Method and device for processing missing partial input of AI network
By using target input or default values to replace missing input in the AI network, the problem of missing input caused by discontinuous resource allocation is solved, and the performance of AI network and communication systems is improved.
Patent Information
- Application Number
- CN202011323033.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-11-23
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2040-11-23
AI Technical Summary
In the field of communication, the lack of inputs caused by discontinuous resource allocation of AI networks affects the performance of AI networks.
When there are missing inputs in multiple inputs of the AI network, the target input or default value is used to replace the missing input, and the output of the AI network is obtained, specifically including directly replacing the missing with the target input or processed results. Input, or perform operations on multiple target inputs instead of missing inputs.
It effectively solves the problem of missing input caused by discontinuous resource allocation, and improves the performance of AI network and the performance of communication systems.
Smart Images

Figure CN114531696B_ABST
Abstract
Description
Technical Field
[0001] The present application belongs to the field of communication technology, and specifically relates to a method and device for processing missing partial input of an AI network. Background Art
[0002] Artificial Intelligence (AI) is a new technical science that studies and develops theories, methods, technologies and application systems for simulating, extending and expanding human intelligence. By applying AI networks to actual business, various problems in various fields can be solved.
[0003] In the field of communications, the introduction of AI networks into the communication process is conducive to improving the performance of the communication system. However, in the current actual environment, since resource allocation cannot be guaranteed to be continuous, for example, discontinuity in the time domain and frequency domain, the input of the AI network will be missing, which will have a great impact on the performance of the AI network. Summary of the invention
[0004] The embodiments of the present application provide a method and device for processing missing input of an AI network, which can solve the problem that missing input of the AI network affects the performance of the AI network.
[0005] In a first aspect, a method for processing missing inputs of an AI network is provided, the method comprising: when there is a missing input among multiple inputs of the AI network, a first communication device replaces the missing input with a target input or a default value to obtain an output of the AI network.
[0006] In a second aspect, a first communication device is provided, comprising: a processing module, configured to, when there is a missing input among multiple inputs of an AI network, replace the missing input with a target input or a default value to obtain an output of the AI network.
[0007] In a third aspect, a terminal is provided, comprising a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the method described in the first aspect.
[0008] In a fourth aspect, a network side device is provided, which includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the method described in the first aspect.
[0009] In a fifth aspect, a readable storage medium is provided, on which a program or instruction is stored. When the program or instruction is executed by a processor, the method described in the first aspect is implemented.
[0010] In a sixth aspect, a computer program product is provided, which includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the method described in the first aspect is implemented.
[0011] In a seventh aspect, a chip is provided, comprising a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run a program or instruction to implement the method described in the first aspect.
[0012] In an embodiment of the present application, when there is a missing input among multiple inputs of the AI network, the first communication device uses a target input or a default value to replace the missing input to obtain the output of the AI network. By setting the target input or the default value, the problem of missing input caused by discontinuous resource allocation in the actual environment can be effectively solved, which is convenient for improving the performance of the AI network and the performance of the communication system. BRIEF DESCRIPTION OF THE DRAWINGS
[0013] Figure 1 is a schematic diagram of a wireless communication system according to an embodiment of the present application;
[0014] Figure 2 is a schematic flow chart of a method for processing missing inputs of an AI network according to an embodiment of the present application;
[0015] Figure 3 is a schematic diagram of a partial missing input of an AI network according to an embodiment of the present application;
[0016] Figure 4 is a schematic structural diagram of a first communication device according to an embodiment of the present application;
[0017] Figure 5 is a schematic structural diagram of a communication device according to an embodiment of the present application;
[0018] Figure 6 is a schematic diagram of the structure of a terminal according to an embodiment of the present application;
[0019] Figure 7 It is a schematic diagram of the structure of a network side device according to an embodiment of the present application. DETAILED DESCRIPTION
[0020] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of this application.
[0021] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first" and "second" are generally of the same type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally represents that the objects associated with each other are in an "or" relationship.
[0022] It is worth noting that the technology described in the embodiments of the present application is not limited to the Long Term Evolution (LTE) / LTE-Advanced (LTE-A) system, but can also be used in other wireless communication systems, such as Code Division Multiple Access (CDMA), Time Division Multiple Access (TDMA), Frequency Division Multiple Access (FDMA), Orthogonal Frequency Division Multiple Access (OFDMA), Single-carrier Frequency-Division Multiple Access (SC-FDMA) and other systems. The terms "system" and "network" in the embodiments of the present application are often used interchangeably, and the described technology can be used for the systems and radio technologies mentioned above as well as for other systems and radio technologies. The following description describes a New Radio (NR) system for illustrative purposes, and NR terminology is used in most of the following descriptions, but these technologies can also be applied to applications other than NR system applications, such as the 6th generation (6 th Generation, 6G) communication system.
[0023] Figure 1A schematic diagram of a wireless communication system applicable to an embodiment of the present application is shown. The wireless communication system includes a terminal 11 and a network side device 12. Among them, the terminal 11 can also be called a terminal device or a user terminal (User Equipment, UE), and the terminal 11 can be a mobile phone, a tablet computer (Tablet Personal Computer), a laptop computer (Laptop Computer) or a notebook computer, a personal digital assistant (Personal Digital Assistant, PDA), a handheld computer, a netbook, an ultra-mobile personal computer (ultra-mobile personal computer, UMPC), a mobile Internet device (Mobile Internet Device, MID), a wearable device (Wearable Device) or a vehicle-mounted device (VUE), a pedestrian terminal (PUE) and other terminal side devices, and the wearable device includes: a bracelet, a headset, glasses, etc. It should be noted that the specific type of the terminal 11 is not limited in the embodiment of the present application. The network side device 12 can be a base station or a core network, wherein the base station can be referred to as a node B, an evolved node B, an access point, a base transceiver station (Base Transceiver Station, BTS), a radio base station, a radio transceiver, a basic service set (Basic Service Set, BSS), an extended service set (Extended Service Set, ESS), a B node, an evolved B node (eNB), a next generation node B (gNB), a home B node, a home evolved B node, a WLAN access point, a WiFi node, a transmitting and receiving point (Transmitting Receiving Point, TRP) or some other suitable term in the field. As long as the same technical effect is achieved, the base station is not limited to a specific technical vocabulary. It should be noted that in the embodiment of the present application, only the base station in the NR system is taken as an example, but the specific type of the base station is not limited.
