Network node, server, wireless device, and computer-readable storage medium

Abstract

According to the present invention, a network node of a mobile communication network comprises: a determining means that, in response to receiving a request from a server to determine the availability of sidelink communication in a plurality of wireless devices belonging to a group, determines a pair of two wireless devices that are capable of sidelink communication from among the plurality of wireless devices, and determines whether or not a first learning model with a higher performance level than a second learning model is permitted to be transmitted and received between the two wireless devices of the pair by sidelink communication; and a transmitting means that transmits a determination result by the determining means to the server.

Description

Network node, server, wireless device and computer-readable storage medium

[0001] The present disclosure relates to a technique for transmitting and receiving learning models between wireless devices.

[0002] Non-Patent Document 1 discloses various applications that utilize learning models generated by machine learning in wireless devices (WDs) that can access a mobile communication network. Furthermore, Non-Patent Document 1 also discloses transmitting and receiving learning models between WDs that have been grouped in advance.

[0003] Non-Patent Document 2 discloses a system for proximity services (ProSe) that transmit and receive data between WDs. According to Non-Patent Document 2, two WDs can transmit and receive data via a "sidelink," a wireless link connecting the two WDs, without going through a mobile communication network. Furthermore, Non-Patent Document 2 also discloses switching communication between two WDs from communication via the sidelink to communication via a mobile communication network in response to deterioration of the quality of the sidelink, etc. In the following description, communication between two WDs via the sidelink connecting the two WDs is referred to as "sidelink communication," and communication between the two WDs via a mobile communication network is referred to as "communication via a network."

[0004] Non-Patent Document 3 discloses that a learning model generated by machine learning includes information about the learning data used to generate the learning model. Therefore, if the learning data includes privacy information such as WD location information, the learning model also includes the privacy information. Note that there is generally a trade-off between the privacy information included in the learning model and the performance level of the learning model. In other words, the less privacy information included in the learning model, the lower the performance level of the learning model.

[0005] 3GPP TR 22.876 V19.0.0, June 2023 3GPP TS 23.304 V18.2.0, June 2023 Nasr, Milad, Reza Shokri, and Amir Houmansadr, "Comprehensive privacy analysis of deep learning", Proceedings of IEEE Symposium on Security and Privacy, 2018

[0006] For example, when transmitting and receiving a first learning model that has a high performance level but contains a lot of privacy information between WDs, it is preferable from a security perspective to use side link communication rather than communication via the network. On the other hand, when side link communication is not available and communication via the network is used, it is preferable to transfer a second learning model that has a lower performance level than the first learning model but contains less privacy information than the first learning model.

[0007] However, while the first learned model is being transferred via sidelink communication, the quality of the sidelink communication may deteriorate, causing a switch to communication via the network. Therefore, when the first WD transmits a learned model to the second WD capable of sidelink communication, it is preferable to introduce a mechanism that can determine whether to transmit the first learned model or the second learned model, taking into account the possibility of switching to communication via the network due to quality deterioration of the sidelink communication.

[0008] According to one aspect of the present disclosure, a network node of a mobile communication network is provided with: a determination means for determining, in response to receiving a determination request from a server regarding the availability of sidelink communication among a plurality of wireless devices belonging to a group, whether a pair of two wireless devices among the plurality of wireless devices that are capable of sidelink communication is permitted to transmit and receive a first learning model having a higher performance level than a second learning model between the two wireless devices of the pair via sidelink communication; and a transmission means for transmitting the determination result by the determination means to the server.

[0009] According to the present disclosure, a wireless device can determine the performance level of a learned model transmitted and received via sidelink communication.

[0010] Other features and advantages of the present invention will become apparent from the following description taken in conjunction with the accompanying drawings, in which the same or similar elements are designated by the same reference numerals.

[0011] A system configuration diagram according to an embodiment. A sequence diagram according to an embodiment. A sequence diagram according to an embodiment. A diagram showing an example of a determination result by an NWDAF. A diagram showing an example of a determination result by an NWDAF. A configuration diagram of a network node according to an embodiment. A configuration diagram of a ProSe AS according to an embodiment. A configuration diagram of a WD according to an embodiment.

