Apparatus, methods, and media for handover in hierarchical federated learning networks

Abstract

The invention relates to an apparatus, a method and a medium for handover in a hierarchical federated learning network. There is provided a network-side electronic device for federated learning, comprising processing circuitry configured to: determine a model aggregation time and a remaining service time for a user equipment, where the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node; under the condition that the model aggregation time and the residual service time meet preset conditions, making a switching decision aiming at the user equipment; and transmitting the handover decision for the user equipment.

Description

Apparatus, method and medium for switching in hierarchical federated learning networks

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to Chinese patent application number 202210936728.X, filed on August 5, 2022, entitled “Switching in Hierarchical Federated Learning Networks,” the entire contents of which are incorporated herein by reference. Technical Field

[0003] The present disclosure relates to apparatus, methods, and media for switching in a hierarchical federated learning network. Background Art

[0004] In a federated learning network, each user device (UE) accesses a base station via a wireless channel and uploads the locally learned model to a server via the base station. The server aggregates the model and then distributes it to each UE via the base station. However, due to the mobility of the UE and changes in the wireless channel between the UE and the base station, the UE may need to switch when executing a federated learning task.

[0005] Summary of the Invention

[0006] The present disclosure provides apparatus, methods, and media for switching in a hierarchical federated learning network.

[0007] According to one aspect of the present disclosure, a network-side electronic device for federated learning is provided, including a processing circuit, wherein the processing circuit is configured to: determine a model aggregation time and a remaining service time for a user device, wherein the user device is directly connected to the network or indirectly connected to the network via an intermediate node; make a switching decision for the user device if the model aggregation time and the remaining service time meet a predetermined condition; and send the switching decision for the user device.

[0008] According to another aspect of the present disclosure, an intermediate node-side electronic device for federated learning is provided, including a processing circuit, wherein the processing circuit is configured to: receive a switching decision for a user equipment from a network, wherein the switching decision is made when a model aggregation time and a remaining service time for the user equipment meet predetermined conditions, and the user equipment is indirectly connected to the network via the intermediate node; and send the switching decision to the user equipment.

[0009] According to another aspect of the present disclosure, a user equipment side electronic device for federated learning is provided, including a processing circuit, wherein the processing circuit is configured to: receive a switching decision for a user equipment from a network, wherein the switching decision is made when a model aggregation time and a remaining service time for the user equipment meet predetermined conditions, and the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node; and perform switching based on the switching decision.

[0010] According to another aspect of the present disclosure, a network-side method for federated learning is provided, including: determining a model aggregation time and a remaining service time for a user device, wherein the user device is directly connected to the network or indirectly connected to the network via an intermediate node; making a switching decision for the user device if the model aggregation time and the remaining service time meet a predetermined condition; and sending the switching decision for the user device.

[0011] According to another aspect of the present disclosure, an intermediate node-side method for federated learning is provided, comprising: receiving a switching decision for a user equipment from a network, wherein the switching decision is made when a model aggregation time and a remaining service time for the user equipment meet predetermined conditions, and the user equipment is indirectly connected to the network via the intermediate node; and sending the switching decision to the user equipment.

[0012] According to another aspect of the present disclosure, a user device-side method for federated learning is provided, comprising: receiving a switching decision for a user device from a network, wherein the switching decision is made when a model aggregation time and a remaining service time for the user device meet predetermined conditions, and the user device is directly connected to the network or indirectly connected to the network via an intermediate node; and performing switching based on the switching decision.

[0013] According to yet another aspect of the present disclosure, a non-transitory computer-readable storage medium is provided, on which program instructions are stored. When the program instructions are executed by a processor, the processor is caused to perform the method of the present disclosure.

[0014] According to yet another aspect of the present disclosure, a computer program product is provided, comprising program instructions, which, when executed by a processor, cause the processor to perform the method of the present disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] A better understanding of the present disclosure may be obtained when the following detailed description of the embodiments is considered in conjunction with the accompanying drawings. The same or similar reference numerals are used in the various drawings to represent the same or similar parts. The accompanying drawings, together with the following detailed description, are incorporated into and form a part of this specification and are used to illustrate the embodiments of the present disclosure and to explain the principles and advantages of the present disclosure.

[0016] Figure 1 shows an exemplary structure of a traditional federated learning network.

[0017] FIG2 shows an exemplary structure of a hierarchical federated learning network according to an embodiment of the present disclosure.

[0018] FIG3 illustrates an exemplary federated learning process of a hierarchical federated learning network according to an embodiment of the present disclosure.

[0019] 4A to 4C illustrate different handover scenarios for a local UE.

[0020] FIG5 shows an exemplary handover process of a local UE according to an embodiment of the present disclosure.

[0021] 6A to 6E illustrate situations where the remaining service time of a local UE satisfies different conditions.

[0022] FIG7 shows an exemplary handover process of a global UE according to an embodiment of the present disclosure.

[0023] 8A to 8C illustrate situations where the remaining service time of the global UE meets different conditions.

[0024] Figure 9 shows the 5G core network SBA architecture and some of its network functions (NFs).

[0025] FIG. 10 is a block diagram illustrating an example of a schematic configuration of a computing device to which the technology of the present disclosure may be applied.

[0026] FIG11 is a block diagram illustrating a first example of a schematic configuration of a gNB to which the technology of the present disclosure may be applied.

[0027] FIG12 is a block diagram illustrating a second example of a schematic configuration of a gNB to which the technology of the present disclosure may be applied.

[0028] FIG. 13 is a block diagram illustrating an example of a schematic configuration of a smartphone to which the technology of the present disclosure can be applied.

[0029] FIG. 14 is a block diagram showing an example of a schematic configuration of a car navigation device to which the technology of the present disclosure can be applied. DETAILED DESCRIPTION

[0030] Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. Note that, in this specification and the drawings, structural elements that have substantially the same function and structure are denoted with the same reference numerals, and repeated explanation of these structural elements is omitted.

[0031] FIG1 shows an exemplary structure of a traditional federated learning network. In a traditional federated learning network, UEs 1201, 1202, 1203, ... K-1 , 120 K Directly connect to the server 110 through a base station (not shown), and upload the local model to the server 110 for global aggregation. The specific process is as follows.

[0032] First, UE 120 accesses server 110 and obtains the initial global model w through downlink transmission. 0 :

[0033] in, represents the initial local model of the k-th UE.

[0034] Then, UE 120 uses the locally stored data for learning and completes the local iteration of the r+1th local model update:

[0035] in, represents the local model of the k-th UE after the r-th iteration, represents the gradient of the kth UE, η represents the learning rate, represents the loss function of the kth UE.

[0036] Then, UE 120 transmits the learned local model to the uplink. or gradient Upload to server 110. Server 110 aggregates the local models collected from each UE to complete the update of the global model:

[0037] where p k is the weight of the local model from each UE, usually set to D i represents the dataset of the i-th UE, |D i |Indicates the size of the dataset.

[0038] Finally, the server 110 sends the updated global model w r+1 The data is distributed to UE 120 again, and the above steps are repeated until the model converges.

[0039] However, this network structure has its limitations. First, the base station's coverage is limited, and service cannot be provided to UEs outside of the coverage area. Second, there are areas with low communication speeds within the base station's coverage area. For UEs in these areas, even within the coverage area, service quality cannot be guaranteed. Furthermore, the base station's uplink and downlink communication resources (such as frequency resources and number of carriers) are limited, and cannot support the simultaneous access of too many UEs.

[0040] Figure 2 shows an exemplary structure of a hierarchical federated learning network according to an embodiment of the present disclosure. The network structure consists of two layers: the first layer consists of intermediate nodes 220 and UEs 230 they serve (UEs connected to intermediate nodes are referred to as local UEs herein); the second layer consists of global nodes 210, intermediate nodes 220 connected to them, and UEs 240 directly connected to the global nodes (UEs directly connected to global nodes are referred to as global UEs herein).

[0041] Global nodes are implemented by base stations and / or servers connected to base stations. Intermediate nodes can be implemented by mobile vehicle-mounted relays (VMRs), unmanned aerial vehicles (UAVs), etc. Alternatively, intermediate nodes can be implemented as fixed-position roadside units (RSUs), edge nodes, etc.

[0042] The following describes an exemplary federated learning process of a hierarchical federated learning network according to an embodiment of the present disclosure in conjunction with FIG3 .

[0043] In step S302, local UE 230 (acting as a primary node) performs, for example, k1 local iterations, and then uploads the resulting local model to intermediate node 220 (acting as a secondary node) for intermediate aggregation. In step S304, intermediate node 220 performs an intermediate aggregation on the local model received from local UE 230 to obtain an intermediate model.