[0024] In combination with the accompanying drawings, the following detailed description is given of the method and device for processing missing partial input (i.e., the presence of missing input) of the artificial intelligence (AI) network provided in the embodiment of the present application through specific embodiments and their application scenarios.
[0025] like Figure 2 As shown, an embodiment of the present application provides a processing method 200 for missing partial input of an AI network. The method can be executed by a first communication device. In other words, the method can be executed by software or hardware installed on the first communication device. The first communication device can be a terminal or a network side device. The method includes the following steps.
[0026] S202: When there is a missing input among multiple inputs of the AI network, the first communication device replaces the missing input with a target input or a default value to obtain an output of the AI network.
[0027] like Figure 3 As shown, Figure 3 An AI network is schematically shown, which can obtain output based on N inputs, where N is an integer greater than 2.
[0028] In practical applications, input loss may occur because resource allocation cannot guarantee continuity, for example, discontinuity in the time domain or frequency domain. Figure 3 The input i in is missing. In this way, the target input or the default value can be used to replace the missing input, and then combined with other non-missing inputs to obtain the output of the AI network.
[0029] In one example, multiple inputs of the AI network are sorted in a target order, and the target input can be one or more inputs before the missing input, or one or more inputs after the missing input, or one or more inputs before the missing input and one or more inputs after the missing input. Figure 3 In the example shown, multiple inputs are sorted in the order of numbers 1 to N, and the i-th input is a missing input. The target input may be the (i-1)th input, or the (i-1)th input and the (i-2)th input, etc.; or, the target input may be the (i+1)th input, or the (i+1)th input and the (i+2)th input, etc.; or, the target input may be the (i-1)th input and the (i+1)th input. When there are multiple target inputs, operations may be performed on the multiple target inputs, such as averaging operations, and the operation results may be used to replace the missing inputs.
[0030] Optionally, the use of the target input to replace the missing input to obtain the output of the AI network mentioned in S202 may include: directly replacing the missing input with the target input to obtain the output of the AI network; or the target input may be subjected to certain processing, such as mathematical processing, signal processing, etc., and the processed result may be used to replace the missing input to obtain the output of the AI network. Optionally, the above-mentioned mathematical processing includes a combination of various common mathematical operations such as addition, subtraction, multiplication, division, Kth power, Kth square root, logarithm, derivation, partial derivative, etc. K is an arbitrary number. For example, K can be a positive number or a negative number or 0, a real number or a complex number.
[0031] Optionally, the multiple inputs of the AI network are sorted in a target order (such as sorted by sequence number), and the target input includes at least one of the following:
[0032] 1) The first N1 inputs of the missing input sequence, where N1 is 1, 2, 3, etc. For example, if the i-th input is missing, it is completed by the (i-N1), (i-N1-1), ..., (i-1)-th inputs.
[0033] 2) The N2 inputs after the sequence number of the missing input, where N2 is 1, 2, 3, etc. For example, if the i-th input is missing, it will be completed by the (i+1), (i+2), ..., (i+N2)-th inputs.
[0034] The default values used to replace missing inputs mentioned in various embodiments of this specification include one of the following: a constant value (e.g., 0, 1, etc.), a constant vector, a constant two-dimensional matrix, and a constant multi-dimensional matrix. The above-mentioned constant vectors, constant two-dimensional matrices, and constant multi-dimensional matrices can be determined according to the dimensions of multiple inputs of the AI network. For example, if each input is a two-dimensional vector, the default value for replacing the missing input is also a constant two-dimensional vector.
[0035] In various embodiments of the present specification, when there are multiple target inputs, using the target input to replace the missing input to obtain the output of the AI network may specifically include: replacing the missing input with a result after averaging or combining the multiple target inputs to obtain the output of the AI network.
[0036] Similarly, in various embodiments of the present specification, when there are multiple default values, using the default value to replace the missing input to obtain the output of the AI network may specifically include: replacing the missing input with a result after averaging or combining multiple default values to obtain the output of the AI network.
[0037] The average mentioned in the above two examples can be a linear average in the logarithmic domain or linear domain, a geometric average, a harmonic average, a square average, a weighted average, a minimum maximization, a maximum minimization, other common average methods, and their combinations and simple changes. The combination methods mentioned in the above two examples may also include a combination of various common mathematical operations such as addition, subtraction, multiplication, division, Kth power, Kth square root, logarithm, derivation, partial derivative, etc. K is an arbitrary number. For example, K can be a positive number or a negative number or 0, a real number or a complex number.
[0038] In the method for processing missing inputs of an AI network provided in an embodiment of the present application, when there is a missing input among multiple inputs of the AI network, the first communication device replaces the missing input with a target input or a default value to obtain the output of the AI network. By setting the target input or the default value, the problem of missing inputs caused by discontinuous resource allocation in an actual environment can be effectively solved, which is convenient for improving the performance of the AI network and the performance of the communication system.
