Method and system for implementing a two-sided channel state information (CSI) prediction model

Abstract

A wireless transmit / receive unit (WTRU) comprising a processor configured to receive configuration information for a channel state information (CSI) prediction model, store previous instance of estimated CSI in a circular queue to be used as inputs to the CSI prediction model, determine a current estimated CSI associated with a channel at a current time based on CSI reference signals (CSI-RSs), and determine a predicted CSI associated with the channel at the current time based on the previous instances of estimated CSI in the circular queue.

Description

METHOD AND SYSTEM FOR IMPLEMENTING A TWO-SIDED CHANNEL STATE INFORMATION (CSI) PREDICTION MODELCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of United States Provisional Application Nos. 63 / 581 ,177 filed on September 7, 2023 and 63 / 589,153 filed on October 10, 2023, the entire contents of each of which are incorporated herein by reference.BACKGROUND

[0002] CSI parameters contain information about the state of a channel that may be extracted from the estimated channel. In conventional 5G radio networks, the CSI feedback parameters may include the Channel Quality Indicator (CQI), the precoding matrix indicator (PMI) with different codebook sets, and the rank indicator (Rl). The WTRU may measure and compute the CSI parameters using the CSI reference signals (CSI-RS) received from the access network node (gNB). The WTRU may then report CSI parameters to the gNB as part of the CSI feedback report. Upon receiving the CSI parameters, the gNB may schedule downlink data transmissions with attributes such as modulation scheme, code rate, number of transmission layers, and MIMO precoding.

[0003] CSI refers to the information about the state of a channel. Traditionally, CSI has included CQI, PMI, and Rl. For AI / ML models, CSI may further include the actual channel matrix, an index to an entry in a codebook of CSI information, a set of eigenvectors of the channel matrix, or a compressed and / or quantized version of those.

[0004] In conventional 5G radio networks, CSI feedback consumes a large portion of the uplink bandwidth, resulting in additional feedback overhead. The more the amount of uplink data used for CSI feedback can be reduced, the more bandwidth will be available for the user data.SUMMARY

[0005] A wireless transmit / receive unit (WTRU) may comprise a processor. The processor may be configured to receive configuration information for a channel state information (CSI) prediction model. The processor may be configured to store previous instance of estimated CSI in a circular queue to be used as inputs to the CSI prediction model. The processor may be configured to determine a current estimated CSI associated with a channel at a current time based on CSI reference signals (CSI-RSs). The processor may be configured to determine apredicted CSI associated with the channel at the current time based on the previous instances of estimated CSI in the circular queue.

[0006] The configuration information may be, for example, a CSI prediction model ID, CSI compression model ID, residual compression model ID, residual threshold, quantization configuration parameters, or beamforming future channel index.

[0007] The processor may be configured to receive a reconfiguration message. The processor may be configured to transition to a startup mode based on the reconfiguration message. The processor may be configured to flush the circular queue to empty at the transition to the startup mode. The reconfiguration message may be, for example, an indication of a communication failure.

[0008] The processor may be configured to provide a null output from the CSI prediction model while the circular queue is empty.

[0009] The processor may be configured to determine a number of previous instances of CSI currently within the circular queue. The processor may be configured to provide an instance of CSI identical to the CSI within the circular queue based on the determination that one previous instance of CSI currently resides within the circular queue.

[0010] The processor may be configured to select the CSI prediction model ID based on one or more parameters. For example, the CSI prediction model ID may be based on one or more WTRU capabilities, a CSI feedback delay, or a detected channel environment.

[0011] In some examples, the one or more WTRU capabilities may include, for example, computation power and / or memory limits. In other examples, the detected channel environment may be, for example, doppler and delay spread.

[0012] In some examples, the CSI prediction model ID may include, for example, an observation window size and / or a prediction window size. The observation window size may include, for example, a number of past instances of CSI information that are configured to be used as input to the CSI prediction model. The prediction window size may include, for example, a number of future instances of CSI information predicted by the CSI prediction model.

[0013] A WTRU may be configured to perform a method that includes one or more of the following steps. The method may include receiving configuration information for a channel state information (CSI) prediction model. The method may include storing previous instance of estimated CSI in a circular queue to be used as inputs to the CSI prediction model. The method may include determining a current estimated CSI associated with a channel at a current time based on CSI reference signals (CSI-RSs). The method may include determining a predictedCSI associated with the channel at the current time based on the previous instances of estimated CSI in the circular queue.

[0014] The configuration information may be, for example, a CSI prediction model ID, CSI compression model ID, residual compression model ID, residual threshold, quantization configuration parameters, or beamforming future channel index.

[0015] The method may include receiving a reconfiguration message. The method may include transitioning to a startup mode based on the reconfiguration message, wherein the circular queue is flushed to empty at the transition to the startup mode. The reconfiguration message may be, for example, an indication of a communication failure.

[0016] The method may include providing a null output from the CSI prediction model while the circular queue is empty.

[0017] The method may include determining the number of previous instances of CSI currently within the circular queue. The method may include providing an instance of CSI identical to the CSI within the circular queue based on the determination that one previous instance of CSI currently resides within the circular queue.

[0018] The method may include selecting the CSI prediction model ID based on one or more parameters. For example, the CSI prediction model ID may be selected based on one or more WTRU capabilities, a CSI feedback delay, or a detected channel environment.

[0019] In some examples, the one or more WTRU capabilities may include, for example, computation power and / or memory limits. In other examples, the detected channel environment may be, for example, doppler and delay spread.

[0020] In some examples, the CSI prediction model ID may include, for example, an observation window size and / or a prediction window size. The observation window size may include, for example, a number of past instances of CSI information that are configured to be used as input to the CSI prediction model. The prediction window size may include, for example, a number of future instances of CSI information predicted by the CSI prediction model.BRIEF DESCRIPTION OF THE DRAWINGS

[0021] FIG. 1A is a system diagram illustrating an example communications system in which one or more disclosed embodiments may be implemented.

[0022] FIG. 1 B is a system diagram illustrating an example wireless transmit / receive unit (WTRU) that may be used within the communications system illustrated in FIG. 1A according to an embodiment.

[0023] FIG. 1C is a system diagram illustrating an example radio access network (RAN) and an example core network (CN) that may be used within the communications system illustrated in FIG. 1A according to an embodiment.

[0024] FIG. 1 D is a system diagram illustrating a further example RAN and a further example CN that may be used within the communications system illustrated in FIG. 1A according to an embodiment.

[0025] FIG. 2 is a system diagram illustrating an example transmission of CSI feedback within a 5G radio network.

[0026] FIG. 3 is a schematic diagram illustrating an example end-to-end pipeline for calculating residual information using CSI.

[0027] FIG. 4 is a schematic diagram illustrating an example end-to-end alternative pipeline for calculating residual information using compressed versions of CSI.

[0028] FIG. 5 is a schematic diagram illustrating an example sequential deep neural network (DNN) based CSI prediction model.

[0029] FIG. 6 is a schematic diagram illustrating an example preparation of residual CSI feedback at the WTRU.

[0030] FIG. 7 is a schematic diagram illustrating an example end-to-end pipeline for compression, quantization, and entropy coding.

[0031] FIG. 8 is a schematic diagram illustrating an example transformer-based models utilized for encoder and decoder blocks.

[0032] FIG. 9 is a flow chart illustrating an example WTRU method for calculating and transmitting residual information using CSI.

[0033] FIG. 10 is a flow chart illustrating an example gNB method for calculating and transmitting residual information using CSI.

[0034] FIG. 11 is a schematic diagram of an example training architecture for an AI / ML CSI prediction solution.

[0035] FIG. 12 is a schematic diagram of example structures for AI / ML CSI prediction models.DETAILED DESCRIPTION

[0036] FIG. 1A is a diagram illustrating an example communications system 100 in which one or more disclosed embodiments may be implemented. The communications system 100 may be a multiple access system that provides content, such as voice, data, video, messaging, broadcast, etc., to multiple wireless users. The communications system 100 may enablemultiple wireless users to access such content through the sharing of system resources, including wireless bandwidth. For example, the communications systems 100 may employ one or more channel access methods, such as code division multiple access (CDMA), time division multiple access (TDMA), frequency division multiple access (FDMA), orthogonal FDMA (OFDMA), single-carrier FDMA (SC-FDMA), zero-tail unique-word DFT-Spread OFDM (ZT UW DTS-s OFDM), unique word OFDM (UW-OFDM), resource block-filtered OFDM, filter bank multicarrier (FBMC), and the like.

[0037] As shown in FIG. 1A, the communications system 100 may include wireless transmit / receive units (WTRUs) 102a, 102b, 102c, 102d, a RAN 104 / 113, a CN 106 / 115, a public switched telephone network (PSTN) 108, the Internet 110, and other networks 112, though it will be appreciated that the disclosed embodiments contemplate any number of WTRUs, base stations, networks, and / or network elements. Each of the WTRUs 102a, 102b, 102c, 102d may be any type of device configured to operate and / or communicate in a wireless environment. By way of example, the WTRUs 102a, 102b, 102c, 102d, any of which may be referred to as a “station” and / or a “STA,” may be configured to transmit and / or receive wireless signals and may include a user equipment (UE), a mobile station, a fixed or mobile subscriber unit, a subscription-based unit, a pager, a cellular telephone, a personal digital assistant (PDA), a smartphone, a laptop, a netbook, a personal computer, a wireless sensor, a hotspot or Mi-Fi device, an Internet of Things (loT) device, a watch or other wearable, a head-mounted display (HMD), a vehicle, a drone, a medical device and applications (e.g., remote surgery), an industrial device and applications (e.g., a robot and / or other wireless devices operating in an industrial and / or an automated processing chain contexts), a consumer electronics device, a device operating on commercial and / or industrial wireless networks, and the like. Any of the WTRUs 102a, 102b, 102c, 102d may be interchangeably referred to as a WTRU. Further, any description herein that is described with reference to a UE may be equally applicable to a WTRU (or vice versa). For example, a WTRU may be configured to perform any of the processes or procedures described herein as being performed by a UE (or vice versa).

[0038] The communications systems 100 may also include a base station 114a and / or a base station 114b. Each of the base stations 114a, 114b may be any type of device configured to wirelessly interface with at least one of the WTRUs 102a, 102b, 102c, 102d to facilitate access to one or more communication networks, such as the CN 106 / 115, the Internet 110, and / or the other networks 112. By way of example, the base stations 114a, 114b may be a base transceiver station (BTS), a Node-B, an eNode B, a Home Node B, a Home eNode B, a gNB, a NR NodeB, a site controller, an access point (AP), a wireless router, and the like. While thebase stations 114a, 114b are each depicted as a single element, it will be appreciated that the base stations 114a, 114b may include any number of interconnected base stations and / or network elements.

