Device and method for identifying models of wireless communication systems
By optimizing the two-sided model in the wireless communication system, reducing the frequency and amount of data updates, and using neural network training models, the problems of power consumption and excessive processor utilization caused by data updates are solved, thereby improving system performance.
Patent Information
- Application Number
- CN202580024333.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-04-04
- Filing Date
- 2025-03-19
- Publication Date
- 2026-11-17
AI Technical Summary
In existing wireless communication systems, the excessively high data update frequency and quantity of bilateral models lead to increased power consumption and processor utilization, affecting system performance.
By reducing the number of data updates or lowering the update frequency, the encoding and decoding process is optimized. The encoder and decoder devices using a two-sided model manage parameters, and the model is trained and updated using a neural network module, thereby reducing data transmission, power consumption, and processor utilization.
It improves the overall performance of wireless communication systems, reduces power consumption and processor utilization, and enhances system efficiency.
Smart Images

Figure CN123125812A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to wireless communication, and more specifically to the identification of models of wireless communication systems. Background Technology
[0002] A wireless communication system may include one or more network communication devices (e.g., base stations) that support wireless communication with one or more user communication devices (which may otherwise be referred to as user equipment (UE) or other suitable terms). The wireless communication system can support wireless communication with one or more user communication devices by utilizing its own resources (e.g., time resources (e.g., symbols, time slots, subframes, frames, etc.) or frequency resources (e.g., subcarriers, carriers, etc.)). Furthermore, the wireless communication system can support wireless communication across various radio access technologies, including third-generation (3G), fourth-generation (4G), fifth-generation (5G), and other suitable radio access technologies besides 5G (e.g., sixth-generation (6G)). Summary of the Invention
[0003] The article “a(a)” preceding an element is not limited and should be understood to mean “at least one” or “one or more” of those elements. The terms “a,” “at least one,” “one or more,” and “at least one of one or more” are interchangeable. As used herein and contained in the claims, the use of “or” in a list of items (e.g., a list of items ending with phrases such as “at least one of…”, “one or more of…”, or “one or both of…”) indicates a comprehensive list, such that (for example) a list of at least one of A, B, or C means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, as used herein, the phrase “based on” should not be construed as a reference to a set of closing conditions. For example, an example step described as “based on condition A” may be based on both condition A and condition B without departing from the scope of this disclosure. In other words, as used herein, the phrase “based on” should be interpreted in the same manner as the phrase “at least partially based on.” Furthermore, as used herein and included in the claims, "group" may comprise one or more elements.
[0004] Various aspects of this disclosure relate to wireless communication, including improved methods and apparatus for supporting the identification of models (e.g., encoders and decoders) of wireless communication systems. A first device (e.g., a base station, a UE, a node of a base station, or a node of a UE, etc.) can determine the first set of information based on an input sample and a first index associated with the input sample or a first set of information. The first device can determine the input sample. Alternatively or additionally, the first device can receive the input sample, for example, from another device (e.g., a base station, a UE, a node of a base station, or a node of a UE, etc.). The first index can be associated with at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device. The first device can generate one or more datasets, each of which is a subset of the first set of information. The first device can also transmit at least one of the one or more datasets. Attached Figure Description
[0005] Figure 1 Illustrated examples of wireless communication systems according to various aspects of this disclosure.
[0006] Figure 2 Illustrated examples of wireless networks according to various aspects of this disclosure.
[0007] Figure 3 Illustrated examples of block diagrams illustrating various aspects of the two-sided model according to this disclosure.
[0008] Figure 4 Illustrated examples of UEs based on various aspects of this disclosure.
[0009] Figure 5 Illustrated examples of processors according to various aspects of this disclosure.
[0010] Figure 6 Illustrated examples of network equipment (NE) according to various aspects of this disclosure.
[0011] Figure 7 The diagram illustrates a flowchart of a method performed by a first device according to various aspects of this disclosure.
[0012] Figure 8 The diagram illustrates a flowchart of a method performed by a second device according to various aspects of this disclosure.
[0013] Figure 9 The diagram illustrates a flowchart of a method performed by a third device according to various aspects of this disclosure. Detailed Implementation
[0014] Various aspects of this disclosure relate to a two-sided model supporting (e.g., configuring, enabling) a wireless communication system. The two-sided model may be associated with an encoder of a first wireless device (which may be referred to as an encoder device) and a decoder of either the first or second wireless device (which may be referred to as a decoder device). Some functions of the model (e.g., operation, behavior, characteristics, constraints) may be performed by the first wireless device (e.g., the encoder device), and other functions of the model (e.g., operation, behavior, characteristics, constraints) may be performed by either the first or second wireless device (e.g., the decoder device). In some embodiments, one or more of the first wireless device (e.g., the encoder device) or the second wireless device (e.g., the decoder device) may manage (e.g., update, adjust, modify) a set of one or more parameters of the two-sided model at a frequency (e.g., rate, type, interval). However, in some cases, excessive data may be used, for example, based on the amount (e.g., quantity) of data transmitted to update the two-sided model and the frequency (e.g., rate) of the update.
[0015] A device (e.g., a wireless device) that performs one or more of the encoding or decoding processes can reduce power consumption, reduce processor utilization, reduce data utilization, and improve overall system performance by reducing one or more of the transmission (including the amount of data used to update the bilateral model, e.g., quantity) or reducing the frequency (e.g., rate) of updating the bilateral model.
[0016] Various aspects of this disclosure are described in the context of wireless communication systems.
[0017] Figure 1 The illustration depicts examples of a wireless communication system 100 according to various aspects of this disclosure. The wireless communication system 100 may include one or more NEs 102, one or more UEs 104, and a core network (CN) 106. The wireless communication system 100 may support various radio access technologies. In some embodiments, the wireless communication system 100 may be a 4G network, such as an LTE network or an advanced LTE (LTE-A) network. In some other embodiments, the wireless communication system 100 may be a new radio (NR) network, such as a 5G network, an advanced 5G (5G-A) network, or a 5G ultra-wideband (5G-UWB) network. In other embodiments, the wireless communication system 100 may be a combination of 4G and 5G networks, or include other suitable radio access technologies such as IEEE 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), and IEEE 802.20. The wireless communication system 100 may support radio access technologies other than 5G (e.g., 6G). In addition, the wireless communication system 100 can support technologies such as Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), or Code Division Multiple Access (CDMA).
[0018] One or more NEs 102 may be geographically dispersed to form a wireless communication system 100. One or more of the NEs 102 described herein may be, include, or be referred to as a network node, base station, network element, network function, network entity, radio access network (RAN), NodeB, eNodeB (eNB), next-generation NodeB (gNB), or other suitable terms. NEs 102 and UEs 104 may communicate via a communication link, which may be a wireless or wired connection. For example, NEs 102 and UEs 104 may perform wireless communication (e.g., receiving signaling, transmitting signaling) via a Uu interface.
[0019] NE 102 can provide a geographic coverage area, and NE 102 can support service for one or more UEs 104 within said geographic coverage area. For example, NE 102 and UE 104 can support wireless communication of signals related to services (e.g., voice, video, packet data, messaging, broadcasting, etc.) according to one or more radio access technologies. In some embodiments, NE 102 can be mobile, such as a satellite associated with an NTN. In some embodiments, different geographic coverage areas associated with the same or different radio access technologies can overlap, but different geographic coverage areas can be associated with different NEs 102.
[0020] One or more UEs 104 may be geographically dispersed throughout the wireless communication system 100. UE 104 may include or be referred to as a remote unit, mobile device, wireless device, remote device, subscriber device, transmitter device, receiver device, or some other suitable term. In some implementations, UE 104 may be referred to as a unit, station, terminal, or client, and other instances thereof. Alternatively or additionally, UE 104 may be referred to as an Internet of Things (IoT) device, Internet of Everything (IoE) device, or Machine-Type Communication (MTC) device, and other instances thereof.