[0039] In various embodiments of the present application, the first communication device may process the target service according to the AI network, and the input or output of the AI network may reflect the purpose of the AI network to a certain extent. Optionally, the input and / or output of the AI network includes one of the following:
[0040] 1) Reference signal. For example, AI networks are used for reference signal processing, including reference signal detection, filtering, equalization, etc. The reference signal may include a demodulation reference signal (DMRS), a sounding reference signal (SRS), a synchronization signal and a physical broadcast block (SSB), a tracking reference signal (TRS), a phase tracking reference signal (PTRS), a channel state information reference signal (CSI-RS), etc.
[0041] 2) Channel. For example, the AI network is used to send and receive signals, signaling or data on the channel. The channel includes, for example, the physical downlink control channel (PDCCH), the physical downlink shared channel (PDSCH), the physical uplink control channel (PUCCH), the physical uplink shared channel (PUSCH), the physical random access channel (PRACH), the physical broadcast channel (PBCH), etc.
[0042] 3) Channel state information.
[0043] For example, the AI network is used to obtain channel state information. Optionally, the channel state information may include one of the following: channel state information feedback information, and frequency division duplex (FDD) uplink and downlink partial reciprocal channel state information.
[0044] Among them, the channel state information feedback information includes: channel-related information, channel matrix-related information, channel characteristic information, channel matrix characteristic information, precoding matrix indication (Pre-coding Matrix Indicator, PMI), rank indication (Rank Indication, RI), channel quality indication (Channel Quality Indicator, CQI), channel state information reference signal resource indicator (CSI-RS Resource Indicator, CRI), layer indication (Layer Indicator, LI), etc.
[0045] Regarding the channel state information of partial reciprocity of FDD uplink and downlink, for the FDD system, based on partial reciprocity, the network side device obtains angle and delay information based on the uplink channel, and can notify the terminal of the angle information and delay information, or other channel state information with partial reciprocity, or directly estimated downlink channel information through the channel state information reference signal (Channel State Information-Reference Signal, CSI-RS) precoding or direct indication method. The terminal reports according to the indication of the network side device or selects and reports within the indication range of the network side device, or the terminal uses these downlink channel signals by itself, thereby reducing the terminal's calculation amount and the overhead of reporting the channel state information (Channel State Information, CSI).
[0046] 4) Beam information. The beam information includes beam quality, beam indication information (reference signal ID), beam failure indication information, new beam indication information in beam failure recovery, etc. For example, the AI network is used for beam management, including, for example, beam measurement, beam reporting, beam prediction, beam failure detection, beam failure recovery, and new beam indication in beam failure recovery.
[0047] 5) Channel prediction information. For example, the AI network is used to predict channel state information and beam prediction.
[0048] 6) Interference information. For example, the AI network is used to evaluate interference information. The interference information includes, for example, intra-cell interference, inter-cell interference, out-of-band interference, intermodulation interference, etc.
[0049] 7) Positioning information. For example, the AI network is used to obtain the positioning information of the first communication device. The positioning information may also include trajectory information, for example, estimating the specific position of the terminal or the possible future movement trajectory through a reference signal (such as SRS), and the position may include a horizontal position and / or a vertical position. Or information to assist in position estimation or trajectory estimation.
[0050] 8) Prediction information of high-level services or parameters. For example, the AI network is used to obtain prediction information of high-level services or parameters. The high-level services or parameters may include throughput, required data packet size, service requirements, mobile speed, noise information, etc.
[0051] 9) Management information of high-level services or parameters. For example, the AI network is used to obtain management information of high-level services or parameters. The high-level services or parameters may include throughput, required data packet size, service requirements, mobile speed, noise information, etc.
[0052] 10) Control signaling. For example, the AI network is used to generate control signaling, such as power control related signaling and beam management related signaling.
[0053] Optionally, the resource domains occupied by the multiple inputs of the AI network mentioned in various embodiments of the present application include at least one of the following: frequency domain, time domain, code domain, space domain, delay domain, Doppler domain, Fourier transform domain, S domain, and Z domain.
[0054] Specifically, in one example, multiple inputs of the AI network are resource blocks (RB) 1 to RB N in the frequency domain of time domain A (such as symbol A), where N is the total number of multiple inputs, and the resource domains occupied by the multiple inputs of the AI network include the frequency domain and the time domain.
[0055] The S domain mentioned above refers to the use of the imaginary exponent exp(jωt) as the basic signal in the frequency domain analysis to decompose the signal into many imaginary exponential components of different frequencies; the Z domain is the domain obtained by Z transform of the reference signal.
[0056] In the resource domain occupied by multiple inputs of the above-mentioned AI network, frequency domain resources can be divided by subcarrier, resource block (RB), physical resource block (PRB), subband, physical precoding resource block group (PRG), bandwidth part (BWP), etc.; time domain resources can be divided by symbol, time slot, half time slot, frame, subframe, radio frame, millisecond, second, and other common time units; code domain resources can be divided by orthogonal code, quasi-orthogonal code, semi-orthogonal code, etc.; spatial domain resources can be divided by antenna, antenna element, antenna panel, transmitting and receiving unit, beam, layer, rank, and antenna angle.
[0057] Optionally, each embodiment of this specification may also include the following steps: sending the output of the AI network to a second communication device; wherein the first communication device is a terminal and the second communication device is a network side device; or the first communication device is a network side device and the second communication device is a terminal; or the first communication device is a terminal and the second communication device is a terminal; or the first communication device is a network side device and the second communication device is a network side device.
[0058] In this embodiment, the second communication device may process the received output using a matching AI network, or may process the received output using a non-AI method, or the received output may be used directly.