[0044] In step S306, intermediate node 220 determines whether a total of, for example, k2 intermediate aggregations have been completed. If the total number of intermediate aggregations is less than k2, intermediate node 220 distributes the intermediate model to local UE 230. After receiving the intermediate model, local UE 230 updates its local model to the intermediate model and performs a new round of local iteration. If the total number of intermediate aggregations has reached k2, intermediate node 220 uploads the intermediate model to global node 210 for global aggregation.

[0045] In step S308 , if there is a global UE 240 directly served by the global node 210 , the global UE 240 performs, for example, k1k2 local iterations and then uploads the obtained local model to the global node 210 .

[0046] In step S310, the global node 210 (or a network device such as a server connected to it) globally aggregates the intermediate model uploaded by the intermediate node 220 and the local model (if any) uploaded by the global UE 240, and distributes the resulting global model to the intermediate node 220 and the global UE 240 (if any). After receiving the global model, the intermediate node 220 updates the intermediate model to the global model and distributes the global model to the local UE 230 it serves. After receiving the global model, the local UE 230 and the global UE 240 update their local models to the global model.

[0047] In the hierarchical federated learning network structure of the embodiment of the present disclosure, local UEs are connected to intermediate nodes, which shortens the communication distance and ensures the quality of communication services. In addition, multiple local UEs communicate directly with the intermediate nodes, and the intermediate nodes are then connected to the global nodes. Although the intermediate node serves multiple local UEs, each communication between it and the global node only uploads the aggregated intermediate model. The amount of data is equivalent to the amount of data when a global UE communicates with a global node or a local UE communicates with an intermediate node. The load on the global node is greatly reduced, alleviating the problem of insufficient communication resources at the global node. In addition, the network structure makes full use of the function of the intermediate node, that is, the intermediate node not only acts as a relay for transmission, but also participates in the calculation as an aggregator of the first layer structure to complete the intermediate aggregation of the model.

[0048] The following will introduce in detail the specific federated learning processing in the local UE, intermediate node, global UE and global node according to the embodiment of the present disclosure. Assume that there are C global UEs, M intermediate nodes, and the number of local UEs served by the i-th intermediate node is n i During the federated learning process, UEs and intermediate nodes can upload the federated learning model by uploading model parameters or gradients. Therefore, the federated learning process for uploading model parameters and gradients will be described separately below.

[0049] First, we introduce the federated learning process of uploading model parameters. The global node first initializes the global model w 0 , and distribute the initialized global model to the global UE and intermediate nodes. The intermediate nodes then distribute the initialized global model to the local UE. At this point, the global node, global UE, intermediate nodes, and local UE have the same federated learning model:

[0050] in,

[0051] ----Model parameters of intermediate node #i,

[0052] ----Model parameters of the lth local UE served by intermediate node #i,

[0053] n i ----The number of local UEs served by intermediate node #i.

[0054] Each local UE performs local iteration based on locally stored data:

[0055] in,

[0056] ----Local model parameters of the lth local UE served by intermediate node #i after the r1+1th local iteration,

[0057] ----Local model parameters of the lth local UE served by intermediate node #i after the r1th local iteration,

[0058] ----The local gradient of the lth local UE served by intermediate node #i at the r1+1th local iteration,

[0059] ----Local loss function of the lth local UE served by intermediate node #i at the r1+1th local iteration.

[0060] The local UE obtains the local model parameters after k1 local iterations And upload it to the intermediate node. After receiving the local models uploaded by all the local UEs it serves, the intermediate node performs an intermediate aggregation:

[0061] in,

[0062] ----Intermediate model parameters after the intermediate aggregation of intermediate node #i,

[0063] ----Intermediate model parameters after the intermediate aggregation of intermediate node #i,

[0064] p i,l The weight corresponding to the local model uploaded by the lth local UE served by intermediate node #i is usually determined by using the previous intermediate model as a reference.

[0065] r1=k1r2, that is, each time the intermediate node completes an intermediate aggregation, the local UE it serves completes k1 local model updates.

[0066] Then, the intermediate node distributes the model obtained by performing intermediate aggregation (referred to as the intermediate model herein) to the local UEs it serves.

[0067] Repeat the process of equations (2) and (3). When r2 is not an integer multiple of k2, the intermediate node distributes the intermediate aggregated model to each local UE it serves. The local UE uses the received intermediate model to update the local model. When r2 is an integer multiple of k2, after completing the k2th intermediate aggregation, the intermediate node uploads the obtained intermediate model to the global node.

[0068] In addition, in parallel with the processes of equations (2) and (3), the global UE performs a separate local iteration based on local data, which is similar to the process of equation (2):

[0069] in,

[0070] ----Local model parameters of the jth global UE after the r1+1th local iteration,

[0071] ----Local model parameters of the jth global UE after the r1th local iteration,

[0072] ----The local gradient of the jth global UE at the r1+1th local iteration,

[0073] ----The local loss function of the jth global UE at the r1+1th local iteration.

[0074] The global UE performs k1k2 local iterations to obtain the local model parameters And upload to the global node.

[0075] After receiving the intermediate models uploaded by all intermediate nodes and the local models uploaded by all global UEs, the global node performs global aggregation:

[0076] in,

[0077] ----Global model parameters of the global node after the r3+1th global aggregation,

[0078] ----Intermediate model parameters uploaded by the i-th intermediate node,

[0079] ----Local model parameters uploaded by the jth global UE,

[0080] p i The weight corresponding to the intermediate model uploaded by the i-th intermediate node usually needs to be determined using the previous global model as a reference.

[0081] p c,j The weight corresponding to the local model uploaded by the jth global UE is usually determined by using the previous global model as a reference.

[0082] r2=k2r3, that is, every time a global node completes a global aggregation, the intermediate nodes connected to it complete k2 intermediate aggregations.

[0083] The global node performs global model aggregation to obtain global model parameters And distributed to each intermediate node and global UE.

[0084] Repeat the process from (2) to (5) until the global model converges.

[0085] Next, we will introduce the federated learning process of UE and intermediate nodes uploading gradients. First, the global node initializes the global model w 0 , and distributes it to the global UE and intermediate nodes. The intermediate nodes then distribute the initialized global model to the local UEs. At this point, the global node, global UE, intermediate nodes, and global UE all have the same learning model:

[0086] in,

[0087] ----Local model parameters of intermediate node #i,

[0088] ----Local model parameters of the lth UE served by intermediate node #i,

[0089] n i ----The number of local UEs served by intermediate node #i.

[0090] Each local UE performs local iteration based on locally stored data:

[0091] in,

[0092] ----Local model parameters of the lth local UE served by intermediate node #i after the r1+1th local iteration,

[0093] ----Local model parameters of the lth local UE served by intermediate node #i after the r1th local iteration,

[0094] ----The local gradient of the lth local UE served by intermediate node #i at the r1+1th local iteration,

[0095] ----Local loss function of the lth local UE served by intermediate node #i at the r1+1th local iteration.

[0096] The local UE obtains the local model parameters after k1 local iterations And calculate the sum of the gradients of k1 local iterations

[0097] Local UE upload gradient To the intermediate node. After receiving the local gradients uploaded by all the local UEs it serves, the intermediate node performs an intermediate aggregation:

[0098] in,

[0099] ----Intermediate model parameters after the intermediate aggregation of intermediate node #i,

[0100] ----Intermediate model parameters after the intermediate aggregation of intermediate node #i,

[0101] ----The gradient of the intermediate node #i during the r2+1th intermediate aggregation,

[0102] p i,l ----The weight corresponding to the local model (gradient) uploaded by the lth local UE served by intermediate node #i,

[0103] r1=k1r2, that is, each time the intermediate node completes an intermediate aggregation, the local UE it serves completes k1 local model updates.

[0104] Repeat the process from (7) to (9). When r2 is not an integer multiple of k2, the intermediate node distributes the intermediate aggregated model (herein referred to as the intermediate model) to each local UE it serves. The local UE uses the received intermediate model to update the local model. When r2 is an integer multiple of k2, after completing the k2th intermediate aggregation, the intermediate node uploads the obtained intermediate model to the global node.

[0105] The intermediate nodes are aggregated k2 times to obtain the intermediate model parameters. And calculate the sum of the gradients of k2 intermediate aggregations

[0106] The intermediate nodes will be gradient Upload to the global node for global aggregation.