[0039] The base station 114a may be part of the RAN 104 / 113, which may also include other base stations and / or network elements (not shown), such as a base station controller (BSC), a radio network controller (RNC), relay nodes, etc. The base station 114a and / or the base station 114b may be configured to transmit and / or receive wireless signals on one or more carrier frequencies, which may be referred to as a cell (not shown). These frequencies may be in licensed spectrum, unlicensed spectrum, or a combination of licensed and unlicensed spectrum. A cell may provide coverage for a wireless service to a specific geographical area that may be relatively fixed or change over time. The cell may further be divided into cell sectors. For example, the cell associated with the base station 114a may be divided into three sectors. Thus, in one embodiment, the base station 114a may include three transceivers, i.e., one for each sector of the cell. In an embodiment, the base station 114a may employ multiple-input multiple output (MIMO) technology and may utilize multiple transceivers for each sector of the cell. For example, beamforming may be used to transmit and / or receive signals in desired spatial directions.

[0040] The base stations 114a, 114b may communicate with one or more of the WTRUs 102a, 102b, 102c, 102d over an air interface 116, which may be any suitable wireless communication link (e.g., radio frequency (RF), microwave, centimeter wave, micrometer wave, infrared (IR), ultraviolet (UV), visible light, etc.). The air interface 116 may be established using any suitable radio access technology (RAT).

[0041] More specifically, as noted above, the communications system 100 may be a multiple access system and may employ one or more channel access schemes, such as CDMA, TDMA, FDMA, OFDMA, SC-FDMA, and the like. For example, the base station 114a in the RAN 104 / 113 and the WTRUs 102a, 102b, 102c may implement a radio technology such as Universal Mobile Telecommunications System (UMTS) Terrestrial Radio Access (UTRA), which may establish the air interface 115 / 116 / 117 using wideband CDMA (WCDMA). WCDMA may include communication protocols such as High-Speed Packet Access (HSPA) and / or Evolved HSPA (HSPA+). HSPA may include High-Speed Downlink (DL) Packet Access (HSDPA) and / or High-Speed UL Packet Access (HSUPA).

[0042] In an embodiment, the base station 114a and the WTRUs 102a, 102b, 102c may implement a radio technology such as Evolved UMTS Terrestrial Radio Access (E-UTRA),which may establish the air interface 116 using Long Term Evolution (LTE) and / or LTE- Advanced (LTE-A) and / or LTE-Advanced Pro (LTE-A Pro).

[0043] In an embodiment, the base station 114a and the WTRUs 102a, 102b, 102c may implement a radio technology such as NR Radio Access, which may establish the air interface 116 using New Radio (NR).

[0044] In an embodiment, the base station 114a and the WTRUs 102a, 102b, 102c may implement multiple radio access technologies. For example, the base station 114a and the WTRUs 102a, 102b, 102c may implement LTE radio access and NR radio access together, for instance using dual connectivity (DC) principles. Thus, the air interface utilized by WTRUs 102a, 102b, 102c may be characterized by multiple types of radio access technologies and / or transmissions sent to / from multiple types of base stations (e.g., a eNB and a gNB).

[0045] In other embodiments, the base station 114a and the WTRUs 102a, 102b, 102c may implement radio technologies such as IEEE 802.11 (i.e. , Wreless Fidelity (WiFi), IEEE 802.16 (i.e. , Worldwide Interoperability for Microwave Access (WiMAX)), CDMA2000, CDMA2000 1X, CDMA2000 EV-DO, Interim Standard 2000 (IS-2000), Interim Standard 95 (IS-95), Interim Standard 856 (IS-856), Global System for Mobile communications (GSM), Enhanced Data rates for GSM Evolution (EDGE), GSM EDGE (GERAN), and the like.

[0046] The base station 114b in FIG. 1 A may be a wireless router, Home Node B, Home eNode B, or access point, for example, and may utilize any suitable RAT for facilitating wireless connectivity in a localized area, such as a place of business, a home, a vehicle, a campus, an industrial facility, an air corridor (e.g., for use by drones), a roadway, and the like. In one embodiment, the base station 114b and the WTRUs 102c, 102d may implement a radio technology such as IEEE 802.11 to establish a wireless local area network (WLAN). In an embodiment, the base station 114b and the WTRUs 102c, 102d may implement a radio technology such as IEEE 802.15 to establish a wireless personal area network (WPAN). In yet another embodiment, the base station 114b and the WTRUs 102c, 102d may utilize a cellularbased RAT (e.g., WCDMA, CDMA2000, GSM, LTE, LTE-A, LTE-A Pro, NR, etc.) to establish a picocell or femtocell. As shown in FIG. 1A, the base station 114b may have a direct connection to the Internet 110. Thus, the base station 114b may not be required to access the Internet 110 via the CN 106 / 115.

[0047] The RAN 104 / 113 may be in communication with the CN 106 / 115, which may be any type of network configured to provide voice, data, applications, and / or voice over internet protocol (VoIP) services to one or more of the WTRUs 102a, 102b, 102c, 102d. The data may have varying quality of service (QoS) requirements, such as differing throughput requirements,latency requirements, error tolerance requirements, reliability requirements, data throughput requirements, mobility requirements, and the like. The CN 106 / 115 may provide call control, billing services, mobile location-based services, pre-paid calling, Internet connectivity, video distribution, etc., and / or perform high-level security functions, such as user authentication. Although not shown in FIG. 1A, it will be appreciated that the RAN 104 / 113 and / or the CN 106 / 115 may be in direct or indirect communication with other RANs that employ the same RAT as the RAN 104 / 113 or a different RAT. For example, in addition to being connected to the RAN 104 / 113, which may be utilizing NR radio technology, the CN 106 / 115 may also be in communication with another RAN (not shown) employing a GSM, UMTS, CDMA 2000, WiMAX, E-UTRA, or WiFi radio technology.

[0048] The CN 106 / 115 may also serve as a gateway for the WTR Us 102a, 102b, 102c, 102d to access the PSTN 108, the Internet 110, and / or the other networks 112. The PSTN 108 may include circuit-switched telephone networks that provide plain old telephone service (POTS). The Internet 110 may include a global system of interconnected computer networks and devices that use common communication protocols, such as the transmission control protocol (TCP), user datagram protocol (UDP), and / or the internet protocol (IP) in the TCP / IP internet protocol suite. The networks 112 may include wired and / or wireless communications networks owned and / or operated by other service providers. For example, the networks 112 may include another CN connected to one or more RANs, which may employ the same RAT as the RAN 104 / 113 or a different RAT.

[0049] Some or all of the WTRUs 102a, 102b, 102c, 102d in the communications system 100 may include multi-mode capabilities (e.g., the WTRUs 102a, 102b, 102c, 102d may include multiple transceivers for communicating with different wireless networks over different wireless links). For example, the WTRU 102c shown in FIG. 1A may be configured to communicate with the base station 114a, which may employ a cellular-based radio technology, and with the base station 114b, which may employ an IEEE 802 radio technology.

[0050] FIG. 1 B is a system diagram illustrating an example WTRU 102. As shown in FIG. 1 B, the WTRU 102 may include a processor 118, a transceiver 120, a transmit / receive element 122, a speaker / microphone 124, a keypad 126, a display / touchpad 128, non-removable memory 130, removable memory 132, a power source 134, a global positioning system (GPS) chipset 136, and / or other peripherals 138, among others. It will be appreciated that the WTRU 102 may include any sub-combination of the foregoing elements while remaining consistent with an embodiment.

[0051] The processor 118 may be a general purpose processor, a special purpose processor, a conventional processor, a digital signal processor (DSP), a plurality of microprocessors, one or more microprocessors in association with a DSP core, a controller, a microcontroller, Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) circuits, any other type of integrated circuit (IC), a state machine, and the like. The processor 118 may perform signal coding, data processing, power control, input / output processing, and / or any other functionality that enables the WTRU 102 to operate in a wireless environment. The processor 118 may be coupled to the transceiver 120, which may be coupled to the transmit / receive element 122. While FIG. 1 B depicts the processor 118 and the transceiver 120 as separate components, it will be appreciated that the processor 118 and the transceiver 120 may be integrated together in an electronic package or chip.

[0052] The transmit / receive element 122 may be configured to transmit signals to, or receive signals from, a base station (e.g., the base station 114a) over the air interface 116. For example, in one embodiment, the transmit / receive element 122 may be an antenna configured to transmit and / or receive RF signals. In an embodiment, the transmit / receive element 122 may be an emitter / detector configured to transmit and / or receive IR, UV, or visible light signals, for example. In yet another embodiment, the transmit / receive element 122 may be configured to transmit and / or receive both RF and light signals. It will be appreciated that the transmit / receive element 122 may be configured to transmit and / or receive any combination of wireless signals.

[0053] Although the transmit / receive element 122 is depicted in FIG. 1 B as a single element, the WTRU 102 may include any number of transmit / receive elements 122. More specifically, the WTRU 102 may employ MIMO technology. Thus, in one embodiment, the WTRU 102 may include two or more transmit / receive elements 122 (e.g., multiple antennas) for transmitting and receiving wireless signals over the air interface 116.

[0054] The transceiver 120 may be configured to modulate the signals that are to be transmitted by the transmit / receive element 122 and to demodulate the signals that are received by the transmit / receive element 122. As noted above, the WTRU 102 may have multi-mode capabilities. Thus, the transceiver 120 may include multiple transceivers for enabling the WTRU 102 to communicate via multiple RATs, such as NR and IEEE 802.11 , for example.

[0055] The processor 118 of the WTRU 102 may be coupled to, and may receive user input data from, the speaker / microphone 124, the keypad 126, and / or the display / touchpad 128 (e.g., a liquid crystal display (LCD) display unit or organic light-emitting diode (OLED) display unit). The processor 118 may also output user data to the speaker / microphone 124, the keypad 126, and / or the display / touchpad 128. In addition, the processor 118 may access information from,and store data in, any type of suitable memory, such as the non-removable memory 130 and / or the removable memory 132. The non-removable memory 130 may include random-access memory (RAM), read-only memory (ROM), a hard disk, or any other type of memory storage device. The removable memory 132 may include a subscriber identity module (SIM) card, a memory stick, a secure digital (SD) memory card, and the like. In other embodiments, the processor 118 may access information from, and store data in, memory that is not physically located on the WTRU 102, such as on a server or a home computer (not shown).