[0059] To explain in detail the target input mentioned in each embodiment of the present application, several specific examples will be introduced below.
[0060] In one example, multiple inputs of the AI network occupy multiple resource domains, that is, the missing input also occupies multiple resource domains. For example, the multiple inputs of the AI network are RB 1 to RB N in the frequency domain of symbol A, the missing input is RB i in the frequency domain of symbol A, and the resource domains occupied by the missing input include the time domain and the frequency domain. The target input is obtained according to at least one of the following:
[0061] 1) The missing input corresponds to a resource in at least one resource domain among the multiple resource domains; for example, symbol A; or symbol A and RB i; or RB i.
[0062] 2) The missing input is one or more resources preceding the corresponding resource in at least one of the multiple resource domains; for example, one or more symbols adjacent to symbol A; or one or more symbols adjacent to symbol A and one or more RBs adjacent to RB i; or one or more RBs adjacent to RB i.
[0063] 3) The missing input is one or more resources following the corresponding resource in at least one of the multiple resource domains; for example, one or more symbols adjacent to symbol A; or one or more symbols adjacent to symbol A and one or more RBs adjacent to RB i; or one or more RBs adjacent to RB i.
[0064] In a specific example, the missing input includes the jth resource in the second resource domain of the i-th resource in the first resource domain, for example, information of the j-th RB in the i-th time slot is missing, and the target input includes one of the following:
[0065] 1) The (j-M1)th to (j-1)th resources of the second resource domain of the i-th resource of the first resource domain. For example, the missing input is the 5th (i.e., j=5)th RB of the 4th (i.e., i=4)th time slot, M1=2, and the target input in this example is: the 3rd RB of time slot 4 and the 4th RB of time slot 4.
[0066] 2) The (j+1)th to (j+M2)th resources of the second resource domain of the i-th resource of the first resource domain. For example, the missing input is the 5th (i.e., j=5)th RB of the 4th (i.e., i=4)th time slot, M2=2, and the target input in this example is: the 6th RB of time slot 4 and the 7th RB of time slot 4.
[0067] 3) The j-th resource in the second resource domain of the (i-N1)-th to (i-1)-th resources in the first resource domain.
[0068] 4) The (j-M1) to (j-1) resources of the (i-N1) to (i-1) resources of the first resource domain. For example, the missing input is the 5th (i.e., j=5) RB of the 4th (i.e., i=4) time slot, N1=2, M1=2, and the target inputs in this example are: the 3rd RB of time slot 2 and the 4th RB of time slot 2, the 3rd RB of time slot 3 and the 4th RB of time slot 3. Optionally, the value of N1 may be different for different resources of the first resource domain; the value of M1 may be different for different resources of the second resource domain.
[0069] 5) The (j+1)th to (j+M2)th resources in the second resource domain of the (i-N1)th to (i-1)th resources in the first resource domain. Optionally, the value of N1 may be different for resources in different first resource domains; or, the value of M1 may be different for resources in different second resource domains.
[0070] 6) The j-th resource in the second resource domain of the (i+1)-(i+N1)-th resources in the first resource domain.
[0071] 7) The (j-M1)th to (j-1)th resources in the second resource domain of the (i+1)th to (i+N1)th resources in the first resource domain. Optionally, the value of N1 may be different for different resources in the first resource domain; or, the value of M1 may be different for different resources in the second resource domain.
[0072] 8) The (j+1)th to (j+M2)th resources in the second resource domain of the (i+1)th to (i+N1)th resources in the first resource domain. Optionally, the value of N1 may be different for different resources in the first resource domain; or, the value of M1 may be different for different resources in the second resource domain.
[0073] The i, j, N1 and M1 mentioned in the above examples are positive integers. In the above different examples, the values of i, j, N1 and M1 can be the same or different.
[0074] In a specific example, the resource domains occupied by multiple inputs of the AI network include a first resource domain (such as a time domain) and a second resource domain (such as a frequency domain); wherein the missing input includes multiple adjacent second resources of the first resource in the first resource domain of the second resource domain; for example, the multiple inputs of the AI network are RB 1 to RB N in the frequency domain of symbol 5, and the missing input is RB i, RB(i-1) and RB(i+1) in the frequency domain of symbol 5. The target input is multiple second resources of the second resource domain of the resources near the first resource in the first resource domain. For example, the target input is RBi, RB(i-1) and RB(i+1) in the frequency domain of symbol 4; or, the target input is RB i, RB(i-1) and RB(i+1) in the frequency domain of symbol 3; or, the target input is RB i, RB(i-1) and RB(i+1) in the frequency domain of symbol 6; or, the target input is RB i, RB(i-1) and RB(i+1) in the frequency domain of symbol 7. Alternatively, the target input is a collection of one or more of the above examples, for example, the target input is RB i, RB(i-1) and RB(i+1) in the frequency domain of symbol 4, and RB i, RB(i-1) and RB(i+1) in the frequency domain of symbol 6.
[0075] It should be noted that although the resources “near” the first resource are mentioned here, in actual applications, the “near” here can be limited by a certain threshold, for example, the X1 adjacent resources before the first resource, the X2 adjacent resources after the first resource, etc., where X1 and X2 are positive integers.
[0076] In one example, the missing input is a plurality of adjacent inputs, and the target input is the first input before or after the missing input. For example, the multiple inputs of the AI network are RB 1 to RB N in the frequency domain, and the i-th input (ie, RBi) is missing and should be completed by RB i-1, but at this time both RB i and i-1 are missing, then RB i-2 replaces RB i-1.