[0107] In addition, in parallel with the process of equations (7) to (10), the global UE performs a local iteration based on the local data, which is similar to the process of equation (7):

[0108] in,

[0109] ----Local model parameters of the jth global UE after the r1+1th model update,

[0110] ----Local model parameters of the jth global UE before the r1th model update,

[0111] ----The local gradient of the jth global UE at the r1th model update,

[0112] ----The local loss function of the jth global UE at the r1th model update,

[0113] The global UE performs k1k2 local iterations to obtain the local model parameters And upload the gradient To the global node:

[0114] After receiving the intermediate models uploaded by all intermediate nodes and the local models uploaded by all global UEs, the global node performs global aggregation:

[0115] in,

[0116] ----Global model parameters of the global node after the r3+1th global aggregation,

[0117] ----The intermediate model gradient uploaded by the i-th intermediate node,

[0118] ----The local model gradient uploaded by the jth global UE,

[0119] p i ----The weight corresponding to the intermediate model uploaded by the i-th intermediate node,

[0120] pc,j ----The weights corresponding to the local model uploaded by the j-th global UE,

[0121] r3 = k1k2r1, that is, every time the global node completes a global aggregation, the intermediate nodes connected to it complete k2 intermediate aggregations.

[0122] The global node performs global model aggregation and update, and distributes the aggregated model to each intermediate node and global UE.

[0123] Repeat the process of formulas (7) to (13) until the global model converges.

[0124] Due to the mobility of UEs and intermediate nodes, changes in the wireless channel, etc., UEs may need to switch during the execution of the federated learning task. Fig. 4A shows a scenario where one or more local UEs switch from intermediate node #i to intermediate node #j, and there are local UEs served by intermediate node #j before the switch. In this scenario, there is an intermediate model at intermediate node #j, so there is no need to transmit the intermediate model from intermediate node #i to intermediate node #j.

[0125] Assuming that federated learning is not considered, UE#l directly switches when it meets the switching condition (such as the received signal strength RSRP is less than a certain threshold). If the switch of UE#l occurs during the (r2 + 1)-th intermediate aggregation of intermediate node #i (the r2-th intermediate aggregation has been completed, but the (r2 + 1)-th intermediate aggregation has not been completed, that is, k1r2 < r1 < k1(r2 + 1)), the following results will occur.

[0126] First, for intermediate node #i, it has disconnected from UE#l and cannot receive the local model of UE#l during the (r2 + 1)-th intermediate aggregation. Second, for intermediate node #j, it can receive the local model uploaded by UE#l. However, the local model uploaded by UE#l after r1 = k1(r2 + 1) local iterations is trained based on the intermediate model received from intermediate node #i (instead of intermediate node #j) when r1 = k1r2.

[0127] An existing solution is that intermediate node #j discards and does not use the local model uploaded by UE#l after r1 = k1(r2 + 1) local iterations. Therefore, for UE#l, only when it is completely within the coverage of the corresponding intermediate node during the (r2 + 1)-th intermediate aggregation of intermediate node #i (or intermediate node #j) (from the model sent by the r2-th intermediate aggregation to UE#l, to UE#l completing k1(r2 + 1) local iterations and uploading), can it participate in the (r2 + 1)-th intermediate aggregation of the corresponding intermediate node.

[0128] Another existing solution is for intermediate node #j to use the local model uploaded by UE #1 after r1=k1(r2+1) local iterations for its r2+1th intermediate aggregation. However, this can lead to divergence of the intermediate model at intermediate node #j, resulting in a decrease in system performance (such as the global model convergence speed) or global model accuracy.

[0129] Figure 4B shows a scenario where all local UEs served by intermediate node #i are handed over to intermediate node #j, and intermediate node #j has no local UEs served before the handover. This scenario can be caused by the movement of intermediate node #i or the movement of UEs.

[0130] Assume that after completing r2 intermediate aggregations, the UE switches from intermediate node #i to intermediate node #j. Next, intermediate node #j performs the r2+1th intermediate aggregation, which requires the intermediate model after the r2th intermediate aggregation. If the gradient is uploaded, as shown in formula (10), the intermediate model after the r2th intermediate aggregation is required to calculate the intermediate model after the r2+1th intermediate aggregation, but at this time, the intermediate model after the r2th intermediate aggregation is not available at intermediate node #j. Even if the model parameters are uploaded, the weights used to calculate the intermediate model need to be determined using the previous intermediate model parameters.

[0131] In this scenario, there is no intermediate model at intermediate node #j, only the global model initially received from the global node. Therefore, intermediate node #i needs to transmit its intermediate model to intermediate node #j.

[0132] Figure 4C shows a scenario where some local UEs served by intermediate node #i are handed over to intermediate node #j, while intermediate node #j had no local UEs served before the handover. Compared to before the handover, after the handover, both intermediate node #i and intermediate node #j lack some UEs participating in the intermediate aggregation, which can lead to model divergence and reduced accuracy. In other words, handover during the intermediate aggregation process, even if only some UEs are handed over, can lead to intermediate model divergence, reducing the accuracy of the global model.

[0133] In response to the above problems, some embodiments of the present disclosure enable the UE to switch after global aggregation and global model broadcasting. At this time, the models of each UE and the intermediate node are the same, all of which are global models. There is no need for additional model transmission, and the switching cost is minimized. If switching cannot be guaranteed after global aggregation and global model broadcasting, some embodiments of the present disclosure enable switching after the intermediate node completes intermediate aggregation. At this time, the intermediate node has the same model as all local UEs it serves, and can ensure that the training service of the local UE is not interrupted. In addition, some embodiments of the present disclosure also extend the connection time of the UE and / or intermediate node until the intermediate aggregation or global aggregation is completed by increasing the transmission power, lowering the RSRP threshold, allocating more transmission resources (including time resources, frequency resources), etc. This ensures service continuity and improves system performance.

[0134] For the sake of simplicity, first define the following variables:

[0135] T serve ---The estimated value of the remaining time for the global node or intermediate node to provide service to the UE;

[0136] T1---The estimated remaining time required for the global node to complete this round of global aggregation, that is, the time from the current moment to the time when the global node completes this round of global aggregation;

[0137] T2---The time required for the global node to complete the next round of global aggregation, that is, the time from the global node completing the current round of global aggregation to the global node completing the next round of global aggregation;

[0138] T train ---The remaining time required for the global node to complete the current and next rounds of global aggregation, that is, the time from the current moment to the time when the global node completes the next round of global aggregation, that is, T train =T1+T2;

[0139] t1---The remaining time required for the intermediate node to complete this round of intermediate aggregation, that is, the time from the current moment to the intermediate node completing this round of intermediate node aggregation;

[0140] t2---The remaining time required for the intermediate node to complete the next round of intermediate aggregation, that is, the time from the intermediate node completing the current round of intermediate node aggregation to the intermediate node completing the next round of intermediate node aggregation;

[0141] t train ---The remaining time required for the intermediate node to complete the current and next round of global aggregation, that is, the time from the current moment to the intermediate node completing the next round of intermediate node aggregation, that is, t train =t1+t2.

[0142] First, let's discuss the case where UE#1 is a local UE served by intermediate node#i. UE#1 can be switched to be served by intermediate node#j or directly by a global node. FIG5 shows an exemplary handover process for a local UE according to an embodiment of the present disclosure.

[0143] In step S501, UE#1 sends its own status information Info U Sent to intermediate node #i. UE status information Info U It may include one or more of the channel status (such as RSRP), computing power (such as CPU occupancy), local data information (such as the number of samples involved in model training, sample dimension, etc.), power, location and movement information (such as speed, direction, length of stay in a certain place, etc.), and other information.

[0144] In step S502, the intermediate node #i sends its own state information Info V And the status information of the local UE it serves Info U Sent to the global node. Status information of the intermediate node Info V This may include one or more of channel status (RSRP), computing power (e.g., CPU occupancy), location and movement information (e.g., speed, direction, time spent at a location), and other information. The channel status of the intermediate node may include the channel status between the intermediate node and the local UE and the channel status between the intermediate node and the global node.

[0145] In step S503, the global node determines the remaining service time T for UE#1. serve , T serve is the remaining time for the intermediate node #i to provide services to the local UE. serve When the predetermined conditions are met, the global node makes a switching decision.

[0146] For T serve The estimation can be done by the global node based on the state information Info of the intermediate node #i V and UE#1 status information Info U The global node can estimate the link quality and connection time (such as the time when RSRP is greater than a certain threshold) between it and the intermediate node #i, as well as the link quality and connection time (such as the time when RSRP is greater than a certain threshold) between the intermediate node #i and UE #1, and then estimate the time T that the intermediate node #i can serve UE #1. serve For example, the global node may determine the time when the link between it and the intermediate node #i and the link between the intermediate node #i and UE #1 simultaneously meet the corresponding requirements as the time T that the intermediate node #i can serve UE #1.serve .