[0021] UE 104 may be able to support direct wireless communication with other UE 104s via a communication link. For example, UE 104 may support direct wireless communication with another UE 104 via a device-to-device (D2D) communication link. In some implementations (such as vehicle-to-vehicle (V2V) deployments, vehicle-to-everything (V2X) deployments, or cellular V2X deployments), the communication link may be referred to as a sidelink. For example, UE 104 may support direct wireless communication with another UE 104 via a UE-to-UE interface (PC5 interface).
[0022] NE 102 may support communication with CN 106 or with another NE 102, or both. For example, NE 102 may interface with other NE 102 or CN 106 via one or more backhaul links (e.g., S1, N2, N2, or network interfaces). In some embodiments, NE 102 may communicate directly with each other. In some other embodiments, NE 102 may communicate with each other or indirectly (e.g., via CN 106). In some embodiments, one or more NE 102 may include sub-components such as an access network entity, which may be an instance of an Access Node Controller (ANC). The ANC may communicate with one or more UE 104 via one or more other access network transport entities (which may be referred to as a radio head, smart radio head, or transmit-receive point (TRP)).
[0023] CN 106 can support user authentication, access authorization, tracking, connectivity, and other access, routing, or mobility functions. CN 106 can be an evolved packet core (EPC) or a 5G core (5GC), which may include control plane entities that manage access and mobility (e.g., a mobility management entity (MME), access and mobility management function (AMF)) and user plane entities that route packets to or interconnect to external networks (e.g., a serving gateway (S-GW), packet data network (PDN) gateway (P-GW), or user plane function (UPF)). In some implementations, the control plane entity may manage non-access plane (NAS) functions, such as mobility, authentication, and bearer management (e.g., data bearers, signaling bearers, etc.), for one or more UEs 104 served by one or more NEs 102 associated with CN 106.
[0024] CN 106 may communicate with a packet data network via one or more backhaul links (e.g., via S1, N2, N2, or another network interface). The packet data network may contain an application server. In some implementations, one or more UEs 104 may communicate with the application server. UE 104 may establish a session (e.g., a Protocol Data Unit (PDU) session, etc.) with CN 106 via NE 102. CN 106 may use the established session (e.g., an established PDU session) to route services (e.g., control information, data, etc.) between UE 104 and the application server. A PDU session may be an instance of a logical connection between UE 104 and CN 106 (e.g., one or more network functions of CN 106).
[0025] In the wireless communication system 100, NE 102 and UE 104 can use the resources of the wireless communication system 100 (e.g., time resources (e.g., symbols, time slots, subframes, frames, etc.) or frequency resources (e.g., subcarriers, carriers)) to perform various operations (e.g., wireless communication). In some embodiments, NE 102 and UE 104 can support different resource structures. For example, NE 102 and UE 104 can support different frame structures. In some embodiments, such as in 4G, NE 102 and UE 104 can support a single frame structure. In some other embodiments, such as in 5G and other suitable radio access technologies, NE 102 and UE 104 can support various frame structures (i.e., multiple frame structures). NE 102 and UE 104 can support various frame structures based on one or more parameter sets (numerology).
[0026] The wireless communication system 100 may support one or more parameter sets, and the parameter sets may include subcarrier spacing and cyclic prefixes. A first parameter set (e.g., μ=0) may be associated with a first subcarrier spacing (e.g., 15 kHz) and a regular cyclic prefix. In some embodiments, the first parameter set (e.g., μ=0) associated with the first subcarrier spacing (e.g., 15 kHz) may utilize one time slot per subframe. A second parameter set (e.g., μ=1) may be associated with a second subcarrier spacing (e.g., 30 kHz) and a regular cyclic prefix. A third parameter set (e.g., μ=2) may be associated with a third subcarrier spacing (e.g., 60 kHz) and a regular or extended cyclic prefix. A fourth parameter set (e.g., μ=3) may be associated with a fourth subcarrier spacing (e.g., 120 kHz) and a regular cyclic prefix. A fifth parameter set (e.g., μ=4) may be associated with a fifth subcarrier spacing (e.g., 240 kHz) and a regular cyclic prefix.
[0027] The time intervals of resources (e.g., communication resources) can be organized according to frames (also called radio frames). Each frame may have a duration, for example, 10 milliseconds (ms). In some embodiments, each frame may contain multiple subframes. For example, each frame may contain 10 subframes, and each subframe may have a duration, for example, 1 ms. In some embodiments, each frame may have the same duration. In some embodiments, each subframe of a frame may have the same duration.
[0028] Alternatively or concurrently, the time intervals of resources (e.g., communication resources) can be organized according to time slots. For example, a subframe may contain a certain number (e.g., quantity) of time slots. The number of time slots in each subframe may also depend on one or more parameter sets supported in the wireless communication system 100. For example, the first, second, third, fourth, and fifth parameter sets (i.e., μ=0, μ=1, μ=2, μ=3, μ=4) associated with corresponding subcarrier spacings of 15 kHz, 30 kHz, 60 kHz, 120 kHz, and 240 kHz can respectively utilize one time slot per subframe, two time slots per subframe, four time slots per subframe, eight time slots per subframe, and 16 time slots per subframe. Each time slot may contain a certain number (e.g., quantity) of symbols (e.g., Orthogonal Frequency Division Multiplexing (OFDM) symbols). In some embodiments, the number (e.g., quantity) of time slots in a subframe may depend on the parameter set. For a conventional cyclic prefix, a time slot may contain 14 symbols. For an extended cyclic prefix (e.g., applicable to a 60 kHz subcarrier spacing), a time slot may contain 12 symbols. The relationship between the number of symbols per time slot, the number of time slots per subframe, and the number of time slots per frame for both regular and extended cyclic prefixes may depend on the parameter set. It should be understood that a reference to the first parameter set (e.g., μ=0) associated with the first subcarrier spacing (e.g., 15 kHz) may be used interchangeably between subframes and time slots.
[0029] In the wireless communication system 100, the electromagnetic (EM) spectrum can be divided into various categories, frequency bands, channels, etc., based on frequency or wavelength. By way of example, the wireless communication system 100 can support one or more operating frequency bands, such as frequency ranges specified as FR1 (410 MHz – 7.125 GHz), FR2 (24.25 GHz – 52.6 GHz), FR3 (7.125 GHz – 24.25 GHz), FR4 (52.6 GHz – 114.25 GHz), FR4a or FR4-1 (52.6 GHz – 71 GHz), and FR5 (114.25 GHz – 300 GHz). In some embodiments, NE 102 and UE 104 can perform wireless communication via one or more of the said operating frequency bands. In some embodiments, NE 102 and UE 104, as well as other equipment or devices, can use FR1 for cellular communication services (e.g., control information, data). In some implementations, NE 102 and UE 104, as well as other equipment or devices, may use FR2 for short-range, high-data-rate capabilities.
[0030] FR1 can be associated with one or more parameter sets (e.g., at least three parameter sets). For example, FR1 can be associated with: a first parameter set (e.g., μ=0) containing a 15 kHz subcarrier spacing; a second parameter set (e.g., μ=1) containing a 30 kHz subcarrier spacing; and a third parameter set (e.g., μ=2) containing a 60 kHz subcarrier spacing. FR2 can be associated with one or more parameter sets (e.g., at least two parameter sets). For example, FR2 can be associated with: a third parameter set (e.g., μ=2) containing a 60 kHz subcarrier spacing; and a fourth parameter set (e.g., μ=3) containing a 120 kHz subcarrier spacing.
[0031] Figure 2 The diagram illustrates an example of a wireless network 200 according to various aspects of this disclosure. The wireless network 200 may include NE102-a (e.g., an embodiment of NE 102, gNB), a first UE 104-a (e.g., an embodiment of UE 104, UE1), a second UE 104-b (e.g., an embodiment of UE 104, UE2), and a third UE 104-c (e.g., an embodiment of UE 104, UE3). K (The third UE 104-c represents any number of additional UEs).
[0032] Specifically Figure 2 The demonstration features one NE 102-a (e.g., represented by node B1 and equipped with M antennas) and K UEs 104-a, 104-b, 104-c (e.g., represented by nodes U1, U2, ..., Uc, each with N antennas). k This is an example of a wireless network 200 (represented by [reference]). It can represent B1 and U k A channel at time t in frequency band l (l∈{1,2,…,L}) can form a matrix of size N×M with complex terms, i.e., .