[0012] Hereinafter, the embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the scope of the invention as claimed, and not all combinations of features described in the embodiments are necessarily essential to the invention. Two or more of the features described in the embodiments may be arbitrarily combined. Furthermore, the same reference numerals are used for the same or similar components, and redundant explanations will be omitted.

[0013] FIG. 1 is a system configuration diagram according to an embodiment. A core network 100 of a mobile communication network includes various network functions (NFs). A network data analysis function (NWDAF) 1 in FIG. 1 is an NF that performs processing related to a machine learning learning model defined by 3GPP. NFs 2 in FIG. 2 is a collective term for other NFs different from NWDAF 1. NFs 2 may include functions such as managing the location of a WD, measuring the movement speed of a WD, and enabling a server of an external data network (DN) connected to the core network 100 to access an NF of the core network 100.

[0014] Furthermore, the Proximity Service (ProSe) Application Server (AS) 3 is a device located in an external DN connected to the core network 100, such as the Internet, and is operated by a service provider that provides proximity services. The multiple WDs 5 shown in FIG. 1 are wireless devices of users who use the proximity service and belong to the same group that can transmit and receive learning models. Each WD 5 has the function of communicating with a DN server, such as a ProSe AS 3, via a mobile communication network and the function of directly communicating with other WDs 5 via sidelink communication. Although five WDs 5 are shown in FIG. 1, the number of WDs 5 belonging to the same group can be any number greater than or equal to two. In the following description, when distinguishing between the five WDs 5, they will be referred to as WD #1 to WD #5.

[0015] 2 and 3 are sequence diagrams according to an embodiment. In S10, the ProSe AS 3 requests the NWDAF 1 to determine which set of WDs 1 to 5 belonging to the same group is capable of sidelink communication. In S10, the message transmitted from the ProSe AS 3 to the NWDAF 1 may be transmitted directly from the ProSe AS 3 to the NWDAF 1 or may be transmitted via another NF in the core network 10. In S11, the NWDAF 1 collects information about the WDs 1 to 5 for which the determination is requested, such as their locations, moving speeds, and / or moving directions, from the NFs 2. In S12, the NWDAF 1 determines which set is capable of sidelink communication based on the information collected in S11, such as the current location of WD 5.

[0016] FIG. 4A shows an example of the determination result in S12. According to FIG. 4A, two sets, set #1 and set #2, are determined. Set #1 includes WD #1, WD #2, and WD #3, and set #2 includes WD #4 and WD #5. In this embodiment, when transmitting and receiving data between two pairs of WDs, the other WDs are not used as relays. That is, any two WDs 5 among WD #1, WD #2, and WD #3 in set #1 can directly communicate via sidelink without using other WDs 5. Therefore, according to the determination result in FIG. 4A, the pairs capable of sidelink communication are (WD #1, WD #2), (WD #1, WD #3), (WD #2, WD #3), and (WD #4, WD #5). Thus, the process of S12 also determines the pairs of WDs 5 capable of sidelink communication.

[0017] 4B shows another example of the determination result at S12. According to FIG. 4B, one set #1 is determined. Set #1 includes WD #1, WD #2, WD #3, and WD #4. Therefore, according to the determination result of FIG. 3B, the pairs of WD5 capable of sidelink communication are (WD #1, WD #2), (WD #1, WD #3), (WD #1, WD #4), (WD #2, WD #3), (WD #2, WD #4), and (WD #3, WD #4).

[0018] In S13, NWDAF1 determines, based on predetermined criteria, which of two or more WD5s included in a set capable of sidelink communication is likely to switch to communication via the network within a predetermined period of time, based on the information collected in S11, for example, information on the movement speed of WD5.

[0019] For example, the NWDAF1 can determine, among two or more WDs 5 included in a set capable of sidelink communication, WDs 5 whose moving speed is greater than a predetermined speed as WDs 5 that are likely to switch to communication via the network within a predetermined period. This is based on the idea that sidelink communication can be performed stably between any two WDs 5 in a set of multiple WDs 5 that are close to each other and are substantially stationary, but that sidelink communication cannot be performed within a predetermined period for WDs 5 that are moving faster than a predetermined speed, even if sidelink communication is possible when the NWDAF1 determines that sidelink communication is possible.