[0147] For T1, T2, T train The estimation can be done by the global node based on the status information Info of all UEs U and status information of all intermediate nodes Info V OK. For t1, t2, t train The estimation can be done by the global node based on the state information Info of the intermediate node #i V And the status information Info of all local UEs served by intermediate node #i U Alternatively, for t1, t2, t train The estimation can be done by the intermediate node #i based on its own state information Info V And the status information of all local UEs it serves U Determine and send to the global node in step S502.

[0148] For T serve , T1, T2, T train , t1, t2, t train The estimation of can be performed periodically by the global node and each intermediate node, or can be triggered by certain triggering events, such as sudden movement of the intermediate node or UE.

[0149] In step S506, the global node sends the handover decision to the intermediate node #i. In step S508, the intermediate node #i sends the received handover decision to UE #1. In step S510, UE #1 performs handover based on the received handover decision.

[0150] After making a handover decision, the global node can send it immediately, or it can send it when the global model is broadcast after global aggregation is complete. In the hierarchical federated learning structure of base station, VMR, and UE, the handover decision can be sent through the traditional method, that is, sent to intermediate node #i via the Uu link (downlink), and then from intermediate node #i to UE#1 via PC5 (sidelink). The global model is sent via broadcast and can be received by all intermediate nodes. However, the handover decision of the global node is not broadcast, but is sent only to UE#1 that needs to perform the handover and its connected intermediate node #i.

[0151] 6A-6E are discussed below when UE#1 is a local UE. serve The conditions that are met and whether a switching decision is made under the corresponding conditions.

[0152] FIG. 6A shows a case where UE#1 is a local UE. serve >T trainCase. In this case, the UE can participate in completing the global aggregation of the current round and the next round. Therefore, no handover is performed temporarily before T train ends.

[0153] Figure 6B shows the case where UE#l is a local UE and T1 < T serve < T train Case. In this case, the UE can participate in completing the global aggregation of the current round, but its service time cannot support the completion of the global aggregation of the next round, so a handover is performed after the global aggregation of the current round. The global node broadcasts and sends the global model after the global aggregation of the current round. At this time, each UE and the intermediate node have the same model. For the handover, only the establishment and release of the communication link need to be considered, without considering the transfer of the model, the divergence of the intermediate model, etc.

[0154] Figure 6C shows the case where UE#l is a local UE and t train < T serve < T1. In this case, the service provided by the original intermediate node #i to the UE cannot complete the global model aggregation of the current round, but can complete the intermediate aggregation of the intermediate node #i in the current round and the next round. Therefore, t train ends.

[0155] In some embodiments of the present disclosure, in the case where T serve < T1, the global node can estimate the improved remaining service time T′ on the assumption of performing one or more of the following operations serve : increasing the transmission power of one or more of UE#l, intermediate node #i, and the global node; allocating more transmission resources to one or both of UE#l and intermediate node #i; and reducing the RSRP threshold of one or more of UE#l, intermediate node #i, and the global node. If T′ serve > T1, the global node performs the one or more operations and instructs UE#l to perform a handover after the end of the current global aggregation.

[0156] Figure 6D shows the case where UE#l is a local UE and t1 < T serve < t trainCase. In this case, the service provided by the original intermediate node #i to the UE can participate in completing the intermediate aggregation of the intermediate node #i in this round, but its service time cannot support the completion of the intermediate aggregation of the intermediate node #i in the next round. Therefore, a switch is made after the intermediate aggregation of the intermediate node #i in this round. The intermediate node #i sends the intermediate model after the intermediate aggregation in this round to the UE it serves. At this time, the UE and the intermediate node #i have the same model. If the UE switches to the intermediate node #j and the intermediate node #j has no served users, the intermediate node #i needs to send the intermediate model to the intermediate node #j. If the UE switches to be directly served by the global node, the UE directly uploads the local model to the global node for global aggregation after completing k1k2 local iterations.

[0157] Figure 6E shows the case when UE#l is a local UE at T serve <Case of t1. In this case, the service provided by the original intermediate node #i to UE#l cannot support the completion of the intermediate aggregation of the intermediate node #i in this round. Therefore, a direct switch is made. If UE#l switches to the intermediate node #j and the intermediate node #j has no served users, the intermediate node #i needs to send the intermediate model to the intermediate node #j. If the UE switches to be directly served by the global node, the UE directly uploads the local model to the global node for global aggregation after completing k1k2 local iterations.

[0158] In some embodiments of the present disclosure, at T serve <In the case of t1, the global node can estimate the increased remaining service time T′ under the assumption of performing one or more of the following operations serve : increasing the transmission power of one or more of UE#l, the intermediate node #i, and the global node; allocating more transmission resources to one or both of UE#l and the intermediate node #i; and reducing the RSRP thresholds of one or more of UE#l, the intermediate node #i, and the global node. If T′ serve > t1, the global node performs the one or more operations and instructs UE#l to switch after the end of this intermediate aggregation. Next, the case where UE#l is a global UE directly served by the global node is discussed. UE#l can switch to be served by the intermediate node #j. Figure 7 shows the switching process of the global UE according to an embodiment of the present disclosure.

[0159] In step S791, UE#l sends its own status information Info U to the global node. In step S792, the global node determines the remaining service time T serve for UE#l, where T serve is the remaining time for the global node to serve UE#l. In step S794, at T serveWhen the predetermined conditions are met, the global node makes a switching decision.

[0160] For T serve The estimation can be performed by the global node based on the status information Info of UE#1 U The global node can estimate the link quality and connection time (such as the time when RSRP is greater than a certain threshold) between it and UE#1, and then estimate the time T that the global node can serve UE#1. serve For example, the global node may determine the time when the link between it and UE#1 meets the corresponding requirements as the time T that the global node can serve UE#1. serve .

[0161] For T1, T2, T train The estimation can be done by the global node based on the status information Info of all UEs U and status information of all intermediate nodes Info V OK. For T serve , T1, T2, T train The estimation of can be performed periodically by the global node or triggered by certain triggering events, such as sudden movement of UE.

[0162] In step S796, the global node sends the handover decision directly to UE#1. In step S798, UE#1 performs handover based on the received handover decision.

[0163] After making a handover decision, the global node can send it immediately, or after global aggregation completes and when the global model is broadcast. The handover decision can be sent directly to UE#1 via the Uu link (Downlink). The global model is broadcast and can be received by all intermediate nodes and the global UE. However, the global node's handover decision is not broadcast; it is sent only to UE#1 that requires handover.

[0164] 8A-8C are discussed below when UE#1 is a global UE. serve The conditions that are met and whether a switching decision is made under the corresponding conditions.

[0165] FIG. 8A shows a case where UE#1 is a global UE. serve >T train In this case, UE#1 can participate in completing the global aggregation of this round and the next round. train No switching will be performed until the end.

[0166] FIG. 8B shows T1 when UE#1 is a global UE <T serve <T trainIn this case, UE#l can participate in completing the current round of global aggregation, but its service time cannot support the completion of the next round of global aggregation, so it switches after the current round of global aggregation. The global node broadcasts and sends the global model after the current round of global aggregation. At this time, each UE#l and the intermediate node have the same model. The switch only needs to consider the establishment and release of the communication link, without considering the transfer of the model, the divergence of the intermediate model, etc.

[0167] Figure 8C shows the case of T when UE#l is the global UE serve <The case of T1. In this case, the service provided by the global node to UE#l cannot support the completion of the current round of global aggregation. Therefore, a direct switch is made.

[0168] In some embodiments of the present disclosure, at T serve <In the case of T1, the global node can estimate the increased remaining service time T′ assuming that one or more of the following operations are performed serve : increasing the transmission power of one or both of UE#l and the global node; allocating more transmission resources to UE#l; and reducing the RSRP threshold of one or both of UE#l and the global node. If T′ serve >T1, the global node performs the one or more operations and instructs UE#l to switch after the end of the current global aggregation.

[0169] Embodiments of the present disclosure can be applied to the 5G core network. Figure 9 shows the 5G core network SBA (Service-based Architecture) architecture and some of its network functions (NFs).

[0170] AF (Application Function) refers to various services at the application layer, which can be internal applications of the operator or third-party AFs (such as video servers, game servers).

[0171] NEF (Network Exposure Function) is located between the 5G core network and external third-party application functions and is responsible for managing the对外开放 network data. All external applications that want to access the internal data of the 5G core network must pass through NEF.

[0172] NWDAF (Network Data Analytics Function) can collect data, perform analysis and provide the analysis results to other network functions, such as NEF.

[0173] The Access and Mobility Management Function (AMF) is responsible for registration, connection, reachability, mobility, security, access management, and service authorization.

[0174] The PCF (Policy Control Function) provides all mobility, UE access selection, and PDF session-related policies.