[0056] The processor 118 may receive power from the power source 134 and may be configured to distribute and / or control the power to the other components in the WTRU 102. The power source 134 may be any suitable device for powering the WTRU 102. For example, the power source 134 may include one or more dry cell batteries (e.g., nickel-cadmium (NiCd), nickel-zinc (NiZn), nickel metal hydride (NiMH), lithium-ion (Li-ion), etc.), solar cells, fuel cells, and the like.

[0057] The processor 118 may also be coupled to the GPS chipset 136, which may be configured to provide location information (e.g., longitude and latitude) regarding the current location of the WTRU 102. In addition to, or in lieu of, the information from the GPS chipset 136, the WTRU 102 may receive location information over the air interface 116 from a base station (e.g., base stations 114a, 114b) and / or determine its location based on the timing of the signals being received from two or more nearby base stations. It will be appreciated that the WTRU 102 may acquire location information by way of any suitable location-determination method while remaining consistent with an embodiment.

[0058] The processor 118 may further be coupled to other peripherals 138, which may include one or more software and / or hardware modules that provide additional features, functionality, and / or wired or wireless connectivity. For example, the peripherals 138 may include an accelerometer, an e-compass, a satellite transceiver, a digital camera (for photographs and / or video), a universal serial bus (USB) port, a vibration device, a television transceiver, a hands free headset, a Bluetooth® module, a frequency modulated (FM) radio unit, a digital music player, a media player, a video game player module, an Internet browser, a Virtual Reality and / or Augmented Reality (VR / AR) device, an activity tracker, and the like. The peripherals 138 may include one or more sensors, the sensors may be one or more of a gyroscope, an accelerometer, a hall effect sensor, a magnetometer, an orientation sensor, a proximity sensor, a temperature sensor, a time sensor, a geolocation sensor, an altimeter, a light sensor, a touch sensor, a magnetometer, a barometer, a gesture sensor, a biometric sensor, and / or a humidity sensor.

[0059] The WTRU 102 may include a full duplex radio for which transmission and reception of some or all of the signals (e.g., associated with particular subframes for both the UL (e.g., for transmission) and downlink (e.g., for reception) may be concurrent and / or simultaneous. The full duplex radio may include an interference management unit 139 to reduce and or substantially eliminate self-interference via either hardware (e.g., a choke) or signal processing via a processor (e.g., a separate processor (not shown) or via processor 118). In an embodiment, the WRTU 102 may include a half-duplex radio for which transmission and reception of some or all of the signals (e.g., associated with particular subframes for either the UL (e.g., for transmission) or the downlink (e.g., for reception)).

[0060] FIG. 1C is a system diagram illustrating the RAN 104 and the CN 106 according to an embodiment. As noted above, the RAN 104 may employ an E-UTRA radio technology to communicate with the WTRUs 102a, 102b, 102c over the air interface 116. The RAN 104 may also be in communication with the CN 106.

[0061] The RAN 104 may include eNode-Bs 160a, 160b, 160c, though it will be appreciated that the RAN 104 may include any number of eNode-Bs while remaining consistent with an embodiment. The eNode-Bs 160a, 160b, 160c may each include one or more transceivers for communicating with the WTRUs 102a, 102b, 102c over the air interface 116. In one embodiment, the eNode-Bs 160a, 160b, 160c may implement MIMO technology. Thus, the eNode-B 160a, for example, may use multiple antennas to transmit wireless signals to, and / or receive wireless signals from, the WTRU 102a.

[0062] Each of the eNode-Bs 160a, 160b, 160c may be associated with a particular cell (not shown) and may be configured to handle radio resource management decisions, handover decisions, scheduling of users in the UL and / or DL, and the like. As shown in FIG. 1C, the eNode-Bs 160a, 160b, 160c may communicate with one another over an X2 interface.

[0063] The CN 106 shown in FIG. 1C may include a mobility management entity (MME) 162, a serving gateway (SGW) 164, and a packet data network (PDN) gateway (or PGW) 166. While each of the foregoing elements is depicted as part of the CN 106, it will be appreciated that any of these elements may be owned and / or operated by an entity other than the CN operator.

[0064] The MME 162 may be connected to each of the eNode-Bs 162a, 162b, 162c in the RAN 104 via an S1 interface and may serve as a control node. For example, the MME 162 may be responsible for authenticating users of the WTRUs 102a, 102b, 102c, bearer activation / deactivation, selecting a particular serving gateway during an initial attachment of the WTRUs 102a, 102b, 102c, and the like. The MME 162 may provide a control plane function forswitching between the RAN 104 and other RANs (not shown) that employ other radio technologies, such as GSM and / or WCDMA.

[0065] The SGW 164 may be connected to each of the eNode Bs 160a, 160b, 160c in the RAN 104 via the S1 interface. The SGW 164 may generally route and forward user data packets to / from the WTRUs 102a, 102b, 102c. The SGW 164 may perform other functions, such as anchoring user planes during inter-eNode B handovers, triggering paging when DL data is available for the WTRUs 102a, 102b, 102c, managing and storing contexts of the WTRUs 102a, 102b, 102c, and the like.

[0066] The SGW 164 may be connected to the PGW 166, which may provide the WTRUs 102a, 102b, 102c with access to packet-switched networks, such as the Internet 110, to facilitate communications between the WTRUs 102a, 102b, 102c and IP-enabled devices.

[0067] The CN 106 may facilitate communications with other networks. For example, the CN 106 may provide the WTRUs 102a, 102b, 102c with access to circuit- switched networks, such as the PSTN 108, to facilitate communications between the WTRUs 102a, 102b, 102c and traditional land-line communications devices. For example, the CN 106 may include, or may communicate with, an IP gateway (e.g., an IP multimedia subsystem (IMS) server) that serves as an interface between the CN 106 and the PSTN 108. In addition, the CN 106 may provide the WTRUs 102a, 102b, 102c with access to the other networks 112, which may include other wired and / or wireless networks that are owned and / or operated by other service providers.

[0068] Although the WTRU is described in FIGS. 1A-1D as a wireless terminal, it is contemplated that in certain representative embodiments such a terminal may use (e.g., temporarily or permanently) wired communication interfaces with the communication network.

[0069] In representative embodiments, the other network 112 may be a WLAN.

[0070] A WLAN in Infrastructure Basic Service Set (BSS) mode may have an Access Point (AP) for the BSS and one or more stations (STAs) associated with the AP. The AP may have access or an interface to a Distribution System (DS) or another type of wired / wireless network that carries traffic in to and / or out of the BSS. Traffic to STAs that originates from outside the BSS may arrive through the AP and may be delivered to the STAs. Traffic originating from STAs to destinations outside the BSS may be sent to the AP to be delivered to respective destinations. Traffic between STAs within the BSS may be sent through the AP, for example, where the source STA may send traffic to the AP, and the AP may deliver the traffic to the destination STA. The traffic between STAs within a BSS may be considered and / or referred to as peer-to- peer traffic. The peer-to-peer traffic may be sent between (e.g., directly between) the source and destination STAs with a direct link setup (DLS). In certain representative embodiments, theDLS may use an 802.11e DLS or an 802.11z tunneled DLS (TDLS). A WLAN using an Independent BSS (IBSS) mode may not have an AP, and the STAs (e.g., all of the STAs) within or using the IBSS may communicate directly with each other. The IBSS mode of communication may sometimes be referred to herein as an “ad-hoc” mode of communication.

[0071] When using the 802.11 ac infrastructure mode of operation or a similar mode of operation, the AP may transmit a beacon on a fixed channel, such as a primary channel. The primary channel may be a fixed width (e.g., 20 MHz wide bandwidth) or a dynamically set width via signaling. The primary channel may be the operating channel of the BSS and may be used by the STAs to establish a connection with the AP. In certain representative embodiments, Carrier Sense Multiple Access with Collision Avoidance (CSMA / CA) may be implemented, for example, in 802.11 systems. For CSMA / CA, the STAs (e.g., every STA), including the AP, may sense the primary channel. If the primary channel is sensed / detected and / or determined to be busy by a particular STA, the particular STA may back off. One STA (e.g., only one station) may transmit at any given time in a given BSS.

[0072] High Throughput (HT) STAs may use a 40 MHz wide channel for communication, for example, via a combination of the primary 20 MHz channel with an adjacent or nonadjacent 20 MHz channel to form a 40 MHz wide channel.

[0073] Very High Throughput (VHT) STAs may support 20MHz, 40 MHz, 80 MHz, and / or 160 MHz wide channels. The 40 MHz, and / or 80 MHz, channels may be formed by combining contiguous 20 MHz channels. A 160 MHz channel may be formed by combining 8 contiguous 20 MHz channels or by combining two non-contiguous 80 MHz channels, which may be referred to as an 80+80 configuration. For the 80+80 configuration, the data, after channel encoding, may be passed through a segment parser that may divide the data into two streams. Inverse Fast Fourier Transform (IFFT) processing and time domain processing may be done on each stream separately. The streams may be mapped on to the two 80 MHz channels, and the data may be transmitted by a transmitting STA. At the receiver of the receiving STA, the above described operation for the 80+80 configuration may be reversed, and the combined data may be sent to the Medium Access Control (MAC).

[0074] Sub 1 GHz modes of operation are supported by 802.11 af and 802.11 ah. The channel operating bandwidths and carriers are reduced in 802.11af and 802.11 ah relative to those used in 802.11n, and 802.11ac. 802.11af supports 5 MHz, 10 MHz and 20 MHz bandwidths in the TV White Space (TVWS) spectrum, and 802.11ah supports 1 MHz, 2 MHz, 4 MHz, 8 MHz, and 16 MHz bandwidths using non-TVWS spectrum. According to a representative embodiment, 802.11ah may support Meter Type Control / Machine-Type Communications, such as MTCdevices in a macro coverage area. MTC devices may have certain capabilities, for example, limited capabilities, including support for (e.g., only support for) certain and / or limited bandwidths. The MTC devices may include a battery with a battery life above a threshold (e.g., to maintain a very long battery life).