[0033] At time t and frequency band l, it can be assumed that NE 102-a wants to send a signal to user U. k Transmitting messages (where k={1,2,…,K}), and it uses As a pre-coding vector. U k The received signal at the location Can be written as:
[0034] ,
[0035] in This represents the noise vector at the receiver. In one instance, the transmit (Tx)-receive (Rx) pair could be: , .
[0036] To improve the achievable rate of the link, in one instance, the NE 102-a selects a method that maximizes the received signal-to-interference-plus-noise ratio (SINR). Several different configurations are available for selection. Some of the configurations described therein may be related to I have some understanding of it.
[0037] NE 102-a can be understood by direct measurement (e.g., in Time Division Duplex (TDD) mode and assuming channel reciprocity) or indirectly by using information sent to NE 102-a by one of UEs 104-a, 104-b, and 104-c (e.g., in Frequency Division Duplex (FDD) mode). In the indirect approach, a large amount of feedback can be used to send information about... Accurate information is needed. This can require a large amount of data, especially in situations with numerous antennas and / or a large frequency band.
[0038] In one instance, a single timeslot is analyzed, but this instance can be scaled to instances with more than a single timeslot. Without loss of generality, Usable express.
[0039] H k (t) can be defined as the sum of all frequency bands stacked. (For example, H) k The term at [n,m,l](t) equals The matrix formed by [n,m](t) is of size N×M×L. In general, each of UE 104-a, 104-b and 104-c can provide NE102-a with feedback information about the most recent N×M×L complex numbers.
[0040] A two-sided model, described in this paper, is used to reduce the required feedback information. The encoding part (e.g., at the UE) computes a quantized latent representation of the input data, and the decoding part (e.g., at the gNB) obtains this latent representation and uses it to reconstruct the desired output. The input data can be data based on channel measurements. For example, it could be H... k or The original channel input, or, for example, the precoder calculated based on the channel matrix (e.g., with H for each subband). k The eigenvectors associated with the largest eigenvector.
[0041] Figure 3The diagram illustrates an example of a block diagram of a two-sided model 300 (e.g., an NN-based model) according to various aspects of this disclosure. The two-sided model 300 includes nodes A 302 (e.g., encoder, encoding model, Me) and nodes B 304 (e.g., decoder, decoding model, Md). Figure 3 This illustration only depicts one instance of the two-sided model 300; in other instances, the encoder and decoder positions may be interchanged. Input data 306 is provided to node A 302, followed by a latent representation 308 in communication from node A 302 to node B 304, and node B 304 outputs data 310. In some configurations, node A 302 is located at UE 104, and node B 304 is located at NE 102; however, in other configurations, node A 302 and node B 304 may be located in other devices.
[0042] In some systems, M e and M d It can be used in logical models. That is, node A 302 can choose not to directly use M. e Instead, it develops and uses several internal models based on some private parameters and decisions, such as M. e(1) M e(2) ... M e(k) In these cases, one assumption can be that all these internal models should at least satisfy the original M. e With M d Performance during pairing. Under this interpretation, node B 304 does not need to be concerned with the actual model used at node A 302. For M... d This also applies to some internal models at node B 304.
[0043] Several methods exist for training the neural network (NN) modules of a two-sided model, including centralized training, simultaneous training, and individual training. Similarly, updating the two-sided model can be performed centrally on one entity, simultaneously on different entities, or individually. In individual training and / or model updates, the NN modules of node A 302 (e.g., UE) and node B 304 (e.g., gNB) can be trained in different training sessions (e.g., there is no forward or backward propagation path between the two parts). One advantage of individual training is that nodes A 302 and B 304 do not need to know the internal structure of the NN module on the other side.
[0044] Node A 302 and Node B 304 may decide not to use a single encoder-decoder pair, but instead construct different encoder-decoder pairs suitable for different conditions of Node A 302 and / or Node B 304. This is because samples in different states may have different statistical properties compared to other settings, and therefore, it is reasonable to construct two different encoder-decoder models for these two scenarios rather than attempting to train a single model that generalizes well to different settings. For example, for a Channel State Information (CSI) feedback system, having separate encoder-decoder pairs for the two environments is advantageous compared to constructing a single encoder-decoder pair covering both unlicensed mobile access (UMA) and indoor environments. Therefore, it can be assumed that the design M... e [i]-M d [i] (i={1,2,…,K}) can be beneficial, where the i-th pair is trained based on training data collected when node A 302 and node B 304 are in a state or a set of states. A state can refer to the situation when the parameters of node A 302 and node B 304 are in a specific state.
[0045] If multiple models exist, it may be difficult to distinguish between them and / or to activate / deactivate them. Furthermore, nodes A 302 and B 304 may be unwilling to disclose certain parameters that still affect their status.
[0046] In some embodiments described herein, a scheme may exist in which nodes involved in training and inference can exchange information that enables consistency between the state in which the model is trained and the state in which it will be used for inference, without sending sensitive information from one side to the other. For example, a UE may create separate models for situations where it has low battery and is outdoors, low battery and is indoors, high battery and is outdoors, and high battery and is indoors, but it may be unwilling to inform the other side whether its battery level is low or high at any stage of lifecycle management (LCM) (e.g., data collection, model training, model inference).
[0047] One way to differentiate between different models developed under different conditions is to determine the model's identifier (ID) when the UE and / or gNB develop the model. This ID can be used to activate and deactivate the model. The disadvantages of this approach include the need to develop several different models for different operating modes and their combinations (since some models can be developed to support more than one operating mode). There are also complexities regarding how to exchange information between nodes and how to associate nodes with model IDs to keep the internal state of a node transparent to the other side.
[0048] Some systems may use dataset IDs. This could mean assigning an ID to each dataset whose samples have been used and collected for model training, and then constructing a model ID based on the dataset ID. In such systems, it may not be entirely certain how the conditions of node A and node B can be considered when assigning dataset IDs. Furthermore, a dataset may contain samples from different UE and / or gNB conditions; therefore, if a single ID is assigned to a dataset, creating different models for it based on associated samples from different UE and / or gNB conditions may not be straightforward.
[0049] The examples in this paper include model identification for a given dataset (or subset thereof) of multiple models (e.g., different model architectures, different complexities, different model parameterizations, different input-output parameter relationships, different preprocessing of input data, different model outputs (e.g., size, dimensionality), different quantizations, etc.) based on the conditions and / or additional conditions of a first node (e.g., node A 302) and / or a second node (node B 304). The conditions and / or additional conditions of node A and / or node B may include parameters based on model characteristics or characteristics used to identify the model.
[0050] Although the description in this disclosure is primarily concerned with two-sided models, the techniques are also applicable to one-sided models.
[0051] In one instance, training a two-sided model M j (j={1,2,…,K}), where each of the models is designed for a set of conditions for node A and node B. Each model may contain encoder models residing at node A and node B respectively. and decoder model It should be noted that the model may be located at a single-sided node (e.g., where no node exists). or (Partially). Furthermore, there may be configurations where the model depends only on one side of the situation (e.g., only node A or node B). The examples in this paper are also applicable to these configurations. It is also possible to train different models for the same situation of node A and node B. For example, both a high-complexity model and a low-complexity model may be trained based on a single dataset collected under specific situations of node A and node B.
[0052] The status of Node A and Node B may be related to some status and / or parameters of each of Node A and Node B, and may also depend on other nodes in the network. These status and / or parameters may be transmitted during UE capability reporting communications, UE configuration, measurement configuration and / or reporting, and during initialization, joining and / or synchronization.
[0053] One example assumes the UE is located indoors, equipped with two Rx antenna ports and an antenna spacing δ. On the gNB side, the gNB may be equipped with 32 Tx antenna ports on two panels, each of which is a virtualization of four antenna elements.
[0054] As used herein, a status parameter may correspond to information that can be shared between the UE and NW sides, wherein the information is parameterized as one or more parameters. Depending on the specific instance, the number of antenna ports at the gNB and UE may be possible status parameters. It should be noted that status parameters can also be referred to as metadata.