[0175] NWDAF can analyze the movement of intermediate nodes and the switching time to provide information to AF so that AF can calculate the optimal federated learning time information, and send the time information to AMF to influence the mobility management of UE to achieve efficient federated learning. For example, NWDAF can estimate T serve , T1, T2, T train , t1, t2, t train The estimation result is output to the AF, or relevant information is output to the AF for estimation.

[0176] In addition, the AF may also send the handover rules of the embodiment of the present disclosure to the PCF, and the PCF may send the mobility management policy to the AMF to control the handover of the UE. For example, the AF sends the rule that the UE should try to maintain the connection with the intermediate node before global aggregation or intermediate aggregation is completed to the PCF, and the PCF further sends policies such as allocating more transmission resources, increasing the transmit power, lowering the RSRP threshold, or reducing the transmission rate to the AMF.

[0177] In addition, AF can also obtain UE handover information through NEF to control gNB and federated learning applications in the cloud.

[0178] <Application Examples>

[0179] The technology disclosed herein can be applied to various products. The base station can be implemented as any type of evolved Node B (eNB), gNB or TRP (Transmit Receive Point), such as macro eNB / gNB and small eNB / gNB. Small eNB / gNB can be an eNB / gNB that covers a cell smaller than a macro cell, such as pico eNB / gNB, micro eNB / gNB and home (femto) eNB / gNB. Alternatively, the base station can be implemented as any other type of base station, such as NodeB and base transceiver station (BTS). The base station may include: a main body (also called a base station device) configured to control wireless communications; and one or more remote radio heads (RRHs) arranged in a place different from the main body. In addition, the various types of terminals described below can all work as base stations by temporarily or semi-persistently performing base station functions.

[0180] The user equipment may be implemented as a mobile terminal (such as a smartphone, a tablet personal computer (PC), a notebook PC, a portable game terminal, a portable / dongle-type mobile router, and a digital camera) or an in-vehicle terminal (such as a car navigation device). The user equipment may also be implemented as a terminal that performs machine-to-machine (M2M) communication (also known as a machine type communication (MTC) terminal).

[0181] Furthermore, both the base station and the user equipment may be implemented as various types of computing devices.

[0182] [Application examples for computing devices]

[0183] 10 is a block diagram illustrating an example of a schematic configuration of a computing device 700 to which the technology of the present disclosure may be applied. The computing device 700 includes a processor 701 , a memory 702 , a storage device 703 , a network interface 704 , and a bus 706 .

[0184] The processor 701 may be, for example, a central processing unit (CPU) or a digital signal processor (DSP), and controls the functions of the server 700. The memory 702 includes a random access memory (RAM) and a read-only memory (ROM), and stores data and programs executed by the processor 701. The storage device 703 may include a storage medium such as a semiconductor memory and a hard disk.

[0185] The network interface 704 is a wired communication interface for connecting the server 700 to a wired communication network 705. The wired communication network 705 may be a core network such as an evolved packet core (EPC) or a packet data network (PDN) such as the Internet.

[0186] The bus 706 connects the processor 701, the memory 702, the storage device 703, and the network interface 704 to each other. The bus 706 may include two or more buses each having a different speed (such as a high-speed bus and a low-speed bus).

[0187] [Application examples for base stations]

[0188] (First application example)

[0189] 11 is a block diagram illustrating a first example of a schematic configuration of a gNB to which the present disclosure can be applied. The gNB 800 includes one or more antennas 810 and a base station device 820. The base station device 820 and each antenna 810 can be connected to each other via an RF cable.

[0190] Each of the antennas 810 includes a single or multiple antenna elements (such as multiple antenna elements included in a multiple-input multiple-output (MIMO) antenna) and is used for base station device 820 to transmit and receive wireless signals. As shown in FIG11 , gNB 800 may include multiple antennas 810. For example, the multiple antennas 810 may be compatible with multiple frequency bands used by gNB 800. Although FIG11 shows an example in which gNB 800 includes multiple antennas 810, gNB 800 may also include a single antenna 810.

[0191] The base station device 820 includes a controller 821 , a memory 822 , a network interface 823 , and a wireless communication interface 825 .

[0192] The controller 821 may be, for example, a CPU or DSP, and operates various higher-layer functions of the base station device 820. For example, the controller 821 generates data packets based on data in the signal processed by the wireless communication interface 825 and transmits the generated packets via the network interface 823. The controller 821 may bundle data from multiple baseband processors to generate bundled packets and transmit the generated bundled packets. The controller 821 may have logic functions for performing control such as radio resource control, radio bearer control, mobility management, admission control, and scheduling. This control may be performed in conjunction with a nearby gNB or core network node. The memory 822 includes RAM and ROM and stores programs executed by the controller 821 and various types of control data (such as terminal lists, transmission power data, and scheduling data).

[0193] The network interface 823 is a communication interface for connecting the base station device 820 to the core network 824. The controller 821 can communicate with the core network node or another gNB via the network interface 823. In this case, the gNB 800 and the core network node or other gNB can be connected to each other via a logical interface (such as an S1 interface and an X2 interface). The network interface 823 can also be a wired communication interface or a wireless communication interface for a wireless backhaul line. If the network interface 823 is a wireless communication interface, the network interface 823 can use a higher frequency band for wireless communication than the frequency band used by the wireless communication interface 825.

[0194] The wireless communication interface 825 supports any cellular communication scheme, such as Long Term Evolution (LTE) and LTE-Advanced, and provides wireless connectivity to terminals located in the cell of the gNB 800 via the antenna 810. The wireless communication interface 825 may typically include, for example, a baseband (BB) processor 826 and RF circuitry 827. The BB processor 826 can perform various signal processing functions, such as encoding / decoding, modulation / demodulation, and multiplexing / demultiplexing, and performs various types of signal processing for layers such as Layer 1 (L1), Medium Access Control (MAC), Radio Link Control (RLC), and Packet Data Convergence Protocol (PDCP). In place of the controller 821, the BB processor 826 may perform some or all of the aforementioned logical functions. The BB processor 826 may be a memory storing communication control programs, or a module including a processor configured to execute programs and associated circuitry. Program updates can modify the functionality of the BB processor 826. This module may be a card or blade inserted into a slot in the base station device 820. Alternatively, it may be a chip mounted on the card or blade. Meanwhile, the RF circuit 827 may include, for example, a mixer, a filter, and an amplifier, and transmit and receive wireless signals via the antenna 810 .

[0195] As shown in Figure 11 , the wireless communication interface 825 may include multiple BB processors 826. For example, multiple BB processors 826 may be compatible with multiple frequency bands used by the gNB 800. As shown in Figure 11 , the wireless communication interface 825 may include multiple RF circuits 827. For example, multiple RF circuits 827 may be compatible with multiple antenna elements. While Figure 11 illustrates an example in which the wireless communication interface 825 includes multiple BB processors 826 and multiple RF circuits 827, the wireless communication interface 825 may also include a single BB processor 826 or a single RF circuit 827.

[0196] (Second application example)

[0197] FIG12 is a block diagram illustrating a second example of a schematic configuration of a gNB to which the techniques of this disclosure can be applied. A gNB 830 includes one or more antennas 840, a base station device 850, and an RRH 860. The RRH 860 and each antenna 840 can be connected to each other via an RF cable. The base station device 850 and the RRH 860 can also be connected to each other via a high-speed line such as a fiber optic cable.

[0198] Each of the antennas 840 includes a single or multiple antenna elements (such as multiple antenna elements included in a MIMO antenna) and is used for RRH 860 to transmit and receive wireless signals. As shown in FIG12 , gNB 830 may include multiple antennas 840. For example, the multiple antennas 840 may be compatible with multiple frequency bands used by gNB 830. Although FIG12 shows an example in which gNB 830 includes multiple antennas 840, gNB 830 may also include a single antenna 840.

[0199] Base station device 850 includes a controller 851, a memory 852, a network interface 853, a wireless communication interface 855, and a connection interface 857. Controller 851, memory 852, and network interface 853 are the same as controller 821, memory 822, and network interface 823 described with reference to FIG.

[0200] The wireless communication interface 855 supports any cellular communication scheme (such as LTE and LTE-Advanced) and provides wireless communication to terminals located in the sector corresponding to the RRH 860 via the RRH 860 and the antenna 840. The wireless communication interface 855 may generally include, for example, a BB processor 856. The BB processor 856 is identical to the BB processor 826 described with reference to FIG. 11 , except that the BB processor 856 is connected to the RF circuit 864 of the RRH 860 via the connection interface 857. As shown in FIG. 12 , the wireless communication interface 855 may include multiple BB processors 856. For example, multiple BB processors 856 may be compatible with multiple frequency bands used by the gNB 830. Although FIG. 12 shows an example in which the wireless communication interface 855 includes multiple BB processors 856, the wireless communication interface 855 may also include a single BB processor 856.