[0020] Furthermore, if the NWDAF 1 can acquire information on the movement direction of each WD 5, the NWDAF 1 can also take the movement direction into consideration to determine which WD 5 is likely to switch to communication via the NW within a predetermined period of time. In other words, the NWDAF 1 can be configured to determine, based on the movement speed and movement direction, which WD 5 has a high possibility of switching to communication via the NW if the distance to other WD 5 in the set exceeds a threshold within a predetermined period of time.

[0021] 4A also shows an example of the determination result in S13. According to FIG. 4A , it is determined that WD #3 included in set #1 is likely to switch to communication via the network within the predetermined period. That is, the NWDAF1 determines that the pair (WD #1, WD #3) and the pair (WD #2, WD #3) are capable of sidelink communication but are likely to switch to communication via the network within the predetermined period. On the other hand, the NWDAF1 determines that the pair (WD #1, WD #2) and the pair (WD #4, WD #5) are capable of sidelink communication but are unlikely to switch to communication via the network within the predetermined period. In this way, the process of S13 also determines the likelihood of switching to communication via the network within the predetermined period for each of the two WD5 pairs capable of sidelink communication.

[0022] 4B also shows an example of the determination result in S13. According to FIG. 4B , it is determined that WD #3 and WD #4 included in set #1 are likely to switch to communication via the network. That is, the NWDAF1 determines that the pair (WD #1, WD #3), the pair (WD #1, WD #4), the pair (WD #2, WD #3), and the pair (WD #3, WD #4) are capable of sidelink communication but are likely to switch to communication via the network within a predetermined period. On the other hand, the NWDAF1 determines that the pair (WD #1, WD #2) is capable of sidelink communication but is unlikely to switch to communication via the network within a predetermined period.

[0023] In S14, the NWDAF 1 transmits the determination results of S12 and S13, that is, the information shown in FIG. 4A or 4B, to the ProSe AS 3.

[0024] In S20 to S24 of FIG. 3, the ProSe AS 3 notifies each of WD #1 to WD #5 of policy information regarding sidelink communication. The policy information includes set information indicating WD #5 belonging to the same set and switch information indicating whether or not there is a high probability of switching to communication via the network within a predetermined period. The set information corresponds to the information indicated by "WD belonging to the set" in FIGS. 4A and 4B. The switch information corresponds to the information indicated by "WD with high probability of switching" in FIGS. 4A and 4B. For example, the switch information may be information indicating, with a first value, that there is a high probability of switching to communication via the network within a predetermined period and with a second value, that there is a low probability of switching to communication via the network within a predetermined period.

[0025] First, the policy information notified to each WD5 when the determination results in S12 and S13 are as shown in Fig. 4A will be described. The set information for WD#1 and WD#2 indicates that WD#1 to WD#3 are the same set, and the switch information for WD#1 and WD#2 indicates that there is a low possibility of switching to communication via the NW within a predetermined period. The set information for WD#3 indicates that WD#1 to WD#3 are the same set, and the switch information for WD#3 indicates that there is a high possibility of switching to communication via the NW within a predetermined period. The set information for WD#4 and WD#5 indicates that WD#4 and WD#5 are the same set, and the switch information for WD#4 and WD#5 indicates that there is a low possibility of switching to communication via the NW within a predetermined period.

[0026] Next, the policy information notified to each WD5 when the determination results in S12 and S13 are as shown in FIG. 4B will be described. The set information for WD#1 and WD#2 indicates that WD#1 to WD#4 are the same set, and the switch information for WD#1 and WD#2 indicates that there is a low possibility of switching to communication via the network within a predetermined period. The set information for WD#3 and WD#4 indicates that WD#1 to WD#4 are the same set, and the switch information for WD#3 and WD#4 indicates that there is a high possibility of switching to communication via the network within a predetermined period. The set information for WD#5 indicates that it does not belong to any set, that is, that sidelink communication with any WD is not possible. Therefore, the switch information for WD#5 may be a null value.