[0075] WLAN systems, which may support multiple channels, and channel bandwidths, such as 802.11 n, 802.11ac, 802.11af, and 802.11 ah, include a channel which may be designated as the primary channel. The primary channel may have a bandwidth equal to the largest common operating bandwidth supported by all STAs in the BSS. The bandwidth of the primary channel may be set and / or limited by a STA, from among all STAs in operating in a BSS, which supports the smallest bandwidth operating mode. In the example of 802.11ah, the primary channel may be 1 MHz wide for STAs (e.g., MTC type devices) that support (e.g., only support) a 1 MHz mode, even if the AP, and other STAs in the BSS support 2 MHz, 4 MHz, 8 MHz, 16 MHz, and / or other channel bandwidth operating modes. Carrier sensing and / or Network Allocation Vector (NAV) settings may depend on the status of the primary channel. If the primary channel is busy, for example, due to a STA (which supports only a 1 MHz operating mode), transmitting to the AP, the entire available frequency bands may be considered busy even though a majority of the frequency bands remain idle and may be available.

[0076] In the United States, the available frequency bands, which may be used by 802.11 ah, are from 902 MHz to 928 MHz. In Korea, the available frequency bands are from 917.5 MHz to 923.5 MHz. In Japan, the available frequency bands are from 916.5 MHz to 927.5 MHz. The total bandwidth available for 802.11 ah is 6 MHz to 26 MHz depending on the country code.

[0077] FIG. 1 D is a system diagram illustrating the RAN 113 and the CN 115 according to an embodiment. As noted above, the RAN 113 may employ an NR radio technology to communicate with the WTRUs 102a, 102b, 102c over the air interface 116. The RAN 113 may also be in communication with the CN 115.

[0078] The RAN 113 may include gNBs 180a, 180b, 180c, though it will be appreciated that the RAN 113 may include any number of gNBs while remaining consistent with an embodiment.The gNBs 180a, 180b, 180c may each include one or more transceivers for communicating with the WTRUs 102a, 102b, 102c over the air interface 116. In one embodiment, the gNBs 180a, 180b, 180c may implement MIMO technology. For example, gNBs 180a, 108b may utilize beamforming to transmit signals to and / or receive signals from the gNBs 180a, 180b, 180c. Thus, the gNB 180a, for example, may use multiple antennas to transmit wireless signals to and / or receive wireless signals from the WTRU 102a. In an embodiment, the gNBs 180a, 180b, 180c may implement carrier aggregation technology. For example, the gNB 180a may transmitmultiple component carriers to the WTRU 102a (not shown). A subset of these component carriers may be on unlicensed spectrum while the remaining component carriers may be on licensed spectrum. In an embodiment, the gNBs 180a, 180b, 180c may implement Coordinated Multi-Point (CoMP) technology. For example, WTRU 102a may receive coordinated transmissions from gNB 180a and gNB 180b (and / or gNB 180c).

[0079] The WTRUs 102a, 102b, 102c may communicate with gNBs 180a, 180b, 180c using transmissions associated with a scalable numerology. For example, the OFDM symbol spacing and / or OFDM subcarrier spacing may vary for different transmissions, different cells, and / or different portions of the wireless transmission spectrum. The WTRUs 102a, 102b, 102c may communicate with gNBs 180a, 180b, 180c using subframe or transmission time intervals (TTIs) of various or scalable lengths (e.g., containing a varying number of OFDM symbols and / or lasting varying lengths of absolute time).

[0080] The gNBs 180a, 180b, 180c may be configured to communicate with the WTRUs 102a, 102b, 102c in a standalone configuration and / or a non-standalone configuration. In the standalone configuration, WTRUs 102a, 102b, 102c may communicate with gNBs 180a, 180b, 180c without also accessing other RANs (e.g., such as eNode-Bs 160a, 160b, 160c). In the standalone configuration, WTRUs 102a, 102b, 102c may utilize one or more of gNBs 180a, 180b, 180c as a mobility anchor point. In the standalone configuration, WTRUs 102a, 102b, 102c may communicate with gNBs 180a, 180b, 180c using signals in an unlicensed band. In a non-standalone configuration WTRUs 102a, 102b, 102c may communicate with / connect to gNBs 180a, 180b, 180c while also communicating with / connecting to another RAN such as eNode-Bs 160a, 160b, 160c. For example, WTRUs 102a, 102b, 102c may implement DC principles to communicate with one or more gNBs 180a, 180b, 180c and one or more eNode-Bs 160a, 160b, 160c substantially simultaneously. In the non-standalone configuration, eNode-Bs 160a, 160b, 160c may serve as a mobility anchor for WTRUs 102a, 102b, 102c, and gNBs 180a, 180b, 180c may provide additional coverage and / or throughput for servicing WTRUs 102a, 102b, 102c.

[0081] Each of the gNBs 180a, 180b, 180c may be associated with a particular cell (not shown) and may be configured to handle radio resource management decisions, handover decisions, scheduling of users in the UL and / or DL, support of network slicing, dual connectivity, interworking between NR and E-UTRA, routing of user plane data towards User Plane Function (UPF) 184a, 184b, routing of control plane information towards Access and Mobility Management Function (AMF) 182a, 182b and the like. As shown in FIG. 1 D, the gNBs 180a, 180b, 180c may communicate with one another over an Xn interface.

[0082] The CN 115 shown in FIG. 1D may include at least one AMF 182a, 182b, at least one UPF 184a, 184b, at least one Session Management Function (SMF) 183a, 183b, and possibly a Data Network (DN) 185a, 185b. While each of the foregoing elements is depicted as part of the CN 115, it will be appreciated that any of these elements may be owned and / or operated by an entity other than the CN operator.

[0083] The AMF 182a, 182b may be connected to one or more of the gNBs 180a, 180b, 180c in the RAN 113 via an N2 interface and may serve as a control node. For example, the AMF 182a, 182b may be responsible for authenticating users of the WTRUs 102a, 102b, 102c, support for network slicing (e.g., handling of different PDU sessions with different requirements), selecting a particular SMF 183a, 183b, management of the registration area, termination of NAS signaling, mobility management, and the like. Network slicing may be used by the AMF 182a, 182b in order to customize CN support for WTRUs 102a, 102b, 102c based on the types of services being utilized WTRUs 102a, 102b, 102c. For example, different network slices may be established for different use cases, such as services relying on ultra-reliable low latency (URLLC) access, services relying on enhanced massive mobile broadband (eMBB) access, services for machine type communication (MTC) access, and / or the like. The AMF 162 may provide a control plane function for switching between the RAN 113 and other RANs (not shown) that employ other radio technologies, such as LTE, LTE-A, LTE-A Pro, and / or non- 3GPP access technologies such as WiFi.

[0084] The SMF 183a, 183b may be connected to an AMF 182a, 182b in the CN 115 via an N11 interface. The SMF 183a, 183b may also be connected to a UPF 184a, 184b in the CN 115 via an N4 interface. The SMF 183a, 183b may select and control the UPF 184a, 184b and configure the routing of traffic through the UPF 184a, 184b. The SMF 183a, 183b may perform other functions, such as managing and allocating WTRU IP addresses, managing PDU sessions, controlling policy enforcement and QoS, providing downlink data notifications, and the like. A PDU session type may be IP-based, non-IP based, Ethernet-based, and the like.

[0085] The UPF 184a, 184b may be connected to one or more of the gNBs 180a, 180b, 180c in the RAN 113 via an N3 interface, which may provide the WTRUs 102a, 102b, 102c with access to packet-switched networks, such as the Internet 110, to facilitate communications between the WTRUs 102a, 102b, 102c and IP-enabled devices. The UPF 184, 184b may perform other functions, such as routing and forwarding packets, enforcing user plane policies, supporting multi-homed PDU sessions, handling user plane QoS, buffering downlink packets, providing mobility anchoring, and the like.

[0086] The CN 115 may facilitate communications with other networks. For example, the CN 115 may include, or may communicate with, an IP gateway (e.g., an IP multimedia subsystem (IMS) server) that serves as an interface between the CN 115 and the PSTN 108. In addition, the CN 115 may provide the WTRUs 102a, 102b, 102c with access to the other networks 112, which may include other wired and / or wireless networks that are owned and / or operated by other service providers. In one embodiment, the WTRUs 102a, 102b, 102c may be connected to a local Data Network (DN) 185a, 185b through the UPF 184a, 184b via the N3 interface to the UPF 184a, 184b and an N6 interface between the UPF 184a, 184b and the DN 185a, 185b.

[0087] In view of Figures 1A-1 D, and the corresponding description of Figures 1A-1D, one or more, or all, of the functions described herein with regard to one or more of: WTRU 102a-d, Base Station 114a-b, eNode-B 160a-c, MME 162, SGW 164, PGW 166, gNB 180a-c, AMF 182a-ab, UPF 184a-b, SMF 183a-b, DN 185a-b, and / or any other device(s) described herein, may be performed by one or more emulation devices (not shown). The emulation devices may be one or more devices configured to emulate one or more, or all, of the functions described herein. For example, the emulation devices may be used to test other devices and / or to simulate network and / or WTRU functions.

[0088] The emulation devices may be designed to implement one or more tests of other devices in a lab environment and / or in an operator network environment. For example, the one or more emulation devices may perform the one or more, or all, functions while being fully or partially implemented and / or deployed as part of a wired and / or wireless communication network in order to test other devices within the communication network. The one or more emulation devices may perform the one or more, or all, functions while being temporarily implemented / deployed as part of a wired and / or wireless communication network. The emulation device may be directly coupled to another device for purposes of testing and / or may perform testing using over-the-air wireless communications.

[0089] The one or more emulation devices may perform the one or more, including all, functions while not being implemented / deployed as part of a wired and / or wireless communication network. For example, the emulation devices may be utilized in a testing scenario in a testing laboratory and / or a non-deployed (e.g., testing) wired and / or wireless communication network in order to implement testing of one or more components. The one or more emulation devices may be test equipment. Direct RF coupling and / or wireless communications via RF circuitry (e.g., which may include one or more antennas) may be used by the emulation devices to transmit and / or receive data.

[0090] FIG. 2 is a system diagram illustrating an example transmission of CSI feedback within a 5G radio network. For example, transmission of CSI feedback within a 5G radio network 200 may include a gNB 202, a WTRU 204, Channel State Information - Reference Signals (CSI-RS) 206, CSI-Feedback 208, and Downlink Data 210. The WTRU 204 may receive CSI-RS 206 and may transmit CSI-Feedback 208 to the gNB 202. The WTRU may receive downlink data 210 using certain channel information based on the CSI-Feedback 208.

[0091] CSI may refer to the information about the state of a channel. The information may include CQI, PMI, and Rl for traditional models. For AI / ML models, CSI may include the actual channel matrix, an index to an entry in a codebook of CSI, a set of eigenvectors of the channel matrix, and / or a compressed and / or quantized version of this information.