[0201] The connection interface 857 is an interface for connecting the base station device 850 (wireless communication interface 855) to the RRH 860. The connection interface 857 may also be a communication module for connecting the base station device 850 (wireless communication interface 855) to the RRH 860 for communication in the high-speed line.

[0202] The RRH 860 includes a connection interface 861 and a wireless communication interface 863 .

[0203] The connection interface 861 is an interface for connecting the RRH 860 (wireless communication interface 863) to the base station device 850. The connection interface 861 may also be a communication module for communication in the above-mentioned high-speed line.

[0204] The wireless communication interface 863 transmits and receives wireless signals via the antenna 840. The wireless communication interface 863 may generally include, for example, an RF circuit 864. The RF circuit 864 may include, for example, a mixer, a filter, and an amplifier, and transmits and receives wireless signals via the antenna 840. As shown in FIG12 , the wireless communication interface 863 may include multiple RF circuits 864. For example, the multiple RF circuits 864 may support multiple antenna elements. Although FIG12 shows an example in which the wireless communication interface 863 includes multiple RF circuits 864, the wireless communication interface 863 may also include a single RF circuit 864.

[0205] [Application examples of terminals]

[0206] (First application example)

[0207] 13 is a block diagram showing an example of a schematic configuration of a smartphone 900 to which the technology of the present disclosure can be applied. The smartphone 900 includes a processor 901, a memory 902, a storage device 903, an external connection interface 904, a camera 906, a sensor 907, a microphone 908, an input device 909, a display device 910, a speaker 911, a wireless communication interface 912, one or more antenna switches 915, one or more antennas 916, a bus 917, a battery 918, and an auxiliary controller 919.

[0208] The processor 901 may be, for example, a CPU or a system on a chip (SoC), and controls the functions of the application layer and other layers of the smartphone 900. The memory 902 includes RAM and ROM, and stores data and programs executed by the processor 901. The storage device 903 may include storage media such as semiconductor memories and hard disks. The external connection interface 904 is an interface for connecting external devices (such as memory cards and universal serial bus (USB) devices) to the smartphone 900.

[0209] The camera 906 includes an image sensor such as a charge coupled device (CCD) and a complementary metal oxide semiconductor (CMOS) and generates a captured image. The sensor 907 may include a group of sensors such as a measurement sensor, a gyroscope sensor, a geomagnetic sensor, and an acceleration sensor. The microphone 908 converts the sound input to the smartphone 900 into an audio signal. The input device 909 includes, for example, a touch sensor, a keypad, a keyboard, a button, or a switch configured to detect a touch on the screen of the display device 910, and receives an operation or information input from the user. The display device 910 includes a screen such as a liquid crystal display (LCD) and an organic light emitting diode (OLED) display and displays an output image of the smartphone 900. The speaker 911 converts the audio signal output from the smartphone 900 into sound.

[0210] The wireless communication interface 912 supports any cellular communication scheme (such as LTE and LTE-Advanced) and performs wireless communication. The wireless communication interface 912 may generally include, for example, a BB processor 913 and an RF circuit 914. The BB processor 913 may perform, for example, encoding / decoding, modulation / demodulation, and multiplexing / demultiplexing, and perform various types of signal processing for wireless communication. Meanwhile, the RF circuit 914 may include, for example, a mixer, a filter, and an amplifier, and transmit and receive wireless signals via an antenna 916. The wireless communication interface 912 may be a chip module on which the BB processor 913 and the RF circuit 914 are integrated. As shown in FIG13 , the wireless communication interface 912 may include multiple BB processors 913 and multiple RF circuits 914. Although FIG13 shows an example in which the wireless communication interface 912 includes multiple BB processors 913 and multiple RF circuits 914, the wireless communication interface 912 may also include a single BB processor 913 or a single RF circuit 914.

[0211] In addition, in addition to the cellular communication scheme, the wireless communication interface 912 can support other types of wireless communication schemes, such as a short-range wireless communication scheme, a near-field communication scheme, and a wireless local area network (LAN) scheme. In this case, the wireless communication interface 912 may include a BB processor 913 and an RF circuit 914 for each wireless communication scheme.

[0212] Each of the antenna switches 915 switches a connection destination of the antenna 916 between a plurality of circuits (eg, circuits for different wireless communication schemes) included in the wireless communication interface 912 .

[0213] Each of the antennas 916 includes a single or multiple antenna elements (such as multiple antenna elements included in a MIMO antenna) and is used for transmitting and receiving wireless signals via the wireless communication interface 912. As shown in FIG13 , the smartphone 900 may include multiple antennas 916. Although FIG13 shows an example in which the smartphone 900 includes multiple antennas 916, the smartphone 900 may also include a single antenna 916.

[0214] In addition, the smartphone 900 may include an antenna 916 for each wireless communication scheme. In this case, the antenna switch 915 may be omitted from the configuration of the smartphone 900.

[0215] The bus 917 connects the processor 901, the memory 902, the storage device 903, the external connection interface 904, the camera 906, the sensor 907, the microphone 908, the input device 909, the display device 910, the speaker 911, the wireless communication interface 912, and the auxiliary controller 919. The battery 918 supplies power to the various blocks of the smartphone 900 shown in FIG13 via feeders, which are partially shown as dotted lines in the figure. The auxiliary controller 919 operates the minimum necessary functions of the smartphone 900, for example, in sleep mode.

[0216] (Second application example)

[0217] 14 is a block diagram showing an example of a schematic configuration of a car navigation device 920 to which the technology of the present disclosure can be applied. The car navigation device 920 includes a processor 921, a memory 922, a global positioning system (GPS) module 924, a sensor 925, a data interface 926, a content player 927, a storage medium interface 928, an input device 929, a display device 930, a speaker 931, a wireless communication interface 933, one or more antenna switches 936, one or more antennas 937, and a battery 938.

[0218] The processor 921 may be, for example, a CPU or an SoC, and controls a navigation function and other functions of the car navigation apparatus 920. The memory 922 includes a RAM and a ROM, and stores data and programs executed by the processor 921.

[0219] The GPS module 924 measures the position (such as latitude, longitude, and altitude) of the car navigation device 920 using GPS signals received from GPS satellites. The sensor 925 may include a group of sensors such as a gyroscope sensor, a geomagnetic sensor, and an air pressure sensor. The data interface 926 is connected to, for example, the in-vehicle network 941 via an unillustrated terminal and acquires data generated by the vehicle (such as vehicle speed data).

[0220] The content player 927 reproduces content stored in a storage medium (such as a CD or DVD) inserted into the storage medium interface 928. The input device 929 includes, for example, a touch sensor, button, or switch configured to detect a touch on the screen of the display device 930, and receives an operation or information input from the user. The display device 930 includes a screen such as an LCD or OLED display and displays an image of a navigation function or reproduced content. The speaker 931 outputs the sound of the navigation function or the reproduced content.

[0221] The wireless communication interface 933 supports any cellular communication scheme (such as LTE and LTE-Advanced) and performs wireless communication. The wireless communication interface 933 may generally include, for example, a BB processor 934 and an RF circuit 935. The BB processor 934 may perform, for example, encoding / decoding, modulation / demodulation, and multiplexing / demultiplexing, and perform various types of signal processing for wireless communication. Meanwhile, the RF circuit 935 may include, for example, a mixer, a filter, and an amplifier, and transmit and receive wireless signals via an antenna 937. The wireless communication interface 933 may also be a chip module on which the BB processor 934 and the RF circuit 935 are integrated. As shown in Figure 14, the wireless communication interface 933 may include multiple BB processors 934 and multiple RF circuits 935. Although Figure 14 shows an example in which the wireless communication interface 933 includes multiple BB processors 934 and multiple RF circuits 935, the wireless communication interface 933 may also include a single BB processor 934 or a single RF circuit 935.

[0222] In addition, in addition to the cellular communication scheme, the wireless communication interface 933 can support other types of wireless communication schemes, such as a short-range wireless communication scheme, a near field communication scheme, and a wireless LAN scheme. In this case, for each wireless communication scheme, the wireless communication interface 933 can include a BB processor 934 and an RF circuit 935.

[0223] Each of the antenna switches 936 switches a connection destination of the antenna 937 between a plurality of circuits included in the wireless communication interface 933 , such as circuits for different wireless communication schemes.

[0224] Each of the antennas 937 includes a single or multiple antenna elements (such as multiple antenna elements included in a MIMO antenna) and is used for transmitting and receiving wireless signals via the wireless communication interface 933. As shown in FIG14, the car navigation device 920 may include multiple antennas 937. Although FIG14 shows an example in which the car navigation device 920 includes multiple antennas 937, the car navigation device 920 may also include a single antenna 937.