[0027] Next, the transmission of the learning model based on the policy information will be described. In the following description, the WD5 that transmits the learning model will be referred to as the "sending WD," and the WD5 that receives the learning model will be referred to as the "receiving WD." The sending WD is assumed to hold a first learning model and a second learning model with a lower performance level than the first learning model. The second learning model can be generated by modifying the first learning model.

[0028] The transmitting WD determines the receiving WD to which it can transmit the training model via sidelink communication based on the received set information. When establishing a sidelink with the receiving WD, the transmitting WD and receiving WD exchange the received switch information. If the switch information of both the transmitting WD and receiving WD indicates a low probability of switching to communication via a network within a predetermined period, the transmitting WD transmits the first training model to the receiving WD. On the other hand, if the switch information of at least one of the transmitting WD and receiving WD indicates a high probability of switching to communication via a network within a predetermined period, the transmitting WD transmits the second training model to the receiving WD. In this way, the performance level of the training model to be transmitted and received via sidelink communication is determined by combining the switch information held by the transmitting WD and the switch information held by the receiving WD. In other words, the switch information is also judgment information that allows the WD5 to determine, for each pair of two WD5s capable of side link communication, whether the first learning model is permitted to be transmitted and received via side link communication between the two WD5s of the pair, or whether only the second learning model, which has a lower performance level than the first learning model, is permitted to be transmitted and received.

[0029] Therefore, in the case of the determination result of Figure 4A, the second learning model is transmitted for the pair (WD#1, WD#3) and the pair (WD#2, WD#3), and the first learning model is transmitted for the pair (WD#1, WD#2) and the pair (WD#4, WD#5). Also, in the case of the determination result of Figure 4B, the second learning model is transmitted for the pair (WD#1, WD#3), the pair (WD#1, WD#4), the pair (WD#2, WD#3), and the pair (WD#3, WD#4), and the first learning model is transmitted for the pair (WD#1, WD#2).

[0030] For example, a WD5 having only the second learning model may be configured to be unable to become a transmitting WD. Alternatively, a WD5 having only the second learning model may be configured to be able to become a transmitting WD. In this case, a WD5 having only the second learning model always transmits the second learning model.

[0031] The ProSe AS 3 may be configured to periodically repeat the processes of Figures 2 and 3. Alternatively, the ProSe AS 3 may be configured to start the processes of Figures 2 and 3 when a request is received from one of the WDs 5 belonging to the group.

[0032] As described above, in response to the determination request (S10) from the ProSe AS 3, the NWDAF 1 determines whether a pair of two WDs 5 capable of sidelink communication is permitted to transmit and receive the first training model instead of the second training model in communication between the pair of WDs 5, and notifies the ProSe AS 3 of the determination result. The ProSe AS 3 then notifies the WD 5 of information indicating other WDs 5 in the same group that are capable of sidelink communication with the WD 5, and information used by the WD 5 to determine whether the first training model can be transmitted and received when transmitting and receiving a training model with the other WD 5. This configuration can reduce the possibility that the first training model, which contains a large amount of privacy information, is transmitted and received via the mobile communication network.

[0033] Note that information indicating other WDs 5 in the same group that are capable of sidelink communication with WD 5 corresponds to the set information described above. Furthermore, information used by a WD 5 to determine whether it can transmit and receive the first learning model with another WD 5 corresponds to the switch information described above. However, the information indicating other WDs 5 in the same group that are capable of sidelink communication with WD 5 and the information used by a WD 5 to determine whether it can transmit and receive the first learning model with another WD 5 are not limited to the formats of the set information and switch information described above.

[0034] For example, information indicating other WDs 5 in the same group capable of sidelink communication and information used by the WDs 5 to determine whether the first learning model can be transmitted or received with the other WDs 5 can be notified to the WDs in the format of (other WDs, whether transmission or reception of the first learning model is possible). Specifically, in the case of the determination result of FIG. 4A, (WD #2, possible) and (WD #3, not possible) can be notified to WD #1, and in the case of the determination result of FIG. 4B, (WD #2, possible), (WD #3, not possible), and (WD #4, not possible) can be notified to WD #1. In this case, the information used by the WDs 5 to determine whether the first learning model can be transmitted or received with the other WDs 5 directly indicates whether the first learning model can be transmitted or received with the other WDs 5.