[0077] In another example, the missing input is a plurality of adjacent inputs, and the target input is the target input of the first or last input among the plurality of adjacent inputs. In this example, a target input can be set in advance for each input of the AI network. If i, (i+1), ..., (i+M) of domain A are all missing, then the target inputs of these missing inputs are the target input of the i-th input or the target input of the (i+M)-th input.
[0078] Optionally, each of the above embodiments may further include the following steps: if the target input is missing or part of the target input is missing, the target input is replaced by an input near the target input or a target input of the next level of the target input to obtain the output of the AI network. For example, multiple missing inputs are not adjacent (not missing continuously), but the target input of a certain input is missing or part of the target input is missing, which is supplemented by the input near the target input or by the target input of the next level of the target input. In this example, for each input of the AI network, a target input can be set in advance, or a rule for finding the target input can be set in advance.
[0079] Optionally, the first communication device does not expect or the AI network does not allow at least one of the following to occur:
[0080] 1) Adjacent inputs are missing continuously;
[0081] 2) There are more than K1 consecutive missing inputs;
[0082] 3) There are more than K2 missing inputs;
[0083] K1 and K2 are integers greater than or equal to 2.
[0084] Optionally, different domains may be processed differently, for example, the time domain cannot be missing continuously; for example, the frequency domain can be missing continuously. The above K1 and K2 are related to the channel environment, the capability of the first communication device, and the network configuration. For example, when the network / terminal configures resources, the above situation cannot occur.
[0085] Optionally, each of the above embodiments further includes: if at least one of the following occurs, the AI network is no longer used:
[0086] 1) Adjacent inputs are missing continuously;
[0087] 2) There are more than K1 consecutive missing inputs;
[0088] 3) There are more than K2 missing inputs;
[0089] K1 and K2 are integers greater than or equal to 2.
[0090] For example, when the above situation occurs, other methods are used, such as other AI algorithms or other non-AI algorithms. For example, for CSI feedback, the AI method is no longer used, but the traditional Type I / Type II codebook is used.
[0091] Optionally, the AI network allows the missing inputs to satisfy a specific pattern or regularity. For example, some specific inputs are allowed to have missing patterns or regularities. For example, N2 inputs are allowed to be missing every N1 inputs.
[0092] Considering different terminal capabilities, some first communication devices allow missing inputs, while some first communication devices do not support missing inputs. Optionally, in parallel with the above embodiments, in some cases, the method further includes: if missing inputs occur in the AI network, no longer using the AI network; wherein the first communication device does not expect or the AI network does not allow missing inputs to occur.
[0093] Optionally, each of the above embodiments may further include the following step: sending at least one of the following to the second communication device:
[0094] 1) an identification of the missing input;
[0095] 2) the number of missing inputs;
[0096] 3) The pattern or regularity of the missing input;
[0097] 4) using an AI algorithm or a non-AI algorithm outside of the AI network to obtain the output;
[0098] 5) Performance loss caused by the missing input, such as signal-to-noise ratio, signal-to-interference-plus-noise ratio, power, interference, noise, throughput, MCS, etc.
[0099] The above-mentioned indication information may be additional output information of the AI network, or the indication information may be included in the existing output of the AI network. The second communication device may obtain the required information by inputting the indication information into the AI network. For the former, whether the input is missing may affect the overhead of the output information of the AI network.
[0100] It should be noted that the execution subject of the method for processing missing input of the AI network provided in the embodiment of the present application may be a first communication device, or a control module in the first communication device for executing the method for processing missing input of the AI network. In the embodiment of the present application, the first communication device executing the method for processing missing input of the AI network is taken as an example to illustrate the first communication device provided in the embodiment of the present application.
[0101] Figure 4 is a schematic diagram of the structure of a first communication device according to an embodiment of the present application, such as Figure 4 As shown, the first communication device 400 includes: a processing module 402, which can be used to replace the missing input with a target input or a default value to obtain the output of the AI network when there is a missing input among multiple inputs of the AI network.
[0102] In an embodiment of the present application, when there is a missing input among multiple inputs of the AI network, the first communication device uses a target input or a default value to replace the missing input to obtain the output of the AI network. By setting the target input or the default value, the problem of missing input caused by discontinuous resource allocation in the actual environment can be effectively solved, which is convenient for improving the performance of the AI network and the performance of the communication system.
[0103] Optionally, as an embodiment, the target input includes at least one of the following:
[0104] one or more inputs preceding the missing input;
[0105] one or more inputs following the missing input;
[0106] Wherein, the multiple inputs of the AI network are sorted according to the target order.
[0107] Optionally, as an embodiment, the resource domains occupied by the multiple inputs of the AI network include at least one of the following: frequency domain, time domain, code domain, space domain, delay domain, Doppler domain, Fourier transform domain, S domain, and Z domain.
[0108] Optionally, as an embodiment, the missing input occupies multiple resource domains, and the target input is obtained according to at least one of the following:
[0109] The missing input corresponds to a resource in at least one resource domain among the multiple resource domains;
[0110] The missing input is one or more resources preceding a corresponding resource in at least one resource domain among the plurality of resource domains;
[0111] The missing input is one or more resources subsequent to a corresponding resource in at least one resource domain among the plurality of resource domains.
