service node 103 comprises a processing circuitry 902; and a memory 903 coupled with the processing circuitry 902, wherein the memory includes instructions 901 that when executed by the processing circuitry 902 causes the application boost service node 103 to perform operations according to method 400 and embodiments thereof. The processing circuitry is exemplified as a processor 902 in Figure 9. The application boost service node 103 may comprise one or more processors. The instructions are exemplified as a computer program 901 comprising computer-executable instructions in Figure 9. The memory 903 stores the computer program 901 comprising computerexecutable instructions. The computer program 901 comprising the computerexecutable instructions is executed on the processor 902 causing the application boost service node 103 to perform operations according to method 400 and embodiments thereof. The computer program 901 comprising the computer-executable instructions may be loaded from the memory 903 and executed by the processor 902 causing the application boost service node 103 to perform operations according to method 400 and embodiments thereof. Figure 9 shows a computer program product 904 which comprises a computer readable storage medium on which the computer program 901 is stored.

[0098] The methods of the present disclosure may be implemented in hardware, or as software modules running on one or more processors. The methods may also be carried out according to the instructions of a computer program, and the present disclosure also provides a computer readable medium having stored thereon a program for carrying out any of the methods described herein. A computer program embodying the disclosure may be stored on a computer readable medium, or it could, for example, be in the form of a signal such as a downloadable data signal provided from an Internet website, or it could be in any other form.

[0099] It should be noted that the above-mentioned examples illustrate rather than limit the disclosure, and that those skilled in the art will be able to design many alternative embodiments without departing from the scope of the appended claims. The word “comprising” does not exclude the presence of elements or steps other than those listed in a claim, “a” or “an” does not exclude a plurality, and a single processor or other unit may fulfil the functions of several units recited in the claims. Any reference signs in the claims shall not be construed so as to limit their scope.

Claims

CLAIMS1. A method (200) performed by a user equipment, UE, (101 ) for enabling scheduling of a radio resource to the UE (101 ) in a communication network(100), the method (200) comprising: collecting (201 ) application information from an application (601 ) in the UE(101 ); receiving (202) Radio Access Network, RAN, performance data from a RAN node (102); obtaining (204) an inference result by inputting the application information and the RAN performance data to a trained machine learning, ML, framework (500); transmitting (205) the inference result to the RAN node; and obtaining (206) the radio resource from the RAN node based on the inference result.

2. The method (200) of claim 1 , comprising receiving (203) the trained ML framework (500) from an application boost service node (103).

3. The method (200) of claim 1 , wherein obtaining (204) the inference result comprises obtaining (204) the inference result from an application boost service node (103) by inputting the application information and the RAN performance data to the trained ML framework (500) which is comprised in the application boost service node (103).

4. The method (200) of any one of claims 1-3, comprising transmitting the application information to the application boost service node (103).

5. The method (200) of any one of claims 1 -4, wherein the application information comprises data required for determining a state of the application (601 ).

6. The method (200) of any one of claims 1 -5, wherein the application information comprises performance characteristics of the application (601 ).

7. The method (200) of any one of claims 1 -6, wherein the RAN performance data comprises ore or more of a value of bandwidth, a value of latency, a value of jitter, a value of packet loss, a value of throughput or a combination thereof.

8. The method (300) of any one of claims 1-7, wherein the inference result comprises an indication whether to prioritize the UE (101 ) during scheduling of the radio resource in the communication network (100).

9. A method (300) performed by a Radio Access Network, RAN, node (102) for scheduling a radio resource to a user equipment, UE, (101 ) in a communication network (100), the method (300) comprising: transmitting (301 ) RAN performance data to the UE (101 ); receiving (302) an inference result from the UE (101 ) wherein the inference result comprises an indication whether to prioritize the UE (101 ) during scheduling of the radio resource in the communication network (100); and providing (303) the radio resource to the UE (101 ) based on the inference result.

10. The method (300) of claim 9, comprising transmitting RAN performance data to an application boost service node (103).11 . The method (300) of any one of claims 9-10, wherein the RAN performance data comprises one or more of: a value of bandwidth, a value of latency, a value of jitter, a value of packet loss, a value of throughput or a combination thereof.

12. A method (400) performed by an application boost service node (103) for enabling scheduling of a radio resource to a user equipment, UE, (101 ) in a communication network (100), the method (400) comprising:receiving (401 ) Radio Access Network, RAN, performance data from a RAN node (102); receiving (402) application information from the UE (101 ); and training (403) a machine learning, ML, framework (500) based on the received RAN performance data and the received application information to predict whether to prioritize the UE (101 ) when scheduling the radio resource in the communication network (100).

13. The method (400) of claim 12, comprising transmitting (404) the trained ML framework (500) to the UE (101 ).

14. The method (400) of claim 12, comprising providing an inference result by inputting the application information and the RAN performance data to the trained ML framework (500).