[0035] In this way, the information indicating the other WDs 5 in the same group that are capable of sidelink communication with the WD 5 can be any pair information indicating a pair between the WD 5 and the other WD 5. Furthermore, the information used by the WD 5 to determine whether the WD 5 can transmit and receive the first learning model to and from the other WD 5 can be any form of determination information. The set information is one form of pair information, and the switch information is one form of determination information.

[0036] 5 is a configuration diagram of a network node 6 according to this embodiment. The network node 6 may be, for example, a device that implements the functions of an NWDAF 1. In response to a reception request from a ProSe AS 3 by a receiver 63 to determine whether sidelink communication is possible among multiple WDs 5 belonging to the same group, a collector 61 collects information, such as information on the positions and moving speeds of these WDs 5, from the NFs 2, and a determiner 62 determines a pair of two WDs 5 that are capable of sidelink communication.

[0037] Furthermore, the determination unit 62 determines, for each pair of two WDs 5 capable of side link communication, the possibility of switching from side link communication to communication via the network within a predetermined period, and determines whether transmission and reception of the first learned model is permitted between the two WDs 5 of the pair, or whether only transmission and reception of the second learned model is permitted. The transmission unit 64 transmits the determination result by the determination unit 62 to the ProSe AS 3.

[0038] 6 is a configuration diagram of a ProSe AS 3 according to this embodiment. The storage unit 32 stores information about WDs 5 belonging to the same group. The request unit 31 requests the NWDAF 1 to determine whether sidelink communication is possible among multiple WDs 5 belonging to the same group via the transmission unit 34. The reception unit 33 receives the determination result from the NWDAF 1 in response to the request by the request unit 31. Based on the determination result received by the reception unit 33 from the NWDAF 1, the transmission unit 34 transmits to each WD 5 pair information indicating other WDs 5 in the same group that are capable of sidelink communication with the WD 5, and determination information used by the WD 5 to determine whether the WD 5 can transmit and receive the first learning model to the other WDs 5.

[0039] 7 is a configuration diagram of a WD 5 according to this embodiment. The storage unit 52 stores a learning model. The sidelink communication unit 51 performs sidelink communication with other WDs 5. The transmitter 55 and receiver 54 perform communication with the mobile communication network. For example, the receiver 54 receives pairing information and determination information from the ProSe AS 3. The controller 53 determines other WDs 5 with which sidelink communication is possible based on the pairing information, and determines whether the first learning model can be transmitted and received based on the determination information when transmitting and receiving a learning model via sidelink communication with other WDs 5.

[0040] When the determination information is switch information, the control unit 53 may determine that the first learning model can be transmitted and received via sidelink communication if both the received determination information and the determination information received by another WD 5 performing sidelink communication have the same first value. On the other hand, the control unit 53 determines that the first learning model cannot be transmitted and received via sidelink communication if at least one of the received determination information and the determination information received by another WD 5 performing sidelink communication does not have the first value. In other words, the control unit 53 determines that the first learning model cannot be transmitted and received via sidelink communication if at least one of the received determination information and the determination information received by another WD 5 performing sidelink communication has a second value. On the other hand, the control unit 53 determines that the first learning model can be transmitted and received via sidelink communication if both the received determination information and the determination information received by another WD 5 performing sidelink communication do not have the second value.

[0041] The present disclosure also provides a computer program, which, when executed by one or more processors in an apparatus having one or more processors, causes the apparatus to operate as the network node 6, ProSe AS 3, or WD 5, and a computer-readable storage medium storing the computer program. The present disclosure also provides a method performed by the network node 6, ProSe AS 3, or WD 5 described with reference to Figures 2 and 3, a computer program causing an apparatus having one or more processors to execute the method, and a computer-readable storage medium storing the computer program.

[0042] This application claims priority based on Japanese Patent Application No. 2023-169507, filed September 29, 2023, the entire contents of which are incorporated herein by reference.