[0092] Methods may be implemented to reduce the CSI feedback overhead by predicting the current / future CSI parameters from previous instances and sending only the residual information (the difference between the predicted channel information and the actual estimated channel) in the CSI feedback (e.g. sending the residual CSI to a network node).

[0093] In 5G communication systems, CSI feedback consumes a large portion of the uplink bandwidth, thus introducing additional overhead that may degrade the system performance. The CSI parameters in these uplink messages contain information about the channel at each instance of time. Since the channel information usually does not change significantly in short time intervals, temporal correlations in the CSI parameters are not appropriately exploited in current implementations. Methods are implemented to reduce the CSI feedback bandwidth / overhead using CSI prediction and compressed residual feedback.

[0094] Methods may be implemented to reduce the CSI-feedback bandwidth. The CSI feedback overhead may be reduced by predicting the CSI using the information in previous instances of CSI. This method may reduce the bandwidth by not communicating CSI feedback and using the predicted CSI instead. There are many temporal correlations between subsequent channel state information. Knowing the CSI for the past few time instances can help predict the next instance efficiently. These temporal correlations may help predict CSI. Accordingly, a temporal predictor at gNB may exist that predicts the next CSI based on the previous ones. Methods may be further implemented to ensure that the CSI parameters in the WTRU and the gNB do not drift from each other. For example, a temporal CSI prediction model may be implemented at the gNB. The inputs to this model may be previous instances of CSI available at the gNB. However, since the gNB has no access to the most recent channel information, as time passes, the information fed to its model may drift away from the actual CSI estimated at the WTRU. Theseinaccuracies may result in inaccurate prediction of CSI, which may result in communication problems.

[0095] To address these inaccuracies, a mirrored temporal CSI prediction model may be utilized to predict the following CSI parameters and send the residual information from WTRU to the gNB. The residual values are received by the gNB and applied to its own predicted CSI parameters to correct small inaccuracies, preventing the CSI parameters at the WTRU and the gNB from drifting away from each other.

[0096] Since residual data may contain small values, these values may be quantized and compressed efficiently. In most cases, the predicted CSI may be so accurate that the residual values become very close to zero. As a result, a very short message can indicate zero residual data to the gNB, reducing the feedback bandwidth / overhead significantly. Accordingly, temporal prediction and residual feedback may help reduce CSI-feedback bandwidth and information drift between the WTR and the gNB.

[0097] Additionally, methods may be further implemented to initialize an end-to-end pipeline at startup when the past CSI may not be available. Methods may be implemented to predict future CSI to compensate for the CSI reporting / feedback delay. In some CSI reporting, the CSI may be estimated at the WTRU and signaled back to the gNB for downlink processing, such as choosing a modulation scheme and / or Ml MO precoding for beamforming. The CSI report received at the gNB may be outdated due to feedback delay (e.g., communication delay, processing delay, and / or scheduling delay) encountered in transmission from the WTRU to the gNB. As a result, using a received CSI that has been delayed for downlink processing and scheduling may result in performance loss. To help minimize this performance loss, a strategy utilizing an AI / ML model for predicting the future CSI may be implemented, the AI / ML model compensating for any feedback delay and utilizing the predicted future CSI for downlink processing.

[0098] FIG. 3 is a schematic diagram illustrating an example end-to-end pipeline 300 for calculating residual information using CSI. For example, end-to-end pipeline 300 for calculating residual information using CSI may include WTRU 300A and gNB 300B. The end-to-end pipeline 300 may be implemented at the WTRU 300A based on a Channel 302A, CSI estimate channel Htpredicted CSI (Wt), circular queue 308A, compression 310A, quantization 312A, entropy coding 314A, dequantization 316A, decompression 318A, CSI prediction model 320A, queued CSI (7?t), residual CSI (Kt)324A, decompressed and dequantized residual CSI 325A, and / or previously queued CSI 326A. The end-to-end pipeline 300 may be implemented at the gNB 300B based on predicted CSI (Ht), circular queue 308B, decompression 310B,dequantization 312B, entropy decoding 314B, CSI prediction model 320B, queued CSI Ht, residual CSI 324B, decompressed and dequantized residual CSI Rt, and previously queued CSI.

[0099] FIG. 4 is a schematic diagram illustrating an example end-to-end alternative pipeline 400 for calculating residual information using compressed versions of CSI Ht. For example, end-to- end pipeline 400 for calculating residual information using CSI Htmay include WTRU 300A and gNB 300B. For example, WTRU 300A may perform compression 310A before the residual compressed channel information rtis calculated so that the value of estimated CSI is compressed. WTRU 300A may perform compression 402A so that predicted CSI (Ht) is compressed before being used to calculate residual compressed channel information rt. gNB 300B may perform decompression 310B after the compressed queued CSI is calculated so that the value of compressed queued CSI is decompressed before going to circular queue 308B. gNB 300B may perform compression 402B so that so that predicted CSI (Ht) is compressed before being used to calculate compressed queued CSI. FIG. 4 is similar to FIG. 3 and may contain the same elements, where the difference is that the end-to-end pipeline may use compressed versions of CSI to calculate residual information.

[0100] There may be three modes of operation for calculating CSI within an AI / ML system comprising the WTRU 300A and the gNB 300B, namely configuration mode, startup mode, and operational mode. Since previous instances of CSI are used to predict the next instances of CSI, the methods for such calculations may handle the case of when past CSI is not yet available (e.g., at the startup). In configuration mode, the pipelines at the WTRU 300A and the gNB 300B may be configured to allow the AI / ML system to start communicating and calculating CSI. In configuration mode, legacy CSI feedback may be used for communication while configuring the pipelines. Once the configuration of the pipelines at the WTRU 300A and the gNB 300B is complete, a “configuration complete” message may be sent. Once both sides of communication are ready, the system may go into startup mode at the beginning of the next frame of communication. The configuration message ensures synchronization between the WTRU 300A and the gNB 300B.

[0101] In startup mode, the circular queues 308A and 308B may be empty at WTRU 300A and gNB 300B. The outputs of the CSI prediction models 320A and 308B may be null since no previous instances of CSI are available to feed it. With CSI prediction (Ht) being zero, the residual CSI (Rt) may become identical to the estimated channel (Wt). A CSI compression model 310A may be used in this case to compress the estimated channel. The compressed CSImaybe quantized by quantization 312A, and entropy coded by entropy coding 314A before being transmitted to the gNB 300B.

[0102] At gNB 300B, when the CSI may be received, entropy decoded by entropy decoder 314B, dequantized by dequantization 312B, and decompressed by decompression 31 OB. This results in residual CSI Rt, which is close to the original CSI subject to compression and quantization errors. Since the circular queue 308B may be also empty at gNB 300B, the predicted CSI ( / ?t) may be zero, and the actual channel info used at the gNB Htmay be the same as the received CSI (e.g., Ht= Rt). This CSI may be stored in the circular queue 308B.

[0103] Similarly, at the WTRU 300A, the CSI before the entropy coding may be dequantized by dequantization 316A and decompressed by decompression 318A to extract in Rt(which in this case is equal is to Ht). This value may then be stored in the circular queue 308A. This value (Ht) may be exactly equal to the value that was obtained by the gNB 300B and inserted into its circular queue.

[0104] At this point, the circular queues of the WTRU 300A and the gNB 300B may include the same information, and the AI / ML system may go into operational mode. In operational mode, after processing the first channel information, the circular queue 308A may include an estimation of the previous CSIThis may notbeideal since the CSI prediction model 320A may expect several instances of past CSI as input. However, since the channel information does not change significantly in short periods of time, this information may still be used by the CSI prediction model 320A to predict the next CSI. As such, the CSI prediction model 320A (e.g., a sequential deep neural network (DNN) model) 320A may output the same input if fed with only one instance of past CSI

[0105] As a result, during the first few prediction cycles, the residual CSI Rtand Ktis expected to contain slightly larger values than in normal operation cases where all the previous instances are available for the CSI prediction model 320A. The residual CSI Rtand Ktmay be smaller and smaller as the circular queues are filled with new CSI. The sequential CSI prediction model 320A may start working at its full capacity once the number of CSI in the circular queue 308A becomes equal to the observation window size of the predictive model (e.g., the number of past instances required by the sequential CSI prediction model).

[0106] The communication pipeline of the AI / ML system may comprise configurable parameters that may be configured at the WTRU 300A and / or the gNB 300B. These configurable parameters may comprise a CSI prediction model ID, a CSI compression model ID, a residual compression model ID, a residual threshold, quantization configuration parameters, and / or abeamforming future channel index. The CSI prediction model ID may be a unique identification number for the CSI prediction models 320A, 320B used at the WTRU 300A and / or the gNB 300. The parameters “Observation Window Size” and “Prediction Window Size” may be inferred from the CSI prediction model 320A, 320B based on its model structure. The CSI prediction model 320A may be selected by the WTRU 300A from a set of available models based on the WTRU 300A capabilities (computation power and memory limits), the CSI feedback delay, the detected channel environment (e.g., Doppler and delay spread, and other factors). The selected CSI prediction model ID may be sent to the gNB 300B to use the same CSI prediction model 320B, or the CSI prediction model 320B may be different.

[0107] The gNB 300B may select the CSI prediction model based on the WRTU 300A capabilities, CSI feedback delay, and / or other factors. The selected CSI prediction model ID may be sent to the WTRU 300A to use the same CSI prediction model 320A as gNB 300B, or the CSI prediction model 320A may be different.

[0108] The CSI prediction model 320B selected by gNB 300B may not be available at the WTRU 300A. In this case, the gNB 300B may download the model information to the WTRU 300A and then configure it with the selected model ID.

[0109] Once the CSI prediction model is specified in the WTRU 300A and the gNB 300B, the following parameters may be inferred from the selected model properties. The observation window size may be defined as the number of past instances of CSI used as input to the CSI prediction model 320A, 320B. This may be equal to the minimum size of the circular buffers used at the WTRU 300A and the gNB 300B. Different CSI prediction models may be designed and trained with different observation window sizes.

[0110] The prediction Window Size may be defined as the number of future instances of CSI predicted by the CSI prediction model 320A, 320B. Because of the communication delay, the received CSI feedback at the gNB 300B may correspond to the past state of the communication channel. By the time the next downlink communication happens, significant changes may have occurred to the channel. The future channels predicted by the CSI prediction model 320B may be used to compensate for this delay in the next downlink communication.