[0055] Furthermore, state information can correspond to information that cannot be shared between the UE and NW sides, such as gNB antenna virtualization and Rx antenna spacing. More generally, both shareable and non-shareable information can be correlated with state information.
[0056] In some instances, in order to effectively train, pair, and activate two-sided models, there may be a data collection phase, a training phase, and an inference phase.
[0057] During the data collection phase, node A can perform data collection tasks, and the data collection process can access subsets of the states of nodes A and B. and This indicates that actual condition names and / or values can be used, or virtual names and / or values of node A and / or node B can be used instead of actual condition names and / or values. In some instances, a node-specific condition ID indicates a set of node parameters and associated values (e.g., actual condition parameter names and / or values). In some instances, a local condition index and / or ID is associated with a set of local parameters (and values) at the node (e.g., virtual condition parameter names and / or values, local condition parameters based on model characteristics or characteristics used to identify the model, etc.). In some instances, a node-specific condition ID may also indicate a local condition ID.
[0058] Furthermore, during the data collection phase, the local status ID can be shared and / or passed to another node (e.g., node B) to indicate the local status at node A, without sharing a set of local parameters. Assuming node A is collecting data, it can also be assumed that information about subsets of the status of nodes A and B can be passed to node A. In one instance, at least a subset of the status of node B is passed to node A before the dataset transfer and / or delivery from node A. In another instance, at least a subset of the status of node B is passed to node A after the dataset transfer and / or delivery from node A. A similar explanation can be used when node B is collecting data.
[0059] Under certain assumptions, node A collects samples. As an example, the collected samples may be associated with state information of the channel between node A and node B. Each or each group of collected samples may be associated with a state parameter (e.g., a node-specific state ID and / or a node-local state ID) indicating the state of node A and / or node B. A mapping table can be used to associate IDs with possible combinations and / or pairs of states (the states of node A and / or node B), and then the IDs can be used as state parameters. The mapping table can be passed to the other side so that both sides have the same understanding. The collected samples may also be associated with timing information.
[0060] Next, the dataset collected by node A can be constructed as a union of one or more of the collected samples, and the metadata of the dataset can be determined based on the condition parameters of each or each group of the collected samples present in the dataset. It should be noted that a training dataset may contain samples collected under different conditions of node A and / or node B. The collected dataset can be further transferred to another node. Furthermore, the transferred dataset can be associated with an identifier.
[0061] During sample collection, node A and / or node B can record state information, i.e., S. A / S B The status information may include one or more of its own condition, status, and / or parameters, which may be invisible to the data collection process. In one instance, this set of conditions, status, and / or parameters may correspond to a set of local parameters (and values) and may be associated with a local condition index and / or ID.
[0062] In some instances, training of a two-sided model can be performed. The complete dataset can be constructed from data collected over one or more time periods and can be generated from a single or multiple collection nodes (e.g., node A in the previous section). It should be noted that each sample and / or group of samples in the complete dataset was collected under specific conditions at nodes A and B. This information and S A / S B Information (e.g., where accessible) can be used to create one or more training datasets for creating the first part of the model. Furthermore, the encoder and decoder models can be trained on a single node or on two different nodes.
[0063] For example, suppose we want to train the decoder part of the model (e.g., node B). We can assume that training the decoder part can occur when node B can share its state information S with it. B At the node (e.g., the same node B or a trusted physical and / or virtual node associated with node B). The training node can then use the metadata of each dataset and (where possible) SB The entire dataset can be split into several training datasets, each containing samples collected under similar (or nearly identical) conditions. This helps keep the samples in each training dataset statistically similar to each other, which can help obtain a better-converged, more accurate, or less complex model.
[0064] Next, each of the training datasets (which may contain one training dataset) can be used to train one or more models. Generally, this model can be a two-sided model, a one-sided model, or one side of a two-sided model.
[0065] Furthermore, a single training dataset can be used to train more than one model. For example, set D1 can be used to train a first two-sided model when the encoder and decoder backbones are based on a convolutional neural network architecture, and the same set D1 can be used to train a second two-sided model when the encoder and decoder backbones are based on a transformer network.
[0066] The metadata associated with each model can be based on the metadata of the dataset used during training, state information S A / S B (Where available) and the parameters of the model are used to determine these parameters. These parameters can directly or indirectly represent different models constructed from the same training dataset (e.g., the model backbone).
[0067] In one embodiment, the training dataset can be used to train a full two-sided model. In such a scenario, the full model is available. It should be noted that while the two-sided model is primarily trained using the training dataset, it is also possible to consider that the trained model can satisfy certain requirements (e.g., imposed by another entity). For example, there may be a requirement that the encoder model's output and the predetermined decoder also produce acceptable performance, or a requirement that the decoder can decode the predetermined encoder's output and produce acceptable results. One way to take these additional requirements into account is to include them as weights in the loss function during training.
[0068] In another embodiment, the training dataset can be used to train only one side of a two-sided model. For example, there may be a first case and a second case.
[0069] In the first scenario, each element in the training dataset can be used to train one or more decoder models. While the decoder models are primarily trained using the training dataset, it is also possible to consider that the trained models may satisfy certain requirements (e.g., imposed by another entity). For example, there may be requirements that the trained decoder model also produces acceptable performance for the predetermined encoder, or that the input and / or output of the trained decoder model may resemble that of the predetermined decoder model. One way to take these additional requirements into account is to include them as weights in the loss function during training.
[0070] In the second scenario, each of the training datasets can be used to train one or more encoder models. While the encoder models are primarily trained using the training dataset, the trained models can also satisfy certain requirements imposed by another entity. For example, there may be requirements that the trained encoder model and the predetermined decoder also produce acceptable performance, or that the input and / or output of the trained encoder model can be similar to that of the predetermined encoder model. One way to take these additional requirements into account is to include them as weights in the loss function during training.
[0071] In both of these cases, it may be necessary to train another part of the two-sided model.
[0072] In the first scenario, one or more decoder models may have been trained, and metadata may be associated with each of these decoder models. Assuming the encoder model (associated with each decoder model) is to be trained on individual nodes (e.g., nodes other than node B or the node on which the decoder model is trained), then the following different options exist:
[0073] In the first option, information about the trained decoder model (or a lower-complexity decoder model based on the trained decoder model, for example, using transfer learning to train a smaller decoder model or a reference decoder model based on the trained decoder model, which may be a model supported by another node) is transferred to another node that is training the encoder model.
[0074] In the second option, since a local encoder model may need to be trained during the training of the decoder model, information about the trained local encoder model (or a lower-complexity encoder model based on the local encoder model, for example, using transfer learning to train a smaller encoder model or a reference encoder model based on the trained local encoder model, which may be a model supported by another node) can be sent to another node training the encoder model. The metadata of the local encoder model (or the lower-complexity encoder model, or the reference encoder model) can be determined based on the metadata of the trained decoder model. In one implementation, it may obtain the same metadata. It should be noted that the local encoder model may have a different structure than the actual encoder model.
[0075] In some instances, the transmission of information about the model can take different alternatives:
[0076] In the first alternative, there may be a transfer of the actual model (both its structure and trained parameters) and (if necessary) its associated metadata. In the second alternative, assuming the model structure is known on the other side (e.g., a normalized model), then there may be a transfer of only the trained parameters and (if necessary) their associated metadata.
[0077] In a third alternative, the transfer of a second model (both its structure and trained parameters) and (if necessary) its associated metadata is possible, wherein this model is trained to have the same input and / or output relationships as the trained model. This option can be useful without revealing the actual trained model that can be used on the other side, but still provides this information about how the model works. Furthermore, a teacher-student scheme can be used to train the second model using the actual trained model. The associated metadata of the second model can be determined based on the metadata of the actual trained model. In one implementation, the same metadata can be obtained.
[0078] In the fourth alternative, assuming the structure of the second model is known on the other side, only the parameters of the second model can be transmitted to the other side, and its associated metadata can be transmitted (if needed).