[0112] Optionally, as an embodiment, the missing input includes the jth resource in the second resource domain of the i-th resource in the first resource domain, and the target input includes one of the following:
[0113] the j-M1th to j-1th resources in the second resource domain of the i-th resource in the first resource domain;
[0114] the j+1th to j+M2th resources in the second resource domain of the i-th resource in the first resource domain;
[0115] the jth resource in the second resource domain of the i-N1th to i-1th resources in the first resource domain;
[0116] resources from i-N1 to i-1 in the first resource domain and resources from j-M1 to j-1 in the second resource domain;
[0117] resources from i-N1 to i-1 in the first resource domain and resources from j+1 to j+M2 in the second resource domain;
[0118] the jth resource in the second resource domain of the i+1th to i+N1th resources in the first resource domain;
[0119] resources j-M1 to j-1 in the second resource domain of resources i+1 to i+N1 in the first resource domain;
[0120] resources from i+1 to i+N1 in the first resource domain and resources from j+1 to j+M2 in the second resource domain;
[0121] Among them, i, j, N1 and M1 are positive integers.
[0122] Optionally, as an embodiment, there are multiple target inputs, and the processing module 402 is used to replace the missing input with a result of averaging or combining the multiple target inputs to obtain the output of the AI network.
[0123] Optionally, as an embodiment, the default value includes one of the following: a constant value, a constant vector, a constant two-dimensional matrix, a constant multi-dimensional matrix.
[0124] Optionally, as an embodiment, the resource domains occupied by the multiple inputs of the AI network include a first resource domain and a second resource domain;
[0125] Wherein, the missing input includes a plurality of adjacent second resources in the second resource domain of the first resource in the first resource domain;
[0126] The target input is a plurality of second resources in the second resource domain of resources near the first resource in the first resource domain.
[0127] Optionally, as an embodiment,
[0128] The missing input is a plurality of adjacent inputs, and the target input is the first input before or after the missing input; or
[0129] The missing input is a plurality of adjacent inputs, and the target input is a target input of the first or last input among the plurality of adjacent inputs.
[0130] Optionally, as an embodiment, the processing module 402 is further configured to:
[0131] If the target input is missing or part of the target input is missing, the target input is replaced by an input near the target input or a next-level target input of the target input to obtain the output of the AI network.
[0132] Optionally, as an embodiment, the first communication device does not expect or the AI network does not allow at least one of the following to occur:
[0133] Adjacent inputs are missing consecutively;
[0134] There are more than K1 consecutive missing inputs;
[0135] There are more than K2 missing inputs in total;
[0136] K1 and K2 are integers greater than or equal to 2.
[0137] Optionally, as an embodiment, the processing module 402 is further configured to: if at least one of the following occurs, no longer use the AI network:
[0138] Adjacent inputs are missing consecutively;
[0139] There are more than K1 consecutive missing inputs;
[0140] There are more than K2 missing inputs in total;
[0141] K1 and K2 are integers greater than or equal to 2.
[0142] Optionally, as an embodiment, the AI network allows the missing input to satisfy a specific pattern or rule.
[0143] Optionally, as an embodiment, the processing module 402 is further configured to: if the AI network has missing input, no longer use the AI network;
[0144] Wherein, the first communication device does not expect or the AI network does not allow missing input.
[0145] Optionally, as an embodiment, the first communication device 400 further includes a sending module, configured to send at least one of the following to the second communication device:
[0146] an identification of the missing input;
[0147] the number of said missing inputs;
[0148] The pattern or regularity of the missing input;
[0149] Using an AI algorithm or a non-AI algorithm outside of the AI network to obtain the output;
[0150] The performance loss caused by the missing input.
[0151] Optionally, as an embodiment, the input and / or output of the AI network includes one of the following: a reference signal, a channel, channel state information, beam information, channel prediction information, interference information, positioning information, prediction information of high-level services or parameters, management information of high-level services or parameters, and control signaling.
[0152] Optionally, as an embodiment, the first communication device 400 further includes a sending module, configured to send the output of the AI network to a second communication device;
[0153] The first communication device 400 is a terminal, and the second communication device is a network side device; or
[0154] The first communication device 400 is a network side device, and the second communication device is a terminal; or
[0155] The first communication device 400 is a terminal, and the second communication device is a terminal; or
[0156] The first communication device 400 is a network side device, and the second communication device is a network side device.
[0157] According to the first communication device 400 of the embodiment of the present application, the process of the method 200 corresponding to the embodiment of the present application can be referred to, and the various units / modules in the first communication device 400 and the above-mentioned other operations and / or functions are respectively for implementing the corresponding processes in the method 200, and can achieve the same or equivalent technical effects. For the sake of brevity, they will not be repeated here.
[0158] The first communication device in the embodiment of the present application may be a device, or a component, an integrated circuit, or a chip in a terminal. The device may be a mobile terminal or a non-mobile terminal. For example, the mobile terminal may include but is not limited to the types of terminals 11 listed above, and the non-mobile terminal may be a server, a network attached storage (Network Attached Storage, NAS), a personal computer (personal computer, PC), a television (television, TV), a teller machine or a self-service machine, etc., which is not specifically limited in the embodiment of the present application.
[0159] The first communication device in the embodiment of the present application may be a device having an operating system. The operating system may be an Android operating system, an iOS operating system, or other possible operating systems, which are not specifically limited in the embodiment of the present application.
[0160] The first communication device provided in the embodiment of the present application can achieve Figure 2 The various processes implemented by the method embodiment and achieving the same technical effect are not described here to avoid repetition.
[0161] Optional, such as Figure 5 As shown, the embodiment of the present application also provides a communication device 500, including a processor 501, a memory 502, and a program or instruction stored in the memory 502 and executable on the processor 501. For example, when the communication device 500 is a terminal, the program or instruction is executed by the processor 501 to implement the various processes of the embodiment of the processing method for missing partial input of the above-mentioned AI network, and the same technical effect can be achieved. When the communication device 500 is a network side device, the program or instruction is executed by the processor 501 to implement the various processes of the embodiment of the processing method for missing partial input of the above-mentioned AI network, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0162] Figure 6 A schematic diagram of the hardware structure of a terminal for implementing an embodiment of the present application.