[0111] The CSI compression model ID may be a unique identification number for the pair of CSI compression (e.g., 310A, 402A, 402B) / decompression (e.g., 310B, 318A) models used at the WTRU 300A and the gNB 300B. In the pipeline 300 illustrated in FIG. 3, these models are used in the startup mode, where the whole CSI is transmitted to the gNB 300B. In the alternative pipeline illustrated in FIG. 4, these models are used in both startup and operational modes. For example, the CSI compression model (e.g., 310A, 402A) may be selected by the WTRU 300Afrom a set of available models based on the WTRU capabilities (e.g., computation power and memory limits) and other factors. The selected CSI compression model ID may be sent to the gNB 300B to use the same CSI compression model (e.g., 402B) as the WTRU 300A. The gNB 300B may select the CSI compression model (e.g., 402B) based on the WTRU capabilities and other factors. The selected CSI compression model ID may be sent to the WTRU 300A to use the same CSI compression model (e.g., 31 OA, 402A) as the gNB 300B.

[0112] The model selected by the gNB 300B may not be available at the WTRU 300A. In this case, the gNB 300B may first download the model information to the WTRU 300A and then configure it with the selected model ID. The residual compression model ID may be a unique identification number for the pair of residual compression / decompression models used at the WTRU 300A and the gNB 300B. In the pipeline 300 illustrated in FIG. 3, these models are used in the operational mode where the residual information is transmitted to the gNB 300B. The residual compression / decompression models 31 OA, 31 OB, 318A may not be used in the alternative pipeline illustrated in FIG. 4. For example, the residual compression model 31 OA may be selected by the WTRU 300A from a set of available models based on the WTRU capabilities (e.g., computation power and memory limits) and other factors. The selected residual compression model ID may be sent to the gNB 300B to use the same residual compression model as the WTRU 300A.

[0113] The gNB 300B may select the residual compression model based on the WTRU capabilities and other factors. In this case, the selected residual compression model ID may be sent to the WTRU 300A to use the same residual compression model as the gNB 300B. The model selected by gNB 300B may not be available at the WTRU 300A. In this case, the gNB 300B may first download the model information to the WTRU 300Aand then configure it with the selected model ID.

[0114] Alternatively or additionally, a simpler compression algorithm such as pruning may be used. The pruning configuration parameters (e.g., pruning threshold) may be extracted from a lookup table based on the residual compression model ID.

[0115] The residual threshold may determine the minimum value of the residual information sent to the gNB 30B. If the residual information is smaller than the threshold value, a short message indicating zero residual (e.g., “zero residual message”) is sent to the gNB 300B. The threshold may indicate a lower bound for all the values in the residual channel matrix, latent vector, or eigenvectors. If all the values are smaller than the threshold value, the zero residual message may be sent. The WTRU 300A may decide the threshold value and send the selected value to the gNB 300B. As another example, the gNB 300B may decide the threshold value(e.g., based on the channel accuracy requirements) and configure the WTRU 300A by sending the selected value to WTRU 300A.

[0116] The quantization configuration parameters may comprise a quantization type, a codebook indicator, and / or a quantization range and bin size. The quantization type indicates the type of quantization. Different types of quantization may be used, such as uniform or codebook-based quantization. The codebook indicator may specify which codebook should be used for the quantization for codebook-based quantization cases. The quantization range may specify an upper and lower limit for the quantized values for the case of uniform quantization. The input values to the quantization process may be clipped to be in the specified range. The bin size specifies the granularity of the quantization, which may determine how many bits can be used to transmit each real value.

[0117] The beamforming future channel index may compensate for the delays associated with the CSI feedback. The gNB 300B may use the k’th instance of the predicted future CSI to calculate the precoding information. In this case, the WTRU 300A may use the same instance of the predicted future CSI to match the precoding information created and used at the gNB 300B. Since WTRU 300A and gNB 300B use the same future instance of the CSI, the input to the beamforming calculation at the WTRU and gNB exactly match, resulting in better beamforming performance. The WTRU 300A may decide the future instance index to use and send the selected value to the gNB 300B. As another example, the gNB 300B may decide this value based on the CSI feedback delay and configure the WTRU 300A by sending the selected value to WTRU 300A.

[0118] Reconfiguration may become necessary for the system to work efficiently due to changes in the channel environment, bandwidth requirements, communication failure and other factors. For example, the WTRU 300A may send a reconfiguration message based on reconfiguration becoming necessary. For example, the system may fall back to a traditional CSI feedback system while the reconfiguration is in progress. Once all the parameters are reconfigured at the WTRU 300A and gNB 300B, the system can start in the startup mode with the new configuration.

[0119] A CSI prediction model (e.g., CSI prediction model 320A, 320B) may be implemented, as described herein. The AI / ML framework for the CSI prediction model may be based on a sequential DNN model that receives previous instances of CSI (e.g., previous channel matrixes) as inputs and predicts the next CSI information (e.g., current and future channel matrixes).

[0120] FIG. 5 is a schematic diagram illustrating an example sequential deep neural network (DNN) based CSI prediction model. The input to the CSI prediction model 500 may be asequence of ‘m’ previous instances of the channel matrix estimates 502, and the output may be the predicted current channel matrix Ht504 and, optionally, up to ‘n’ future instances of the future channel matrixes. For example, two copies of the same CSI prediction model 500 may be used, one at the WTRU and one at the gNB.

[0121] The CSI prediction model 500 may be as simple as linear or non-linear mathematical extrapolation equations and as complex as cutting edge deep neural network sequential models based on recurrent neural network (RNN) techniques such as LSTMs and GRUs, or Transformer-based models. For example, the CSI prediction model 500 may be based on Convolutional embedding layers coupled with Transformer layers.

[0122] The sequential DNN based CSI prediction model 500 may be designed with several additional features. For example, the CSI prediction model 300B may output zero when no input is available (e.g., the startup mode when the circular queue is empty). The CSI prediction model 300B may output the same input when only one instance of previous CSI exists. The CSI prediction model 300B may predict the current / future CSI instances when given from 2 to ‘m’ number of previous instances of the CSI. The accuracy of the predictions may improve as more data becomes available at the input. The CSI prediction model 500 may predict more than one current / future CSI instance. The gNB, depending on the amount of the round-trip delay from CSI-RS to CSI-feedback, may use one of the future instances, such as Ht+1for the next downlink communication.

[0123] A model architecture may be provided, as described herein. The CSI prediction model may effectively utilize the sequence of past CSI matrices to predict the current or future CSI. The predictive model may effectively exploit the relationships of the input’s spatial (e.g., number of transmit and receive antennas) and temporal (e.g., number of past CSI) dimensions. Different schemes to extract spatial correlations and perform temporal predictions may be utilized.

[0124] In an example framework, a combination of stacked convolutional neural network (CNN) layers may extract spatial correlations and effectively perform temporal predictions. In another example framework, a combination of stacked convolutional neural network (CNN) layers may be used to extract spatial correlations, and multiple stacked long-short term memory (LSTM) and recurrent neural network (RNN) layers may be utilized to extract temporal relationships. In another example framework, a combination of stacked convolutional neural network (CNN) layers may be utilized to extract spatial correlations, and transformer layers with attention-based mechanisms may be utilized to perform temporal predictions. In another example framework, transformer layers may be utilized for spatial correlations, and multiple stacked long-short term memory (LSTM) and recurrent neural network (RNN) layers may be utilized to perform temporalpredictions. In another example framework, transformer layers may be utilized for spatial correlations and temporal predictions.

[0125] An example method for CSI prediction may be implemented, as described herein. The method for CSI prediction may relates to the prediction of current and future CSI from a set of previous instances. The method for CSI prediction may be implemented both at the WTRU and the gNB. The WTRU and gNB may be configured with a 2-sided CSI prediction model. The CSI prediction model may include such information as the AI / ML model IDs, quantization parameters, threshold values, circular buffer parameters, and other information. The WTRU and gNB may initiate startup mode. At the initiation of the startup mode, the circular queues may be flushed empty. While in startup mode, the CSI prediction module may output a null prediction. (Ht= 0). Since the circular queue is empty at startup, the CSI prediction model cannot predict the CSI. The WTRU and gNB may transition to operation mode. In operational mode, the circular queues at the WTRU and gNB may have one or more CSI related to past instances. The WTRU and gNB may fetch as many previous instances of CSI as available in the circular queue. If the circular queues at the WTRU and gNB contain only one previous instance of CSI, the CSI prediction model may output the instance of CSI available without any modification (Ht= Hf-^. If the circular queue at the WTRU and gNB include two or more instances of previous CSI, the CSI prediction model may predict the current CSI and / or up to “n” future CSI instances based on the previous instances from the circular queue. The WTRU and gNB may maintain the operation mode for the next cycles of channel prediction. The WTRU and / or gNB may determine and indicate when to switch to startup mode by flushing their circular queues (e.g., if a communication failure occurs). The WTRU and gNB may switch to startup mode after a reconfiguration.

[0126] A residual CSI feedback method may be implemented, as described herein. The residual CSI feedback may be calculated at the WTRU using the latest estimated and predicted CSI. The residual information may be calculated as a difference between the CSI (e.g., as shown in FIG. 3) or as the difference between the compressed versions of CSI (also known as latent vectors) (e.g., as shown in FIG. 4).

[0127] Since the gNB has its own CSI prediction model, the difference between the predicted and estimated channels at the WTRU may be sent to the gNB. Because the WTRU and gNB are implementing the same CSI prediction model, the WTRU may know what the gNB’s model would predict. Therefore, the residual channel information may be calculated at the WTRU by subtracting the predicted channel information from the estimated channel information.

[0128] FIG. 6 is a schematic diagram illustrating an example preparation of residual CSI feedback at the WTRU. Preparation of residual CSI feedback at the WTRU 600 may implement an estimated channel 602, predicted current instances of CSI 604, previous instances of CSI 606, and / or residual channel information 608.

[0129] At startup, the predicted CSI 604 may be zero (e.g., since no previous instances are available in the circular queue to feed the model). Therefore, the residual channel information 608 may be the same as the actual estimated channel 602. This may occur once in the startup mode. Compression, quantization, and / or entropy coding methods may be implemented, as described herein.

[0130] Assuming good prediction performance of the sequential models, the difference between the predicted CSI 604 and the actual estimated CSI 602 may be insignificant. Therefore, the residual CSI 608 may include small values (e g., below a threshold) that may be compressed and quantized efficiently with low bandwidth requirements. The residual CSI 608 may be so small as to assume that Rt= 0. In such cases, a short message may be sent to inform the gNB that the predicted CSI may be used without any correction.