[0079] In the fifth alternative, a dataset containing samples of the inputs and / or outputs or expected outputs of the model can be generated, and the generated dataset can then be transmitted, along with its associated metadata (if necessary). The metadata associated with the generated dataset can be determined based on the model's metadata. In one implementation, the same metadata as the model used to generate the dataset can be obtained.
[0080] In the first option, the model's input is related to the latent representation of the input data (based on the training dataset of the trained decoder model) (generated using, for example, the output of the local encoder model), and the output is the output of the trained decoder model (e.g., in the case of compression, the output of the trained decoder model may be based on the training dataset).
[0081] In the second option, the model's input data is based on the training dataset of the trained decoder model, and the output data is based on the output of the trained local encoder model.
[0082] In some instances, the transmitted dataset may be associated with identification parameters and / or numbers.
[0083] To reduce overhead, instead of transmitting the entire generated dataset, a portion of the generated dataset can be indicated to another node based on previously exchanged and / or transmitted datasets. The dataset IDs of the previously exchanged datasets can be used to identify those datasets. In some instances, the dataset can be created based on a set of training dataset IDs. A latent representation (e.g., input in a first option, output in a second option) can be associated with output, latent, and / or compressed IDs. The dataset can be indicated based on at least one pair (e.g., training dataset ID, output, latent, and / or compressed IDs). In some instances, a training dataset ID can be associated with multiple output, latent, and / or compressed IDs (e.g., for multiple models). In some instances, an output, latent, and / or compressed ID can be associated with multiple training dataset IDs (e.g., for a model dataset constructed by aggregating multiple training datasets).
[0084] If the first option is used (e.g., sending information about the trained decoder model), then the node training the encoder model uses the information received for the decoder model to train its local decoder model. For example, it directly treats the received model as its decoder, uses a teacher-student approach to train its own decoder model using the received local decoder model, and uses the received dataset to train the decoder so that its input / output relationship is as similar as possible to the input / output relationship of the received dataset. The metadata of the trained local decoder model can be determined based on the received metadata of the model (e.g., using the first to fourth alternatives described herein) or the metadata of the generated dataset (e.g., the fifth alternative). In one implementation, it may obtain the same metadata as it receives from another node. The trained local decoder model can then be fixed, and the node trains the encoder model to match the trained local decoder model. The metadata of the trained encoder model can be determined based on the metadata of the trained local decoder model. In one implementation, it may obtain the same metadata as the trained local decoder model.
[0085] If the second option is used (e.g., sending information about the local encoder model), then the node training the encoder model uses the information received for the local encoder model to train its encoder model. For example, it can directly treat the received model as its encoder, use a teacher-student approach to train its own encoder model using the received encoder model, and / or use the received dataset to train the encoder such that its input / output relationship is as similar as possible to the input / output relationship of the received dataset. The metadata of the trained encoder model can be determined based on the received metadata of the model (e.g., using the first to fourth alternatives described herein) or the metadata of the generated dataset (e.g., the fifth alternative). In one implementation, it can obtain the same metadata as it receives from another node.
[0086] It should be noted that different backbone structures or different state information S A Training multiple encoder models is possible (where available). Therefore, the metadata for each trained encoder model can further contain information about its state S. A (Where applicable) information, and parameters about the model.
[0087] In a specific instance, both the encoder model and the decoder model are trained. It should be noted that the node training the encoder model can further transmit information about the trained encoder model and the metadata associated with it to the node training the decoder model. This information can be sent using any of the five alternatives described herein. The node with the decoder model can use this information to update / retrain its local encoder model, fix the local encoder model, and then update the decoder model based on the updated / retrained local encoder model. The updated / retrained decoder model can determine its associated metadata based on the metadata it receives from the other side. The process explained in the first scenario can be repeated several times and can be stopped based on stopping criteria.
[0088] It should be noted that combinations of the different instances described herein can be used. For example, a first alternative can be used for the first option, allowing the UE to have an initial encoder model. Then, if the performance of the resulting two-sided model is unsatisfactory (e.g., due to a poor match between the UE's input distribution and the received encoder), a third alternative for the second option can be used. The UE can then train, retrain, and / or update its encoder based on the received decoder model.
[0089] In the second scenario, one or more encoder models have been trained, and metadata can be associated with each of the encoder models. Assuming the decoder model (associated with each encoder model) is to be trained on individual nodes (e.g., nodes other than node A or the node on which the encoder model is trained), then the following different options exist:
[0090] In the first option, information about the trained encoder model can be transferred to another node that is training the decoder model.
[0091] In the second option, a local decoder model may also need to be trained during the training of the encoder model. Information about the trained local decoder model can then be sent to another node training the decoder model. The metadata of the local decoder model can be determined based on the metadata of the trained encoder model. In one implementation, it may obtain the same metadata. It should be noted that the local decoder model may have a different structure than the actual decoder model.
[0092] The transmission of information about the model can have different alternatives similar to those described in the first case. However, regarding the input and output of the model only in the fifth alternative, the following differences may exist:
[0093] In the first option, the model's input data is based on the training dataset of the trained encoder model, and the output data is based on the output of the trained encoder model. In the second option, the model's input is related to the latent representation generated by the trained encoder model using the input data (e.g., based on the training dataset used to train the encoder model), and the output is the expected output of the local decoder model.
[0094] If the first option is used (e.g., sending information about the trained encoder model), then the node training the decoder model uses the information received for the trained encoder model to train its local encoder model. For example, it could directly treat the received model as its local encoder, use a teacher-student approach to train its own local encoder model using the received encoder model, and / or use the received dataset to train the local encoder such that its input / output relationship is as similar as possible to the input / output relationship of the received dataset. The metadata of the trained local encoder model can be determined based on the received metadata of the model (e.g., the first to fourth alternatives) or the metadata of the generated dataset (e.g., the fifth alternative). In one implementation, it can obtain the same metadata as it receives from another node. The trained local encoder model can then be fixed, and the node can train a decoder model to match the trained local encoder model. The metadata of the trained decoder model can be determined based on the metadata of the trained local encoder model. In one implementation, it can obtain the same metadata as the trained local encoder model.
[0095] If the second option is used (e.g., sending information about the local decoder model), then the node training the decoder model can use the information received for the local decoder model to train its decoder model. For example, it can directly treat the received model as its decoder, use a teacher-student approach to train its own decoder model using the received decoder model, and / or use the received dataset to train the decoder such that its input / output relationship is as similar as possible to the input / output relationship of the received dataset. The metadata of the trained decoder model can be determined based on the received metadata of the model (e.g., the first to fourth alternatives) or the metadata of the generated dataset (e.g., the fifth alternative). In one implementation, it can obtain the same metadata as it receives from another node.
[0096] It should be noted that this may apply to different backbone structures or different state information S. B (If available) Train multiple decoder models. Therefore, the metadata of each trained decoder model can further contain information about its state S. B (Where applicable) information, and additional parameters about the model. In a specific instance, both the encoder and decoder models can then be trained.
[0097] It should be noted that the node training the decoder model can further transmit information about the trained decoder models (each of which) and the metadata associated with said decoder models to the node training the encoder model. This information can be sent using any of the five alternatives described herein. The node with the encoder model can use this information to update and / or retrain its local decoder model, fix said local decoder model, and then update the encoder model based on the updated / retrained local decoder model. The updated / retrained encoder model can determine its associated metadata based on the metadata it receives from the other side. The process of the second scenario can be repeated several times and can be stopped based on stopping criteria.
[0098] It should be noted that combinations of these schemes can be used. For example, a first alternative can be used for the first option, allowing the UE to have an initial encoder model. Then, if the performance of the resulting two-sided model is unsatisfactory (e.g., due to a poor match between the UE's input distribution and the received encoder), a third alternative for the second option can be used. The UE can then train, retrain, and / or update its encoder based on the received decoder model.
[0099] During the inference phase, it can be assumed that one or more encoder models are trained and stored together with their corresponding metadata in the node A side node, and one or more decoder models are trained and stored together with their corresponding metadata in the node B side node.