[0163] The terminal 600 includes but is not limited to: a radio frequency unit 601, a network module 602, an audio output unit 603, an input unit 604, a sensor 605, a display unit 606, a user input unit 607, an interface unit 608, a memory 609, and a processor 610.
[0164] Those skilled in the art will appreciate that the terminal 600 may also include a power source (such as a battery) for supplying power to various components, and the power source may be logically connected to the processor 610 through a power management system, thereby implementing functions such as managing charging, discharging, and power consumption management through the power management system. Figure 6 The terminal structure shown in the figure does not constitute a limitation on the terminal. The terminal may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently, which will not be described in detail here.
[0165] It should be understood that in the embodiment of the present application, the input unit 604 may include a graphics processor (Graphics Processing Unit, GPU) 6041 and a microphone 6042, and the graphics processor 6041 processes the image data of the static picture or video obtained by the image capture device (such as a camera) in the video capture mode or the image capture mode. The display unit 606 may include a display panel 6061, and the display panel 6061 may be configured in the form of a liquid crystal display, an organic light emitting diode, etc. The user input unit 607 includes a touch panel 6071 and other input devices 6072. The touch panel 6071 is also called a touch screen. The touch panel 6071 may include two parts: a touch detection device and a touch controller. Other input devices 6072 may include, but are not limited to, a physical keyboard, function keys (such as volume control keys, switch keys, etc.), a trackball, a mouse, and a joystick, which will not be repeated here.
[0166] In the embodiment of the present application, the radio frequency unit 601 receives downlink data from the network side device and sends it to the processor 610 for processing; in addition, the uplink data is sent to the network side device. Generally, the radio frequency unit 601 includes but is not limited to an antenna, at least one amplifier, a transceiver, a coupler, a low noise amplifier, a duplexer, etc.
[0167] The memory 609 can be used to store software programs or instructions and various data. The memory 609 can mainly include a program or instruction storage area and a data storage area, wherein the program or instruction storage area can store an operating system, an application program or instruction required for at least one function (such as a sound playback function, an image playback function, etc.), etc. In addition, the memory 609 can include a high-speed random access memory, and can also include a non-volatile memory, wherein the non-volatile memory can be a read-only memory (Read-Only Memory, ROM), a programmable read-only memory (Programmable ROM, PROM), an erasable programmable read-only memory (Erasable PROM, EPROM), an electrically erasable programmable read-only memory (Electrically EPROM, EEPROM) or a flash memory. For example, at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices.
[0168] The processor 610 may include one or more processing units; optionally, the processor 610 may integrate an application processor and a modem processor, wherein the application processor mainly processes an operating system, a user interface, and application programs or instructions, etc., and the modem processor mainly processes wireless communications, such as a baseband processor. It is understandable that the modem processor may not be integrated into the processor 610.
[0169] The processor 610 is used to replace the missing input with a target input or a default value to obtain the output of the AI network when there is a missing input among the multiple inputs of the AI network.
[0170] In an embodiment of the present application, when there is a missing input among multiple inputs of the AI network, the terminal uses a target input or a default value to replace the missing input to obtain the output of the AI network. By setting the target input or the default value, the problem of missing input caused by discontinuous resource allocation in the actual environment can be effectively solved, which is convenient for improving the performance of the AI network and the performance of the communication system.
[0171] The terminal 600 provided in the embodiment of the present application can also implement the various processes of the above-mentioned AI network partial input missing processing method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
[0172] Specifically, the embodiment of the present application also provides a network side device. Figure 7 As shown, the network side device 700 includes: an antenna 71, a radio frequency device 72, and a baseband device 73. The antenna 71 is connected to the radio frequency device 72. In the uplink direction, the radio frequency device 72 receives information through the antenna 71 and sends the received information to the baseband device 73 for processing. In the downlink direction, the baseband device 73 processes the information to be sent and sends it to the radio frequency device 72. The radio frequency device 72 processes the received information and sends it out through the antenna 71.
[0173] The frequency band processing device may be located in a baseband device 73 . The method performed by the network-side device in the above embodiment may be implemented in the baseband device 73 . The baseband device 73 includes a processor 74 and a memory 75 .
[0174] The baseband device 73 may include, for example, at least one baseband board on which a plurality of chips are arranged. Figure 7 As shown, one of the chips is, for example, a processor 74, which is connected to a memory 75 to call a program in the memory 75 to execute the network-side device operations shown in the above method embodiment.
[0175] The baseband device 73 may further include a network interface 76 for exchanging information with the radio frequency device 72 , and the interface may be, for example, a common public radio interface (CPRI for short).
[0176] Specifically, the network side device of the embodiment of the present invention further includes: instructions or programs stored in the memory 75 and executable on the processor 74, and the processor 74 calls the instructions or programs in the memory 75 to execute Figure 4The methods executed by the modules shown achieve the same technical effects, and therefore will not be described here in detail to avoid repetition.
[0177] The embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by the processor, each process of the embodiment of the processing method for missing partial input of the above-mentioned AI network is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.
[0178] The processor may be a processor in the terminal described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.
[0179] An embodiment of the present application further provides a chip, which includes a processor and a communication interface, wherein the communication interface is coupled to the processor, and the processor is used to run programs or instructions to implement the various processes of the above-mentioned AI network partial input missing processing method embodiment, and can achieve the same technical effect. To avoid repetition, it will not be repeated here.
[0180] It should be understood that the chip mentioned in the embodiments of the present application can also be called a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.
[0181] It should be noted that, in this article, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or device including the element. In addition, it should be pointed out that the scope of the methods and devices in the embodiments of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.