Claims

1. A network node of a mobile communications network, comprising: a determination means for determining, in response to receiving a determination request from a server as to whether sidelink communication is possible for a plurality of wireless devices belonging to a group, whether a pair of two wireless devices among the plurality of wireless devices that are capable of sidelink communication is permitted to transmit and receive a first learning model having a higher performance level than a second learning model between the two wireless devices of the pair via sidelink communication; and a transmission means for transmitting a determination result made by the determination means to the server.

2. The network node of claim 1, wherein the determination means determines the pair of two wireless devices capable of sidelink communication based on the positions of the multiple wireless devices, and determines whether or not the first learning model is permitted to be transmitted and received between the two wireless devices of the pair via sidelink communication based on the movement speeds of the multiple wireless devices.

3. A network node according to claim 1 or 2, wherein the server is a server for providing a proximity service (ProSe).

4. A network node according to any one of claims 1 to 3, wherein the network node implements a network data analysis function.

5. A computer-readable storage medium storing a program that, when executed by one or more processors of an apparatus having one or more processors, causes the apparatus to function as a network node according to any one of claims 1 to 4.

6. A server capable of communicating with a network node of a mobile communication network, comprising: a requesting means for requesting the network node to determine whether sidelink communication is possible for a plurality of wireless devices belonging to a group; and a receiving means for receiving from the network node information indicating a pair of two wireless devices among the plurality of wireless devices that are capable of sidelink communication, and whether a first learning model having a higher performance level than a second learning model is permitted to be transmitted and received between the two wireless devices of the pair via sidelink communication.

7. The server described in claim 6, further comprising a transmitting means for transmitting to each of the plurality of wireless devices information indicating other wireless devices with which sidelink communication is possible, and determination information for the wireless device to determine whether or not the first learning model is permitted to be transmitted and received between the other wireless devices and the sidelink communication with the other wireless devices.

8. The server of claim 7, wherein, when both the determination information received by the wireless device and the determination information received by the other wireless device are a first value, the wireless device of each of the multiple wireless devices determines that transmission and reception of the first learning model between the wireless device and the other wireless device is permitted through the sidelink communication with the other wireless device, and when at least one of the determination information received by the wireless device and the determination information received by the other wireless device is not the first value, the wireless device determines that transmission and reception of the first learning model between the wireless device and the other wireless device is not permitted through the sidelink communication with the other wireless device.

9. A server according to any one of claims 6 to 8, wherein the network node implements a network data analysis function.

10. The server according to any one of claims 6 to 9, wherein the server is a server for providing a proximity service (ProSe).

11. A computer-readable storage medium having stored thereon a program that, when executed by one or more processors of an apparatus having one or more processors, causes the apparatus to function as a server according to any one of claims 6 to 10.

12. A wireless device of a mobile communication network, comprising: a receiving means for receiving, from a server of a data network connected to the mobile communication network, information indicating other wireless devices that belong to the same group as the wireless device and are capable of sidelink communication with the wireless device, and determination information for the wireless device to determine whether or not a first learning model having a higher performance level than a second learning model is permitted to be transmitted and received between the other wireless device and the other wireless device through the sidelink communication with the other wireless device; a sidelink communication means for performing sidelink communication with the other wireless device; and a determination means for determining, according to the determination information, which of the first learning model or the second learning model to transmit or receive in the sidelink communication with the other wireless device.

13. The wireless device of claim 12, wherein the determination means determines that, when both the determination information received by the wireless device and the determination information received by the other wireless device are a first value, transmission and reception of the first learning model between the other wireless device and the wireless device is permitted through the side link communication with the other wireless device, and determines that, when at least one of the determination information received by the wireless device and the determination information received by the other wireless device is not the first value, transmission and reception of the first learning model between the other wireless device and the wireless device is not permitted through the side link communication with the other wireless device.

14. A wireless device according to claim 12 or 13, wherein the server is a server for providing a proximity service (ProSe).

15. A computer-readable storage medium having stored thereon a program that, when executed by one or more processors of an apparatus having one or more processors, causes the apparatus to function as a wireless device according to any one of claims 12 to 14.

Images