[0100] During the inference phase, in one implementation scheme, it can be based on and / or The encoder model is determined by the current values of (e.g., the specific state ID of node A and / or the local state ID of node A, the specific state ID of node B and / or the local state ID of node B), where it is assumed that... Communicate with the node on the A side.
[0101] In another implementation, the choice of encoder model may depend on S A The values of the parameters may depend on the parameters that determine the encoder model contained in the metadata of the stored encoder model. In a further embodiment, the selection of the encoder model may depend on the selection of the decoder model. In this case, the node A side node receives some information about the metadata of the selected decoder model and then selects an encoder model that matches the metadata of the decoder model. The metadata of the selected decoder can provide information about... Information, and S B (Where possible) the relevant parameters, and the parameters used during the training of the decoder model.
[0102] Similarly, one or more different implementations can be used to determine the decoder model. In one implementation, it can be based on... and / or The current value determines the decoder model, where it is assumed Communicates with the node on the B side. In another implementation, the choice of decoder model may depend on S. B The values of the parameters may depend on additional parameters that determine the decoder model, contained in the metadata of the stored decoder model.
[0103] In a further implementation, the selection of the decoder model may depend on the selection of the encoder model. In this case, the node B side node receives some information about the metadata of the selected encoder model and then selects a decoder model that matches the metadata of the encoder model. The metadata of the selected encoder can provide information about... Information, and S A (Where possible) the relevant parameters, and the parameters used during the training of the encoder model.
[0104] In one implementation, a two-sided model is intended for Channel State Information (CSI) feedback compression, where node A and node B correspond to the UE side and NW side, respectively. Using this implementation, data collection can be based on a first set of DL RS (e.g., CSI-RS received at the UE based on configured CSI report settings), CSI resource settings, or a combination thereof. The UE can feed back CSI measurements to the network side via the physical uplink channel based on one or more CSI reports or dataset reports.
[0105] The NW side can collect multiple datasets from multiple UEs, wherein the multiple datasets correspond to shared cell ID, shared TCI status, shared antenna configuration, shared CSI report setting, shared CSI resource setting, or a combination thereof.
[0106] In addition to the aforementioned set of conditions and / or parameters, where applicable, the NW side may also train a first model for CSI prediction or beam management based on the multiple datasets. This first model may be trained based on a first model performed at the NW side.
[0107] The NW side can transmit the model ID or a set of parameters corresponding to the model to the UE side. In one instance, all UEs on the UE side are associated with the same model ID or the same set of parameters, while in another instance, each UE is associated with a different model ID or a different set of parameters.
[0108] Each UE on the UE side can use a second set of DL RS to train a second model, wherein the second model is based on a first model transmitted via the NW side. In one instance, the second model includes an additional set of layers, wherein the additional set of layers is based on the second model training performed at the UE.
[0109] Figure 4 The illustration shows an example of a UE 400 according to various aspects of this disclosure. The UE 400 may include a processor 402, a memory 404, a controller 406, and a transceiver 408. The processor 402, memory 404, controller 406, or transceiver 408, or various combinations thereof, or various components thereof, may be examples of components for performing the various aspects of this disclosure described herein. These components may be coupled via one or more interfaces (e.g., operative ground, communicative ground, functional ground, electronic ground, electrical ground).
[0110] Processor 402, memory 404, controller 406, or transceiver 408, or various combinations or components thereof, may be implemented in hardware (e.g., a circuit system). The hardware may include a processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or other programmable logic means configured to or otherwise support components for performing the functions described in this disclosure, or any combination thereof.
[0111] Processor 402 may include intelligent hardware devices (e.g., a general-purpose processor, DSP, CPU, ASIC, field-programmable gate array (FPGA), or any combination thereof). In some embodiments, processor 402 may be configured to operate memory 404. In some other embodiments, memory 404 may be integrated into processor 402. Processor 402 may be configured to execute computer-readable instructions stored in memory 404 to cause UE 400 to perform various functions of this disclosure.
[0112] Memory 404 may include volatile or non-volatile memory. Memory 404 may store computer-readable, computer-executable code containing instructions that, when executed by processor 402, cause UE 400 to perform the various functions described herein. The code may be stored in a non-transitory computer-readable medium, such as memory 404 or another type of memory. Computer-readable media includes both non-transitory computer storage media and communication media, wherein the communication medium includes any medium that facilitates the transfer of a computer program from one location to another. Non-transitory storage media may be any available medium accessible by a general-purpose or special-purpose computer.
[0113] In some implementations, processor 402 and memory 404 coupled to processor 402 may be configured to cause UE 400 to perform one or more of the functions described herein (e.g., instructions stored in memory 404 are executed by processor 402). For example, according to the examples disclosed herein, processor 402 may support wireless communication at UE 400. For example, processor 402 coupled to memory 404 may be configured to cause UE 400 to determine the first set of information based on an input sample and a first index associated with the input sample or the first set of information, wherein the input sample is determined by or received at a first device, and wherein the first index is associated with at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device. UE 400 may also generate one or more datasets, wherein each of the one or more datasets is a subset of the first set of information. UE 400 may transmit at least one of the one or more datasets.
[0114] Controller 406 manages input and output signals for UE 400. Controller 406 can also manage peripheral devices not integrated into UE 400. In some embodiments, controller 406 may utilize an operating system such as iOS®, Android®, Windows®, or other operating systems. In some embodiments, controller 406 may be implemented as part of processor 402.
[0115] In some embodiments, UE 400 may include at least one transceiver 408. In other embodiments, UE 400 may have more than one transceiver 408. Transceiver 408 may represent a wireless transceiver. Transceiver 408 may include one or more receiver chains 410, one or more transmitter chains 412, or a combination thereof.
[0116] Receiver chain 410 may be configured to receive signals (e.g., control information, data, packets) via a wireless medium. For example, receiver chain 410 may include one or more antennas for receiving signals via air or a wireless medium. Receiver chain 410 may include at least one amplifier (e.g., a low-noise amplifier (LNA)) configured to amplify the received signal. Receiver chain 410 may include at least one demodulator configured to demodulate the received signal and obtain the transmitted data through inverse processing of a modulation technique applied during signal transmission. Receiver chain 410 may include at least one decoder for decoding the demodulated signal to receive the transmitted data.
[0117] Transmitter chain 412 may be configured to generate and transmit signals (e.g., control information, data, packets). Transmitter chain 412 may include at least one modulator for modulating data onto a carrier signal to prepare the signal for transmission over a wireless medium. The at least one modulator may be configured to support one or more techniques, such as amplitude modulation (AM), frequency modulation (FM), or digital modulation schemes, such as phase shift keying (PSK) or quadrature amplitude modulation (QAM). Transmitter chain 412 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitable for transmission over a wireless medium. Transmitter chain 412 may also include one or more antennas for transmitting the amplified signal into the air or wireless medium.
[0118] Figure 5 The illustration depicts an example of a processor 500 according to various aspects of this disclosure. Processor 500 may be an example of a processor configured to perform various operations according to the examples described herein. Processor 500 may include a controller 502 configured to perform various operations according to the examples described herein. Processor 500 may optionally include at least one memory 504, for example, said at least one memory may be an L1 / L2 / L3 cache memory. Additionally or alternatively, processor 500 may optionally include one or more arithmetic logic units (ALUs) 506. One or more of these components may be electronically communicated or otherwise coupled (e.g., operative ground, communicative ground, functional ground, electronic ground, electrical ground) via one or more interfaces (e.g., buses).
[0119] Processor 500 may be a processor chipset and include a protocol stack (e.g., a software stack) executed by the processor chipset to perform various operations (e.g., receiving, obtaining, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, and reading) according to the examples described herein. The processor chipset may include one or more cores, one or more cache memories (e.g., memory native to the processor chipset (e.g., processor 500) or included in the processor chipset), or other memories (e.g., random access memory (RAM), read-only memory (ROM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), static RAM (SRAM), ferroelectric RAM (FeRAM), magnetic RAM (MRAM), resistive RAM (RRAM), flash memory, phase-change memory (PCM), and others).