[0131] The residual CSI 608 may include small values (e.g., below a threshold) that are easily compressed and quantized with low bandwidth requirements. However, at the startup, where the predicted CSI 604 is not available (assumed zero), the residual CSI 608 may become equal to the estimated CSI (Rt= Ht). Therefore, in the pipeline for calculating residual information using CSI (e.g., as illustrated in FIG. 3), two types of compression methods for the startup and operational modes may be implemented. Alternatively or additionally, in the alternate pipeline for calculating residual information using compressed versions of CSI (e.g., as illustrated in FIG. 4), one compression model may be implemented, since the compression may happens before the calculation of the residual information.

[0132] FIG. 7 is a schematic diagram illustrating an example end-to-end pipeline 700 for compression, quantization, and entropy coding. End-to-end pipeline 700 for compression, quantization, and entropy coding may include a WTRU 702 and / or a gNB 704. The WTRU 702 may perform compression, quantization and / or entropy encoding of residual or estimated CSI 706 by implementing encoder 708, latent vectors 710, quantizer 712, and / or entropy encoder 714. The gNB 704 may generate residual or estimated CSI 724 by implementing entropy decoding, de-quantization, and / or decompression based on entropy decoder 716, de-quantizer 718, latent vectors 720, decoder 722.

[0133] At startup, the predicted CSI may not be available (e.g., assumed to be zero). In such a case, the residual CSI 706 may include the whole information in the estimated CSI (Rt= Ht).This CSI may be compressed / decompressed using a DNN-based compression model, such as encoder 708 and / or decoder 722. For example, the input CSI may be fed to the encoder, which compresses the information into smaller latent vectors 710 that are then quantized by quantizer 712, and entropy coded by entropy encoder 714 before being transmitted to gNB 704. Upon receiving this data, gNB 704 may entropy decode via entropy decoder 716 and dequantize via de-quantizer 718 the latent vectors 720 and feed them to decoder 722 part of an encoder model to recover the original CSI 724.

[0134] After this first step in the startup mode, the circular queues at the gNB and the WTRU may include one instance of CSI (Ht), which may be used for the next iteration of CSI prediction. In the operational mode, the residual values may be small (e.g., below a threshold), and a simple compression method such as pruning with a small threshold may suffice. Moreover, a compression model may not be implemented as the quantization and entropy coding may provide enough compression. The residual CSI may be so small that it can be assumed that Rt= 0. In such a case, a short message may be sent to inform the gNB that the predicted CSI may be used without any correction. In the alternate pipeline (e.g., as illustrated in FIG. 4), one compression model may be implemented since the compression may occur before the calculation of the residual information. Accordingly, the same compression / decompression model can be used for both startup and operational modes.

[0135] FIG. 8 is a schematic diagram illustrating an example transformer-based models 800 utilized for encoder and decoder blocks. Transformer-based models 800 may include encoder 802 and decoder 804. The encoder 802 may include a convolutional neural network (CNN) layer 808a configured to receive input data. The encoder 802 may include transformer block layers 806a. The encoder 802 may include a fully connected layer 808a configured to output of the encoded information. The decoder 804 may receive the encoded information at the fully connected layer 808b. The decoder 804 may include transformation blocks 806b and / or a CNN layer 808b configured to output the decoded information. As shown in FIG. 8, the transformer block layers 806a, 806b may each include an attention layer or multi-head attention, one or more add & norm layers and / or a feed forward layer.

[0136] A transformer-based encoder-decoder model may be used for both the CSI compression and the residual compression use cases. The type of encoder-decoder models used may have no bearing on the overall framework. Accordingly, different kinds of AI / ML or non-AI / ML encoder-decoder models may be utilized with the same framework.

[0137] FIG. 9 is a flow chart illustrating an example WTRU method 900 for calculating and transmitting residual information using CSI. As shown in FIG. 9, at 902, the WTRU may beconfigured with a 2-sided CSI prediction model. The CSI prediction model may include such information as the AI / ML model IDs, quantization parameters, threshold values, circular buffer parameters, and other relevant information. The WTRU may start at the startup mode, and the circular queue may be flushed empty.

[0138] At 904, the WTRU may estimate the CSI (Hf) using a received CSI-RS information. At 906, the WTRU may compute the predicted CSI for a current time (Ht). If the circular queue is empty, the WTRU is in startup mode, and the predicted CSI for the current time is set to zero (Ht= 0). If the circular queue has only one entry (Ht-t), then the predicted CSI for the current time is set to that one entry (Ht= Ht- . Otherwise, the WTRU is in operational mode, and the WTRU computes the predicted CSI (t) using the configured CSI prediction module.

[0139] At 908, the WTRU may evaluate the residual CSI by subtracting the predicted CSI for a current time from the estimated CSI (Rt= Ht- Ht). At 910, the WTRU may compress the residual CSI. If in startup mode, the WTRU may use the configured CSI compression model to compress the CSI. If in operational mode, the WTRU uses the configured residual compression model to compress the residual CSI. At 912, the WTRU may quantize the compressed residual CSI using a configured quantizer. At 914, the WTRU may dequantize the quantized residual CSI. At 916, the WTRU may decompress the dequantized residual CSI (Rf) to obtain the dequantized residual CSI (Rt). The WTRU may decompress the dequantized residual CSI (Rt) using the CSI decompression model for the startup mode or the residual CSI decompression model for the operational model to obtain the dequantized residual CSI (Rt).

[0140] At 918, the WTRU may determine if a zero residual message needs to be sent instead of the residual information based on a configured threshold value. If it is determined that a zero residual message needs to be sent, at 920, the WTRU may set the CSI residual value to zero Rt= o), may calculate a corrected CSI value (Ht) by adding the CSI residual value (Rt) to the predicted CSI information for a current time (Ht) (Ht= Ht+ Rt) = Ht), and may append the predicted CSI (Ht) to the circular queue at 922. If currently in startup mode, the WTRU may switch to operational mode. At 924, the WTRU may send the zero residual message to the gNB.

[0141] Alternatively, or additionally, if it is determined that the zero residual message does not need to be sent, at 926, the WTRU may calculate a corrected CSI value (Ht) by adding the CSI residual value (Rt) to the predicted CSI (Ht) as outputted by the CSI prediction model and append the corrected CSI (Ht) to the circular queue at 928. If currently in startup mode, the WTRU may switch to operational mode. At 930, the WTRU may use the configured lossless entropy coder to compress the previously quantized residual CSI and send the entropy encodedresidual information to the gNB at 932. The WTRU may continually repeat this method 900 starting at the WTRU estimating the CSI information using the received CSI-RS information (Hc). If a communication failure is detected, the WTRU may repeat the method 900 starting at switching to startup mode. If a reconfiguration is required, the WTRU may repeat the method 900 starting at the configuration of the two-sided CSI prediction model.

[0142] FIG. 10 is a flow chart illustrating an example gNB method 1000 for calculating and transmitting residual information using CSI. As shown in FIG. 10, at 1002, the gNB may be configured with a 2-sided CSI prediction model, which may include information such as the AI / ML model IDs, quantization parameters, threshold values, circular buffer parameters, and other information. The gNB may start at the startup mode, and the circular queues may be flushed empty. If the circular queue is empty, the gNB may be operating in startup mode and Ht= 0. Otherwise, the gNB may be in operational mode, and the gNB may compute the predicted CSI (Ht) using the configured CSI prediction model.

[0143] At 1004, the gNB may receive the feedback message from the WTRU. The gNB may receive the bitstream containing the residual CSI (Rt) from the WTRU. At 1006, the gNB may determine if a zero residual message has been received. If a zero residual message has been received, the gNB may set the residual information to zero (Rt= 0) at 1008. Alternatively, if it is determined that a zero residual message has not been received, at 1010, the gNB may use the configured lossless entropy decoder to decode the received residual CSI (Rt) bitstream. At 1012, the gNB may use the configured de-quantizer to dequantize the decoded residual CSI (R bitstream. At 1014, the gNB may decompress the dequantized residual CSI (Rt) bitstream. If in startup mode, the gNB may use the configured CSI decompression model to decompress the CSI (Rt). If in operational mode, the gNB may use the configured residual decompression model to decompress the residual CSI (Rt).

[0144] At 1018, if the circular queue is empty, the gNB may be in startup mode, and the predicted CSI for the current time may be set to zero (Ht= 0). If the circular queue has one entry then the predicted CSI for the current time may be set to that one entry (Ht= Otherwise, the gNB may be in operational mode, and the gNB may compute the predicted CSI (Ht) using the configured CSI prediction module. At 1016, the gNB may calculate the corrected CSI (Ht) by applying the residual information (Rt) to the predicted CSI Ht) (Ht= Ht+ Rt). At 1020, the gNB may append the corrected CSI ( / ?t) to the circular queue. At 1022, the gNB may use the corrected CSI corresponding to the current time (Wt) or any of thepredicted future CSI (Ht+1to Ht+n) for downlink communication. The gNB may continually repeat the method 1000 starting at determining If the circular queue is empty.

[0145] If a communication failure is detected, the gNB may repeat the method 1000 starting at may switching to startup mode. If a reconfiguration is required, the gNB may repeat the method 1000 starting with the configuration of the 2-sided CSI prediction model.

[0146] An example WTRU method for calculating and transmitting residual information using compressed CSI may be implemented using the end-to-end pipeline as shown in FIG. 4. For example, the WTRU may be configured with a 2-sided CSI prediction model. The CSI prediction model may include such information as the AI / ML model IDs, quantization parameters, threshold values, circular buffer parameters, and other relevant information. The WTRU may start at the startup mode, and the circular queue may be flushed empty. The WTRU may estimate the CSI using the received CSI-RS information ( / 7f). The WTRU may compute the predicted CSI for the current time ( / ?t). If the circular queue is empty, the WTRU may be in startup mode, and the predicted CSI for the current time is set to zero (Ht= 0). If the circular queue has one entrythen the predicted CSI for the current time may be set to that one entry (Ht= Hc_ ,). Otherwise, the WTRU may be in operational mode, and the WTRU may compute the predicted CSI (Ht) using the configured CSI prediction module.

[0147] The WTRU may use the configured CSI compression model to compress the estimated CSI (ht) and to compress the predicted CSI (ht). The WTRU may evaluate the residual compressed CSI (residual latent vector) by subtracting the compressed predicted CSI (ht) from the compressed estimated CSI (ht) (rt= ht- ht). The WTRU may use the configured quantizer to quantize the residual latent vector (rt). The WTRU may dequantize the quantized residual latent vector ft).