[0225] Furthermore, the car navigation device 920 may include an antenna 937 for each wireless communication scheme. In this case, the antenna switch 936 may be omitted from the configuration of the car navigation device 920.

[0226] The battery 938 supplies power to the respective blocks of the car navigation apparatus 920 shown in Fig. 14 via a feeder line, which is partially shown as a dotted line in the figure. The battery 938 accumulates the power supplied from the vehicle.

[0227] The technology of the present disclosure can also be implemented as an in-vehicle system (or vehicle) 940 including a car navigation device 920, an in-vehicle network 941, and one or more blocks of a vehicle module 942. The vehicle module 942 generates vehicle data (such as vehicle speed, engine speed, and fault information) and outputs the generated data to the in-vehicle network 941.

[0228] The various illustrative blocks and components described in conjunction with the present disclosure may be implemented or executed using a general-purpose processor, a digital signal processor (DSP), an ASIC, an FPGA or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. A general-purpose processor may be a microprocessor, but alternatively, the processor may be any conventional processor, controller, microcontroller, and / or state machine. The processor may also be implemented as a combination of computing devices, such as a DSP and a microprocessor, a plurality of microprocessors, one or more microprocessors in combination with a DSP core, and / or any other such configuration.

[0229] The functions described herein can be implemented in hardware, software executed by a processor, firmware, or any combination thereof. If implemented in software executed by a processor, the functions can be stored on a non-transient computer-readable medium or transmitted as one or more instructions or codes on a non-transient computer-readable medium. Other examples and implementations are within the scope and spirit of this disclosure and the appended claims. For example, given the nature of software, the functions described above can be performed using software executed by a processor, hardware, firmware, hard wiring, or any combination thereof. Features that implement the functions can also be physically placed in various locations, including being distributed so that parts of the functions are implemented in different physical locations.

[0230] In addition, the disclosure of components contained within or separate from other components should be considered exemplary, as potentially a variety of other architectures can be implemented to achieve the same functionality, including incorporating all, most, and / or some of the elements as part of one or more single structures or separate structures.

[0231] Non-transitory computer readable media can be any available non-transitory media that can be accessed by a general-purpose computer or a special-purpose computer. By way of example and not limitation, non-transitory computer readable media can include RAM, ROM, EEPROM, flash memory, CD-ROM, DVD or other optical disk storage, magnetic disk storage or other magnetic storage device, or any other medium that can be used to carry or store desired program code components in the form of instructions or data structures and can be accessed by a general-purpose or special-purpose computer or a general-purpose or special-purpose processor.

[0232] The preceding description of the present disclosure is provided to enable those skilled in the art to make or use the present disclosure. Various modifications to the present disclosure will be apparent to those skilled in the art, and the general principles defined herein may be applied to other variations without departing from the scope of the present disclosure. Therefore, the present disclosure is not limited to the examples and designs described herein, but rather corresponds to the widest scope consistent with the principles and novel features disclosed.

[0233] The present disclosure also includes the following embodiments.

[0234] 1. A network-side electronic device for federated learning, comprising a processing circuit, wherein the processing circuit is configured to:

[0235] Determine the model aggregation time and the remaining service time T for the user device serve , wherein the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node;

[0236] At the model aggregation time and the remaining service time T serve When a predetermined condition is met, making a handover decision for the user equipment; and

[0237] The handover decision for the user equipment is sent.

[0238] 2. The electronic device according to embodiment 1, wherein the processing circuit is further configured to:

[0239] receiving an intermediate aggregation model from the intermediate node;

[0240] generating a global aggregation model based at least on the intermediate aggregation model; and

[0241] The global aggregate model is broadcasted.

[0242] 3. The electronic device according to embodiment 1, wherein the processing circuit is further configured to receive status information of the user equipment, and the model aggregation time and the remaining service time T serve The status information is determined based on at least the user equipment.

[0243] 4. The electronic device as described in embodiment 3, wherein the status information of the user equipment includes one or more of the following: channel status, computing capability, local data information, power level, location and movement information.

[0244] 5. The electronic device according to embodiment 3, wherein the processing circuit is further configured to receive status information of the intermediate node, and the model aggregation time and the remaining service time T serve The determination is also based on the status information of the intermediate node.

[0245] 6. The electronic device according to Embodiment 5, wherein the status information of the intermediate node includes one or more of the following: channel status, computing power, location, and movement information.

[0246] 7. The electronic device according to Embodiment 1, wherein

[0247] the model aggregation time includes the remaining time T1 required for the current global aggregation and the time T2 required for the next global aggregation, and

[0248] if T1 < T serve < T1+T2, a handover decision for the user equipment is made, and the handover decision instructs the user equipment to perform a handover after the end of the current global aggregation.

[0249] 8. The electronic device according to Embodiment 7, wherein if T serve < T1, the processing circuit is further configured to estimate an increased remaining service time T′ assuming the execution of one or more of the following operations serve :

[0250] increasing the transmission power of one or more of the user equipment, the intermediate node, and the global node;

[0251] allocating more transmission resources to one or both of the user equipment and the intermediate node; and

[0252] decreasing the RSRP threshold of one or more of the user equipment, the intermediate node, and the global node,

[0253] wherein if T′ serve > T1, the one or more operations are performed, and a handover decision for the user equipment is made, and the handover decision instructs the user equipment to perform a handover after the end of the current global aggregation.

[0254] 9. The electronic device according to Embodiment 1 or 7, wherein the model aggregation time includes the remaining time t1 required for the current intermediate aggregation at the intermediate node and the time t2 required for the next intermediate aggregation, and

[0255] if t1 < T serve < t1+t2, a handover decision for the user equipment is made, and the handover decision instructs the user equipment to perform a handover after the end of the current intermediate aggregation.

[0256] 10. The electronic device according to Embodiment 9, wherein if T serve<If T is less than t1, a handover decision for the user equipment is made, and the handover decision instructs the user equipment to perform a handover immediately.

[0257] 11. The electronic device according to embodiment 9, wherein if T serve <is less than t1, the processing circuit is further configured to estimate an increased remaining service time T' on the assumption that one or more of the following operations are performed serve :

[0258] Increase the transmission power of one or more of the user equipment, the intermediate node, and the global node;

[0259] Allocate more transmission resources to one or both of the user equipment and the intermediate node; and

[0260] Reduce the RSRP threshold of one or more of the user equipment, the intermediate node, and the global node,

[0261] where if T' serve > t1, perform the one or more operations, and make a handover decision for the user equipment, the handover decision instructing the user equipment to perform a handover after the current intermediate aggregation ends.

[0262] 12. The electronic device according to embodiment 9, wherein if the user equipment switches to another intermediate node and there is no intermediate aggregation model at the other intermediate node, the intermediate aggregation model at the intermediate node is sent to the other intermediate node.

[0263] 13. An intermediate node-side electronic device for federated learning, comprising a processing circuit configured to:

[0264] Receive a handover decision for a user equipment from a network, wherein the handover decision is made when a model aggregation time and a remaining service time T for the user equipment serve meet a predetermined condition, and the user equipment is indirectly connected to the network via the intermediate node;

[0265] Send the handover decision to the user equipment.

[0266] 14. The electronic device according to embodiment 13, wherein the processing circuit is further configured to:

[0267] Receive a local model from the user equipment, generate an intermediate aggregation model based at least on the local model, and send the intermediate aggregation model to the network; and

[0268] A global aggregation model is received from the network, and the global aggregation model is sent to the user equipment.

[0269] 15. The electronic device according to embodiment 13, wherein the processing circuit is further configured to:

[0270] receiving status information of the user equipment;

[0271] sending the status information of the user equipment and the status information of the intermediate node to the network,

[0272] Among them, the model aggregation time and the remaining service time T serve The state information of the intermediate node is determined based on at least the state information of the user equipment and the state information of the intermediate node.

[0273] 16. The electronic device according to embodiment 13, wherein the processing circuit is further configured to:

[0274] receiving status information of the user equipment;

[0275] determining at least a portion of the model aggregation time based at least on state information of the user equipment and state information of the intermediate node;

[0276] sending the state information of the user equipment, the state information of the intermediate node, and the at least a portion of the model aggregation time to the network,

[0277] Among them, the other part of the model aggregation time and the remaining service time T serve The state information of the intermediate node is determined based on at least the state information of the user equipment and the state information of the intermediate node.

[0278] 17. A user equipment side electronic device for federated learning, comprising a processing circuit, wherein the processing circuit is configured to:

[0279] Receive a handover decision for the user equipment from the network, wherein the handover decision is based on a model aggregation time and a remaining service time T for the user equipment. serve The request is made when a predetermined condition is met, and the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node;

[0280] A handover is performed based on the handover decision.