[0120] Controller 502 can be configured to manage and coordinate various operations of processor 500 (e.g., signaling, receiving, acquiring, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, and reading) to enable processor 500 to support various operations according to the examples described herein. For example, controller 502 can operate as a control unit of processor 500, thereby generating control signals that manage the operation of various components of processor 500. These control signals include enabling or disabling functional units, selecting data paths, initiating memory access, and coordinating the timing of operations.
[0121] Controller 502 may be configured to fetch (e.g., fetch, retrieve, receive) instructions from memory 504 and determine subsequent instructions to be executed to enable processor 500 to support various operations according to the examples described herein. Controller 502 may be configured to track the memory addresses of instructions associated with memory 504. Controller 502 may be configured to decode instructions to determine the operation to be performed and the operands involved. For example, controller 502 may be configured to interpret instructions and determine control signals to be output to other components of processor 500 to enable processor 500 to support various operations according to the examples described herein. Alternatively or additionally, controller 502 may be configured to manage data flow within processor 500. Controller 502 may be configured to control data transfers between registers, arithmetic logic unit (ALU), and other functional units of processor 500.
[0122] Memory 504 may include one or more cache memories (e.g., memory native to processor 500 or included in the processor), or other memories such as RAM, ROM, DRAM, SDRAM, SRAM, MRAM, flash memory, etc. In some embodiments, memory 504 may reside within or on the processor chipset (e.g., native to processor 500). In some other embodiments, memory 504 may reside external to the processor chipset (e.g., remote from processor 500).
[0123] Memory 504 may store computer-readable, computer-executable code containing instructions that, when executed by processor 500, cause processor 500 to perform the various functions described herein. The code may be stored in a non-transitory computer-readable medium, such as system memory or another type of memory. Controller 502 and / or processor 500 may be configured to execute the computer-readable instructions stored in memory 504 to cause processor 500 to perform various functions. For example, processor 500 and / or controller 502 may be coupled to or to memory 504, processor 500, and controller 502, and memory 504 may be configured to perform the various functions described herein. In some instances, processor 500 may include multiple processors and memory 504 may include multiple memories. One or more of the multiple processors may be coupled to one or more of the multiple memories, which may be individually or jointly configured to perform the various functions described herein.
[0124] One or more ALUs 506 may be configured to support various operations according to the examples described herein. In some embodiments, one or more ALUs 506 may reside within or on a processor chipset (e.g., processor 500). In some other embodiments, one or more ALUs 506 may reside outside the processor chipset (e.g., processor 500). One or more ALUs 506 may perform one or more calculations on data, such as addition, subtraction, multiplication, and division. For example, one or more ALUs 506 may receive input operands and operation codes that determine the operation to be performed. One or more ALUs 506 may be configured with various logic and arithmetic circuitry, including adders, subtractors, shifters, and logic gates, to process and manipulate data according to the operation. Alternatively, one or more ALU 506s may support logical operations such as AND, OR, XOR, NOR, and NAND, enabling one or more ALU 506s to handle conditional operations, comparisons, and bitwise operations.
[0125] According to the examples disclosed herein, processor 500 may support wireless communication. Processor 500 may be configured or operable to support components for: determining the first set of information based on an input sample and a first index associated with said input sample or the first set of information, wherein the input sample is determined by or received at a first device, and wherein the first index is associated with at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device; generating one or more datasets, wherein each of the one or more datasets is a subset of the first set of information; and transmitting at least one of the one or more datasets.
[0126] Figure 6 The illustration shows an example of an NE 600 according to various aspects of this disclosure. The NE 600 may include a processor 602, a memory 604, a controller 606, and a transceiver 608. The processor 602, memory 604, controller 606, or transceiver 608, or various combinations thereof, or various components thereof, may be examples of components for performing the various aspects of this disclosure described herein. These components may be coupled via one or more interfaces (e.g., operative ground, communicative ground, functional ground, electronic ground, electrical ground).
[0127] Processor 602, memory 604, controller 606, or transceiver 608, or various combinations or components thereof, may be implemented in hardware (e.g., a circuit system). The hardware may include a processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or other programmable logic means configured to or otherwise support elements for performing the functions described in this disclosure, or any combination thereof.
[0128] Processor 602 may include intelligent hardware devices (e.g., a general-purpose processor, DSP, CPU, ASIC, FPGA, or any combination thereof). In some embodiments, processor 602 may be configured to operate memory 604. In some other embodiments, memory 604 may be integrated into processor 602. Processor 602 may be configured to execute computer-readable instructions stored in memory 604 to cause NE 600 to perform various functions of this disclosure. For example, processor 602 coupled to memory 604 may be configured to cause NE 600 to perform various functions of this disclosure. 600: Receive one or more datasets from at least one of a first device or another device, wherein a first sample of each of the one or more datasets is associated with a first index, and wherein the first index is associated with at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device; develop one or more decoder models, wherein each of the one or more decoder models is based on one or more datasets or a subset of one or more datasets; assign decoder metadata to each of the one or more decoder models, wherein the decoder metadata is based on at least one of a first index, a second index, and one or more model parameters, and wherein the second index represents the first set of one or more parameters of the second device and is associated with a second sample of one or more datasets; and transmit a set of model information associated with the model, wherein the model and associated model metadata are associated with one or more decoder models and decoder metadata, another model or second model and its associated metadata, wherein the other model is developed to behave similarly to one or more of the decoder models, and wherein another metadata of the other model is determined based on the decoder metadata.
[0129] Memory 604 may include volatile or non-volatile memory. Memory 604 may store computer-readable, computer-executable code containing instructions that, when executed by processor 602, cause NE 600 to perform the various functions described herein. The code may be stored in a non-transitory computer-readable medium, such as memory 604 or another type of memory. Computer-readable media includes both non-transitory computer storage media and communication media, wherein the communication medium includes any medium that facilitates the transfer of a computer program from one location to another. Non-transitory storage media may be any available medium accessible by a general-purpose or special-purpose computer.
[0130] In some implementations, processor 602 and memory 604 coupled to processor 602 may be configured to cause NE 600 to perform one or more of the functions described herein (e.g., processor 602 executing instructions stored in memory 604). For example, according to the examples disclosed herein, processor 602 may support wireless communication at NE 600.
[0131] Controller 606 manages input and output signals for NE 600. Controller 606 can also manage peripheral devices not integrated into NE 600. In some embodiments, controller 606 may utilize an operating system such as iOS®, Android®, Windows®, or other operating systems. In some embodiments, controller 606 may be implemented as part of processor 602.
[0132] In some embodiments, NE 600 may include at least one transceiver 608. In other embodiments, NE 600 may have more than one transceiver 608. Transceiver 608 may represent a wireless transceiver. Transceiver 608 may include one or more receiver chains 610, one or more transmitter chains 612, or a combination thereof.
[0133] Receiver chain 610 may be configured to receive signals (e.g., control information, data, packets) via a wireless medium. For example, receiver chain 610 may include one or more antennas for receiving signals via air or a wireless medium. Receiver chain 610 may include at least one amplifier (e.g., a low-noise amplifier (LNA)) configured to amplify the received signal. Receiver chain 610 may include at least one demodulator configured to demodulate the received signal and obtain the transmitted data through inverse processing of a modulation technique applied during signal transmission. Receiver chain 610 may include at least one decoder for decoding the demodulated signal to receive the transmitted data.
[0134] Transmitter chain 612 can be configured to generate and transmit signals (e.g., control information, data, packets). Transmitter chain 612 may include at least one modulator for modulating data onto a carrier signal to prepare the signal for transmission over a wireless medium. The at least one modulator may be configured to support one or more techniques, such as amplitude modulation (AM), frequency modulation (FM), or digital modulation schemes, such as phase shift keying (PSK) or quadrature amplitude modulation (QAM). Transmitter chain 612 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitable for transmission over a wireless medium. Transmitter chain 612 may also include one or more antennas for transmitting the amplified signal into the air or wireless medium.
[0135] Figure 7The diagram illustrates a flowchart of method 700 according to various aspects of this disclosure. Operation of method 700 may be implemented by a first device (e.g., a UE) described herein. In some embodiments, the UE 400 may execute a set of instructions to control functional elements of a processor to perform the described functions.