[0148] The WTRU may determine if a zero residual message is to be sent instead of the residual information based on a configured threshold value.If it is determined that a zero residual message i to be sent, then the WTRU may set the dequantized residual compressed CSI (residual latent vector) to zero (ft= 0). The WTRU may calculate a corrected compressed predicted CSI ht) by adding the dequantized residual compressed CSI (rt) to the compressed predicted CSI ht) and decompresses the corrected compressed predicted CSI (ht) to obtain the corrected CSI (Wt). The WTRU may store the corrected CSI (Wt) into the circular queue and send the zero residual message to the gNB.Alternatively, if it is determined that a zero residual message does not need to be sent, then the WTRU may use the configured lossless entropy coder to compress the residual latent vector (rt) bitstream and send the encoded residual latent vector (rt) bitstream to the gNB.The WTRU may continually repeat the method 1000 starting at the WTRU estimating the CSI information using the received CSI-RS information (Ht).

[0149] If a communication failure is detected, the WTRU may continually repeat the method starting at switching to startup mode. If a reconfiguration is required, the WTRU may repeat the method starting the configuration of the 2-sides CSI prediction model.

[0150] An example gNB method for calculating and transmitting residual information using compressed CSI may be implemented. The gNB may be configured with a 2-sided CSI prediction model, which includes information such as the AI / ML model IDs, quantization parameters, threshold values, circular buffer parameters, and other information. The gNB may start at the startup mode, and the circular queues may be flushed empty. If the circular queue is empty, the gNB may be operating in startup mode and Ht= 0. Otherwise, the gNB may be in operational mode, and the gNB computes the predicted CSI (Ht) using the configured CSI prediction model. The gNB may use the configured CSI compression model to compress the predicted CSI (Ht-> ht). The gNB may receive the feedback message from the WTRU. The gNB may receive the bitstream containing the residual CSI (ft) from the WTRU. The gNB may determine if a zero residual message has been received. If it is determined that a zero residual message has been received, the gNB may set the residual information to zero (rt= 0). Alternatively or additionally, if it is determined that a zero residual message has not been received, the gNB may use the configured lossless entropy decoder to decode the received residual CSI (ft) bitstream and use the configured de-quantizer to dequantize the decoded received residual CSI (rt) bitstream.

[0151] The gNB may calculate the corrected latent vector (ht) by applying the residual latent vector (ft) to the predicted CSI (ht) (ht= ht+ rt). The gNB may decompress the corrected latent vector to obtain the corrected CSI (Ht). The gNB may append the corrected CSI (Ht) to the circular queue. The gNB may use the corrected CSI corresponding to the current time (Ht) or any of the predicted future CSI (Ht+1to Ht+n) for downlink communication. The gNB may continually repeat the method starting with entering the startup mode. If a communication failure is detected, the gNB may repeat the method starting with entering the startup mode. If a reconfiguration is required, the gNB may repeat the method starting with the configuration of the 2-sided prediction model.

[0152] Quantization and entropy encoding may be implemented, as described herein. A simple linear quantization or a codebook-based quantization / dequantization method may be used in the WTRU and gNB. Any variation of lossless arithmetic coding methods may be used to serialize the quantized residual CSI into a bitstream and transmit it from the WTRU to the gNB.

[0153] Duplicate prediction pipelines at the WTRU and the gNB may be implemented. To help prevent the CSI at the WTRU and gNB from drifting away from each other due to small differences, the input to the CSI prediction model may be the same in the WTRU and gNB. Although the actual estimated channel is available to the WTRU, it may not be used as the model input. Instead, the residual information is dequantized and decompressed to get Rtwhich is applied to the predicted CSI (Ht) to get CSI used by the gNB (Ht). This information may be stored in the circular queue and used in the next iteration. As a result, the contents of the circular queues at the WTRU and gNB may always match.

[0154] A method and system for predicting future CSI may be implemented using a set of previously estimated CSI. For example, three different deep neural network (DNN) structures may be implemented for temporal predictions of CSI. For instance, a convolutional / transformer (CNN-Xformer) based deep neural network may be implemented for predicting the future CSI information using the past instances of CSI. A convolutional / Long-Short Term Memory (CNN- LSTM) based deep neural network may be implemented for predicting the future CSI information using the past instances of CSI. A transformer / Long-Short Term Memory (Xformer- LSTM) based deep neural network may be implemented for predicting the future CSI information using the past instances of CSI.

[0155] FIG. 11 is a schematic diagram of an example training architecture 1100 for an AI / ML CSI prediction solution. As shown in FIG. 11 , previous (L-1) instances of CSI 1102 may be fed to a deep neural network 0t, 1104 which predicts a next CSI instance 1106. During the training of the deep neural network 0t1104, the predicted next CSI instance may be compared with ground truth values, and an error metric (MSE) 1108 may be used to update the parameters of the deep neural network 0t1104. The deep neural network 0tmay comprise two internal modules that capture spatial (1110) and temporal correlation (1112) in the input data (e.g., previous (L-1) instances).

[0156] Table 1 below demonstrates how the two internal modules that capture spatial and temporal correlation in the input data may be configured. As shown in Table 1 , the deep neural network 0tmay comprise different types of DNNs that, in combination, capture spatial and temporal correlation in the input data.Table 1

[0157] FIG. 12 is a schematic diagram of example structures 1200 for the AI / ML CSI prediction model 0t. As shown in FIG. 12, the Xformer structure 1202a may be comprised of a Transformer neural network layer and a recurrent neural network (RNN) layer (e.g., Long-Short Term Memory (LSTM) network layer). The CNN-LSTM structure 1204b may be comprised of a convolutional neural network (CNN) layer and a LSTM network layer. The CNN-Xfomner structure 1206c may be comprised of a CNN layer and a transformer network layer.

[0158] The CNN layers may be used to capture the spatial correlations in each CSI instance. The LSTM network layers may be used to capture the temporal correlations between different CSI instances. The transformer network layers may be used to capture both spatial and temporal correlations (Xformer-LSTM and CNN-Xformer structures correspondingly).

[0159] Table 2 below demonstrates example DNN structures of transformer network layers that may be used for spatial and temporal feature extraction within the deep neural network 0tstructure.Table 2

[0160] Table 3 below demonstrates example DNN structures of CNN layers that may be used for spatial feature extraction within the deep neural network 0tstructure.Table 3

[0161] Table 4 below demonstrates example DNN structures of LSTM network layers that may be used for temporal feature extraction within the deep neural network 0tstructure.Table 4

Claims

CLAIMS:

1. A wireless transmit / receive unit (WTRU) comprising: a processor configured to: receive configuration information for a channel state information (CSI) prediction model; store previous instance of estimated CSI in a circular queue to be used as inputs to the CSI prediction model; determine a current estimated CSI associated with a channel at a current time based on CSI reference signals (CSI-RSs); and determine a predicted CSI associated with the channel at the current time based on the previous instances of estimated CSI in the circular queue.

2. The WTRU of claim 1 , wherein the configuration information comprises one or more of a CSI prediction model ID, CSI compression model ID, residual compression model ID, residual threshold, quantization configuration parameters, or beamforming future channel index.

3. The WTRU of claim 1 , wherein the processor is further configured to: receive a reconfiguration message; and transition to a startup mode based on the reconfiguration message, wherein the processor is further configured to flush the circular queue to empty at the transition to the startup mode.

4. The WTRU of claim 3, wherein the reconfiguration message is an indication of a communication failure.

5. The WTRU of claim 1 , wherein the processor is further configured to provide a null output from the CSI prediction model while the circular queue is empty.

6. The WTRU of claim 1 , wherein the processor is further configured to: determine a number of previous instances of CSI currently within the circular queue; and provide an instance of CSI identical to the CSI within the circular queue based on the determination that one previous instance of CSI currently resides within the circular queue.

7. The WTRU of claim 2, wherein the processor is further configured to:select the CSI prediction model ID based on one or more WTRU capabilities, a CSI feedback delay, or a detected channel environment.

8. The WTRU of claim 7, wherein the one or more WTRU capabilities comprise computation power and memory limits, and the detected channel environment comprises doppler and delay spread.

9. The WTRU of claim 2, wherein the CSI prediction model ID comprises an observation window size and a prediction window size.

10. The WTRU of claim 9, wherein the observation window size comprises a number of past instances of CSI information that are configured to be used as input to the CSI prediction model, and the prediction window size comprises a number of future instances of CSI information predicted by the CSI prediction model.

11. A method performed by a wireless transmit / receive unit (WTRU), the method comprising: receiving configuration information for a channel state information (CSI) prediction model; storing previous instance of estimated CSI in a circular queue to be used as inputs to the CSI prediction model; determining a current estimated CSI associated with a channel at a current time based on CSI reference signals (CSI-RSs); and determining a predicted CSI associated with the channel at the current time based on the previous instances of estimated CSI in the circular queue.

12. The method of claim 11 , wherein the configuration information comprises one or more of a CSI prediction model ID, CSI compression model ID, residual compression model ID, residual threshold, quantization configuration parameters, or beamforming future channel index.

13. The method of claim 11 , wherein the method further comprises: receiving a reconfiguration message; and transitioning to a startup mode based on the reconfiguration message, wherein the circular queue is flushed to empty at the transition to the startup mode.

14. The method of claim 13, wherein the reconfiguration message is an indication of a communication failure.

15. The method of claim 11 , wherein the method further comprises: providing a null output from the CSI prediction model while the circular queue is empty.

16. The method of claim 11 , wherein the method further comprises: determining the number of previous instances of CSI currently within the circular queue; and providing an instance of CSI identical to the CSI within the circular queue based on the determination that one previous instance of CSI currently resides within the circular queue.

17. The method of claim 12, wherein the method further comprises: selecting the CSI prediction model ID based on one or more WTRU capabilities, a CSI feedback delay, or a detected channel environment.

18. The method of claim 17, wherein the one or more WTRU capabilities comprises computation power and memory limits, and the detected channel environment comprises doppler and delay spread.

19. The method of claim 12, wherein the CSI prediction model ID comprises an observation window size and a prediction window size.

20. The method of claim 19, wherein the observation window size comprises a number of past instances of CSI information that are used as input to the CSI prediction model, and the prediction window size comprises a number of future instances of CSI information predicted by the CSI prediction model.

Images