15. The method (400) of any one of claims 12-14, wherein the application information comprises data required for determining a state of the application (601 ).

16. The method (400) of any one of claims 12-15, wherein the application information comprises performance characteristics of the application (601 ).

17. The method (400) of any one of claims 12-16, wherein the RAN performance data comprises one or more of: a value of bandwidth, a value of latency, a value of jitter, a value of packet loss, a value of throughput or a combination thereof.

18. The method (400) of any one of claims 12-17, wherein the inference result comprises an indication whether to prioritize the UE (101 ) during scheduling of the radio resource in the communication network (100).

19. The method (400) of any one of claims 12-18, wherein training the ML framework (500) comprises obtaining, by a state predictor (501 ), predicted states of a system comprising the UE (101 ) and the RAN node (102).

20. The method (400) of claim 19, wherein the state predictor (501 ) comprises a predictive model.

21. The method (400) of any one of claims 12-20, wherein training the ML framework (500) comprises determining, by a mis-prediction detector (502), whether the predicted states are mis-predictions or not.

22. The method (400) of any one of claims 12-21 , wherein training the ML framework (500) comprises: computing, by a model performance estimator (504), one or more performance metrics of the state predictor (501 ); monitoring, by the model performance estimator (504), the one or more computed performance metrics of the state predictor (501 ); and transmitting, by the model performance estimator (504), a notification to a model trainer (505) if the one or more performance metrics are below one or more performance metric thresholds.

23. The method (400) of any one of claims 12-22, wherein training the ML framework (500) comprises: receiving the notification from the model performance estimator (504); and re-training, by the model trainer (505), predictive model comprised in the state predictor (501 ) upon receiving the notification.

24. The method (400) of any one of claims 12-23, wherein training the ML framework (500) comprises determining, by an automatic boost controller (506), whether the UE 101 requires prioritization during scheduling of radio resources.

25. A user equipment, UE, (101 ) for enabling scheduling of a radio resource to the UE (101 ) in a communication network (100), the UE (101 ) comprising processing circuitry (802) configured to cause the UE (101 ) to: collect application information from an application (601 ) in the UE (101 ); receive (202) Radio Access Network, RAN, performance data from a RAN node (102); obtain an inference result by inputting the application information and the RAN performance data to a trained machine learning, ML, framework (500); transmit the inference result to the RAN node (102); and obtain the radio resource from the RAN node (102) based on the inference result.

26. The UE (101 ) as claimed in claim 25, wherein the processing circuitry (702) is further configured to cause the UE (101 ) to perform a method according to any one of claims 2 to 8.

27. A Radio Access Network, RAN, node (102) for scheduling a radio resource to a user equipment, UE, (101 ) in a communication network (100), the RAN node (102) comprising processing circuitry (802) configured to cause the RAN node (102) to: transmit RAN performance data to the UE (101 ); receive an inference result from the UE wherein the inference result comprises an indication whether to prioritize the UE (101 ) during scheduling of the radio resource in the communication network (100); and provide the radio resource to the UE (101 ) based on the inference result.

28. The RAN node (102) as claimed in claim 27, wherein the processing circuitry (802) is further configured to cause the RAN node (102) to perform a method according to any of claims 10 to 11 .

29. An application boost service node (103) for enabling scheduling of a radio resource to a user equipment, UE, (101 ) in a communication network (100), the application boost service node (103) comprising processing circuitry (802) configured to cause the application boost service node (103) to: receive RAN performance data from a RAN node (102); receive application information from the UE (101 ); and train a machine learning, ML, framework (500), based on the received RAN performance data and the received application information, to predict whether to prioritize the UE (101 ) when scheduling the radio resource in the communication network (100).

30. The application boost service node (103) as claimed in claim 29, wherein the processing circuitry (902) is further configured to cause the application boost service node (103) to perform a method according to any one of claims 13 to 24.31 . A computer program (701 ), comprising instructions which when executed on a processor (702) of a UE (101 ) causes the UE (101 ) to perform a method according to any one of claims 1 to 8.

32. A computer program product (704) which comprises a computer readable storage medium on which a computer program (701 ) according to claim 31 is stored.

33. A computer program (801 ), comprising instructions which when executed on a processor (802) of a RAN node (102) causes the RAN node (102) to perform a method according to any one of claims 9 to 11 .

34. A computer program product (804) which comprises a computer readable storage medium on which a computer program (801 ) according to claim 33 is stored.

35. A computer program (901 ), comprising instructions which when executed on a processor (902) of an application boost service node (103) causes the application boost service node (103) to perform a method according to any one of claims 12 to 24.

36. A computer program product (904) which comprises a computer readable storage medium on which a computer program (801 ) according to claim 35 is stored.

Images