[0136] At 702, the method may include determining the first set of information based on an input sample and a first index associated with at least one of the input sample or a first set of information, wherein the input sample is determined by or received at a first device, and wherein the first index is based on at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device. Operation of 702 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 702 may be referenced from... Figure 4 The UE execution described.
[0137] At 704, the method may include generating one or more datasets, wherein each of the one or more datasets is a subset of the first set of information. The operation of 704 can be performed according to the examples described herein. In some embodiments, aspects of the operation of 704 may be referenced... Figure 4 The UE execution described.
[0138] At 706, the method may include transmitting at least one of one or more datasets. The operation of 706 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 706 may be referenced... Figure 4 The UE execution described.
[0139] Figure 8 The diagram illustrates a flowchart of another method 800 according to various aspects of this disclosure. Operation of method 800 may be implemented by a second device (e.g., NE) described herein. In some embodiments, the NE 600 may execute a set of instructions to control the functional elements of a processor to perform the described functions.
[0140] At 802, the method may include receiving one or more datasets from at least one of a first device or another device, wherein a first sample of each of the one or more datasets is associated with a first index, and wherein the first index is associated with at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device. Operation of 802 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 802 may be referenced from... Figure 6 The NE execution described.
[0141] At 804, the method may include developing one or more decoder models, wherein each of the one or more decoder models is based on one or more datasets or one or more subsets of one or more datasets. The operation of 804 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 804 may be referenced from... Figure 6 The NE execution described.
[0142] At 806, the method may include assigning decoder metadata to each of one or more decoder models, wherein the decoder metadata is based on at least one of a first index, a second index, and one or more model parameters, and wherein the second index represents a first set of one or more parameters of a second device and is associated with a second sample of one or more datasets. The operation of 806 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 806 may be referenced from... Figure 6 The NE execution described.
[0143] At 808, the method may include transmitting a set of model information associated with a model, wherein the model and associated model metadata are associated with one or more decoder models and decoder metadata, another model or second model and its associated metadata, wherein the other model is developed to behave similarly to one or more of the decoder models, and wherein another metadata of the other model is determined based on the decoder metadata. The operation of 808 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 808 may be referenced from... Figure 6 The NE execution described.
[0144] Figure 9 The diagram illustrates a flowchart of another method 900 according to various aspects of this disclosure. Operation of method 900 may be implemented by a third device (e.g., NE) described herein. In some embodiments, the NE 600 may execute a set of instructions to control the functional elements of a processor to perform the described functions.
[0145] At 902, the method may include developing one or more encoder models, wherein each of the one or more encoder models is based on one or more sets of model information. The operation of 902 can be performed according to the examples described herein. In some embodiments, aspects of the operation of 902 may be referenced... Figure 6 The NE execution described.
[0146] At 904, the method may include assigning encoder metadata to each of one or more encoder models, wherein the encoder metadata is based on at least one of one or more sets of model information, model metadata of one or more sets of model information, and one or more encoder model parameters. The operation of 904 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 904 may be referenced from... Figure 6 The NE execution described.
[0147] At 906, the method may include storing one or more encoder models and metadata for each of the one or more encoder models. The operation of 906 may be performed according to the examples described herein. In some embodiments, aspects of the operation of 906 may be referenced... Figure 6 The NE execution described.
[0148] It should be noted that the methods described herein describe possible implementations, and the operations and steps may be rearranged or otherwise modified, and other implementations are possible.
[0149] The descriptions herein are provided to enable those skilled in the art to make or use this disclosure. Various modifications to this disclosure will be apparent to those skilled in the art, and the general principles defined herein can be applied to other variations without departing from the scope of this disclosure. Therefore, this disclosure is not limited to the examples and designs described herein, but is given the broadest scope consistent with the principles and novel features disclosed herein.
Claims
1. A first device comprising: At least one memory; as well as At least one processor, coupled to the at least one memory and configured to enable the first device to: The first set of information is determined based on an input sample and a first index associated with at least one of the input sample or a first set of information, wherein the input sample is determined by the first device or received at the first device, and wherein the first index is based on at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device. Generate one or more datasets, wherein each of the one or more datasets is a subset of the first set of information; and Transmit at least one of the one or more datasets.
2. The first device of claim 1, wherein the at least one processor is configured to associate each of the one or more datasets with metadata, an identifier, or both.
3. The first device of claim 2, wherein the at least one processor is configured to enable the first device to transmit the identifier or the metadata associated with each of the one or more datasets.
4. The first device of claim 1, wherein the at least one processor is configured to enable the first device to determine a set of input samples.
5. The first device according to claim 4, wherein the set of input samples is based on channel data representation.
6. The first device according to claim 5, wherein the channel data represents one or more reference signals received from the second device.
7. The first device of claim 5, wherein the channel data represents being associated with one or more different transmit Tx-receive Rx pairs communicating via different frequency bands or different time slots.
8. The first device according to claim 1, wherein the first index includes a parameter indicating measurement time, measurement period, or both.
9. The first device of claim 1, wherein the first index is based on an identifier, the identifier being determined based on a mapping table associated with different combinations of the first set of one or more parameters of the first device and the second set of one or more parameters of the second device.
10. The first device of claim 9, wherein the at least one processor is configured to enable the first device to transmit the mapping table.
11. The first device of claim 1, wherein the at least one processor is configured to receive the second set of one or more parameters.
12. The first device of claim 1, wherein the at least one processor is configured to cause the first device to generate status information associated with the second device, the status information comprising one or more of the following: The value of at least one of the one or more parameters in the second group; Time slot; or Duration.
13. The first device of claim 12, wherein the at least one processor is configured to enable the first device to transmit the status information.
14. The first device according to claim 1, wherein the first device is a UE or a node at the UE.
15. The first device according to claim 1, wherein the first device is a base station or a node at the base station.
16. The first device according to claim 1, wherein the second device is a UE or a node at the UE.
17. The first device according to claim 1, wherein the second device is a base station or a node at the base station.
18. A processor for wireless communication, comprising: At least one controller, coupled to at least one memory and configured to enable the processor to: The first set of information is determined based on an input sample and a first index associated with at least one of the input sample or a first set of information, wherein the input sample is determined by the processor or received at the processor, and wherein the first index is based on at least one of a first set of one or more parameters of the processor and a second set of one or more parameters of the second device. Generate one or more datasets, wherein each of the one or more datasets is a subset of the first set of information; and Transmit at least one of the one or more datasets.
19. A second device comprising: At least one memory; as well as At least one processor, coupled to the at least one memory and configured to enable the second device to: Receive one or more datasets from at least one of a first device or another device, wherein each of the one or more datasets or a first sample of each of the one or more datasets is associated with a first index, and wherein the first index is associated with at least one of a first set of one or more parameters of the first device and a second set of one or more parameters of the second device. Develop one or more decoder models, wherein each of the one or more decoder models is based on the one or more datasets or one or more subsets of the one or more datasets; Decoder metadata is assigned to each of the one or more decoder models, wherein the decoder metadata is based on at least one of a first index, a second index, and one or more model parameters, and wherein the second index represents the first set of one or more parameters of the second device and is associated with a second sample of the one or more datasets; and Transmit a set of model information associated with a model, wherein the model and associated model metadata are associated with one or more decoder models and the decoder metadata, another model or a second model and its associated metadata, wherein the other model is developed to behave similarly to one or more of the decoder models, and wherein another metadata of the other model is determined based on the decoder metadata.
20. A third device comprising: At least one memory; as well as At least one processor, coupled to the at least one memory and configured to enable the third device to: Receive one or more sets of model information; Develop one or more encoder models, wherein each encoder model in the one or more encoder models is based on the one or more sets of model information; The encoder metadata is assigned to each of the one or more encoder models, wherein the encoder metadata is based on at least one of the one or more sets of model information, the model metadata of the one or more sets of model information, and one or more encoder model parameters. and Store the metadata of the one or more encoder models and each of the one or more encoder models.