[0281] 18. The electronic device according to embodiment 17, wherein the processing circuit is further configured to:

[0282] Use local data to train local models;

[0283] sending the local model to the network or the intermediate node; and

[0284] receiving a global aggregate model from the network;

[0285] The local model is updated to the global aggregate model.

[0286] 19. The electronic device according to embodiment 17, wherein the processing circuit is further configured to:

[0287] sending the status information of the user equipment to the network or the intermediate node,

[0288] Among them, the model aggregation time and the remaining service time T serve The status information is determined based on at least the user equipment.

[0289] 20. A network-side method for federated learning, comprising:

[0290] Determine the model aggregation time and the remaining service time T for the user device serve , wherein the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node;

[0291] At the model aggregation time and the remaining service time T serve When a predetermined condition is met, making a handover decision for the user equipment; and

[0292] The handover decision for the user equipment is sent.

[0293] 21. An intermediate node side method for federated learning, comprising:

[0294] A handover decision for the user equipment is received from the network, wherein the handover decision is received at a model aggregation time and a remaining service time T for the user equipment. serve The connection is made when a predetermined condition is met, and the user equipment is indirectly connected to the network via the intermediate node;

[0295] The handover decision is sent to the user equipment.

[0296] 22. A user device-side method for federated learning, comprising:

[0297] Receive a handover decision for the user equipment from the network, wherein the handover decision is based on a model aggregation time and a remaining service time T for the user equipment. serve The request is made when a predetermined condition is met, and the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node;

[0298] A handover is performed based on the handover decision.

[0299] 23. A non-transitory computer-readable storage medium having program instructions stored thereon, which, when executed by a processor, cause the processor to perform the method according to any one of embodiments 20 to 22.

[0300] 24. A computer program product comprising program instructions, which, when executed by a processor, cause the processor to perform the method according to any one of embodiments 20 to 22.

Claims

1. A network-side electronic device for federated learning, comprising a processing circuit, wherein the processing circuit is configured to: Determine the model aggregation time and the remaining service time T for the user device serve , wherein the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node; At the model aggregation time and the remaining service time T serve When a predetermined condition is met, making a handover decision for the user equipment; and The handover decision for the user equipment is sent.

2. The electronic device according to claim 1, wherein The processing circuit is further configured to: receiving an intermediate aggregation model from the intermediate node; generating a global aggregation model based at least on the intermediate aggregation model; and The global aggregate model is broadcasted.

3. The electronic device according to claim 1, wherein: The processing circuit is further configured to receive state information of the user equipment, and the model aggregation time and the remaining service time T serve The status information is determined based on at least the user equipment.

4. The electronic device according to claim 3, wherein: The status information of the user equipment includes one or more of the following: channel status, computing capability, local data information, power level, location, and movement information.

5. The electronic device according to claim 3, wherein: The processing circuit is further configured to receive the state information of the intermediate node, and the model aggregation time and the remaining service time T serve The determination is also based on the status information of the intermediate node.

6. The electronic device according to claim 5, wherein: The status information of the intermediate node includes one or more of the following: channel status, computing capability, location, and movement information.

7. The electronic device according to claim 1, wherein: The model aggregation time includes the remaining time T1 required for this global aggregation and the time T2 required for the next global aggregation, and If T1 < T serve < T1 + T2, a handover decision for the user equipment is made, and the handover decision indicates that the user equipment performs a handover after the end of the current global aggregation.

8. The electronic device according to claim 7, wherein: If T serve < T1, then the processing circuit is further configured to estimate an increased remaining service time T′ assuming execution of one or more of the following individual operations serve : increasing the transmit power of one or more of the user equipment, the intermediate node, and the global node; allocating more transmission resources to one or both of the user equipment and the intermediate node; and reducing the RSRP threshold of one or more of the user equipment, the intermediate node, and the global node, If T′ serve >T1, then perform the one or more operations and make a handover decision for the user equipment, where the handover decision instructs the user equipment to perform handover after the current global aggregation is completed.

9. The electronic device according to claim 1 or 7, wherein: The model aggregation time includes the remaining time t1 required for the current intermediate aggregation at the intermediate node and the time t2 required for the next intermediate aggregation, and If t1 < T serve <and t1 + t2, a handover decision for the user equipment is made, and the handover decision indicates that the user equipment performs a handover after the current intermediate aggregation ends.

10. The electronic device according to claim 9, wherein: If T serve <is less than t1, a handover decision for the user equipment is made, and the handover decision instructs the user equipment to perform a handover immediately.

11. The electronic device according to claim 9, wherein: If T serve <is less than t1, the processing circuit is further configured to estimate an increased remaining service time T' assuming execution of one or more of the following respective operations serve : increasing the transmit power of one or more of the user equipment, the intermediate node, and the global node; allocating more transmission resources to one or both of the user equipment and the intermediate node; and reducing the RSRP threshold of one or more of the user equipment, the intermediate node, and the global node, If T′ serve >t1, then perform the one or more operations and make a handover decision for the user equipment, where the handover decision instructs the user equipment to perform handover after the intermediate aggregation is completed.

12. The electronic device according to claim 9, wherein: If the user equipment is handed over to another intermediate node, and there is no intermediate aggregation model at the other intermediate node, the intermediate aggregation model at the intermediate node is sent to the other intermediate node.

13. An electronic device at an intermediate node side for federated learning, comprising a processing circuit, wherein the processing circuit is configured to: A handover decision for the user equipment is received from the network, wherein the handover decision is received at a model aggregation time and a remaining service time T for the user equipment. serve The connection is made when a predetermined condition is met, and the user equipment is indirectly connected to the network via the intermediate node; The handover decision is sent to the user equipment.

14. The electronic device according to claim 13, wherein: The processing circuit is further configured to: receiving a local model from the user equipment, generating an intermediate aggregate model based at least on the local model, and sending the intermediate aggregate model to the network; as well as A global aggregation model is received from the network, and the global aggregation model is sent to the user equipment.

15. The electronic device according to claim 13, wherein: The processing circuit is further configured to: receiving status information of the user equipment; sending the status information of the user equipment and the status information of the intermediate node to the network, Among them, the model aggregation time and the remaining service time T serve The state information of the intermediate node is determined based on at least the state information of the user equipment and the state information of the intermediate node.

16. The electronic device according to claim 13, wherein: The processing circuit is further configured to: receiving status information of the user equipment; determining at least a portion of the model aggregation time based at least on state information of the user equipment and state information of the intermediate node; sending the state information of the user equipment, the state information of the intermediate node, and the at least a portion of the model aggregation time to the network, Among them, the other part of the model aggregation time and the remaining service time T serve The state information of the intermediate node is determined based on at least the state information of the user equipment and the state information of the intermediate node.

17. A user equipment side electronic device for federated learning, comprising a processing circuit, wherein the processing circuit is configured to: Receive a handover decision for the user equipment from the network, wherein the handover decision is based on a model aggregation time and a remaining service time T for the user equipment. serve The request is made when a predetermined condition is met, and the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node; A handover is performed based on the handover decision.

18. The electronic device according to claim 17, wherein: The processing circuit is further configured to: Use local data to train local models; sending the local model to the network or the intermediate node; and receiving a global aggregate model from the network; The local model is updated to the global aggregate model.

19. The electronic device according to claim 17, wherein: The processing circuit is further configured to: sending the status information of the user equipment to the network or the intermediate node, Among them, the model aggregation time and the remaining service time T serve is based at least on the user device Status information is determined.

20. A network-side method for federated learning, comprising: Determine the model aggregation time and the remaining service time T for the user device serve , wherein the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node; At the model aggregation time and the remaining service time T serve When a predetermined condition is met, making a handover decision for the user equipment; as well as The handover decision for the user equipment is sent.

21. An intermediate node side method for federated learning, comprising: A handover decision for the user equipment is received from the network, wherein the handover decision is received at a model aggregation time and a remaining service time T for the user equipment. serve The connection is made when a predetermined condition is met, and the user equipment is indirectly connected to the network via the intermediate node; The handover decision is sent to the user equipment.

22. A user device-side method for federated learning, comprising: Receive a handover decision for the user equipment from the network, wherein the handover decision is based on a model aggregation time and a remaining service time T for the user equipment. serve The request is made when a predetermined condition is met, and the user equipment is directly connected to the network or indirectly connected to the network via an intermediate node; A handover is performed based on the handover decision.

23. A non-transitory computer-readable storage medium having program instructions stored thereon, which, when executed by a processor, cause the processor to perform the method according to any one of claims 20 to 22.

24. A computer program product comprising program instructions which, when executed by a processor, cause the processor to perform the method according to any one of claims 20 to 22.

Images