[0182] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal (which can be a mobile phone, computer, server, air conditioner, or network side device, etc.) to execute the methods described in each embodiment of the present application.
[0183] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.
Claims
1. A method for processing missing partial input of an artificial intelligence (AI) network, characterized in that: The method comprises: When there is a missing input among the multiple inputs of the AI network, the first communication device replaces the missing input with a target input or a default value to obtain an output of the AI network; The resource domains occupied by the multiple inputs of the AI network include at least one of the following: frequency domain, time domain, code domain, space domain, delay domain, Doppler domain, Fourier transform domain, S domain, and Z domain; The input and / or output of the AI network includes one of the following: reference signal, channel, channel state information, beam information, channel prediction information, interference information, positioning information, prediction information of high-level services or parameters, management information of high-level services or parameters, and control signaling; Among them, the resource domains occupied by multiple inputs of the AI network include a first resource domain and a second resource domain; the missing input includes multiple adjacent second resources in the second resource domain of the first resource in the first resource domain; and the target input is multiple second resources in the second resource domain of the resource near the first resource in the first resource domain.
2. The method according to claim 1, characterized in that The target input includes at least one of the following: one or more inputs preceding the missing input; one or more inputs following the missing input; Wherein, the multiple inputs of the AI network are sorted according to the target order.
3. The method according to any one of claims 1 to 2, characterized in that: The target input is multiple, and the target input is used to replace the missing input to obtain the output of the AI network, including: The missing input is replaced by the result of averaging or combining the multiple target inputs to obtain the output of the AI network.
4. The method according to claim 1, characterized in that: The default value includes one of the following: Constant value, constant vector, constant two-dimensional matrix, constant multidimensional matrix.
5. The method according to claim 1, characterized in that The missing input is a plurality of adjacent inputs, and the target input is the first input before or after the missing input; or The missing input is a plurality of adjacent inputs, and the target input is a target input of the first or last input among the plurality of adjacent inputs.
6. The method according to claim 1, characterized in that The method further comprises: If the target input is missing or part of the target input is missing, the target input is replaced by an input near the target input or a next-level target input of the target input to obtain the output of the AI network.
7. The method according to claim 1, characterized in that The first communication device does not expect or the AI network does not allow at least one of the following to occur: Adjacent inputs are missing consecutively; There are more than K1 consecutive missing inputs; There are more than K2 missing inputs in total; K1 and K2 are integers greater than or equal to 2.
8. The method according to claim 1, characterized in that The method further includes: if at least one of the following occurs, the AI network is no longer used: Adjacent inputs are missing consecutively; There are more than K1 consecutive missing inputs; There are more than K2 missing inputs in total; K1 and K2 are integers greater than or equal to 2.
9. The method according to claim 1, characterized in that: The AI network allows the missing inputs to satisfy a specific pattern or regularity.
10. The method according to claim 1, characterized in that The method further includes: if the AI network has missing input, no longer using the AI network; Wherein, the first communication device does not expect or the AI network does not allow missing input.
11. The method according to claim 1, characterized in that The method further includes: sending at least one of the following to the second communication device: an identification of the missing input; the number of said missing inputs; The pattern or regularity of the missing input; Using an AI algorithm or a non-AI algorithm outside of the AI network to obtain the output; The performance loss caused by the missing input.
12. The method according to claim 1, characterized in that The method further includes: sending an output of the AI network to a second communication device; The first communication device is a terminal, and the second communication device is a network side device; or The first communication device is a network side device, and the second communication device is a terminal; or The first communication device is a terminal, and the second communication device is a terminal; or The first communication device is a network side device, and the second communication device is a network side device.
13. A first communication device, characterized in that: include: A processing module, configured to, when there is a missing input among the multiple inputs of the AI network, replace the missing input with a target input or a default value to obtain an output of the AI network; The resource domains occupied by the multiple inputs of the AI network include at least one of the following: frequency domain, time domain, code domain, space domain, delay domain, Doppler domain, Fourier transform domain, S domain, and Z domain; The input and / or output of the AI network includes one of the following: reference signal, channel, channel state information, beam information, channel prediction information, interference information, positioning information, prediction information of high-level services or parameters, management information of high-level services or parameters, and control signaling; The resource domains occupied by the multiple inputs of the AI network include a first resource domain and a second resource domain; Wherein, the missing input includes a plurality of adjacent second resources in the second resource domain of the first resource in the first resource domain; The target input is a plurality of second resources in the second resource domain of resources near the first resource in the first resource domain.
14. The first communication device according to claim 13, characterized in that: The target input includes at least one of the following: one or more inputs preceding the missing input; one or more inputs following the missing input; Wherein, the multiple inputs of the AI network are sorted according to the target order.
15. The first communication device according to any one of claims 13 to 14, characterized in that: The target input is multiple, and the processing module is used to: The missing input is replaced by the result of averaging or combining the multiple target inputs to obtain the output of the AI network.
16. The first communication device according to claim 13, characterized in that: The default value includes one of the following: Constant value, constant vector, constant two-dimensional matrix, constant multidimensional matrix.
17. The first communication device according to claim 13, characterized in that: The missing input is a plurality of adjacent inputs, and the target input is the first input before or after the missing input; or The missing input is a plurality of adjacent inputs, and the target input is a target input of the first or last input among the plurality of adjacent inputs.
18. The first communication device according to claim 13, characterized in that: The processing module is further used for: If the target input is missing or part of the target input is missing, the target input is replaced by an input near the target input or a next-level target input of the target input to obtain the output of the AI network.
19. The first communication device according to claim 13, characterized in that: The first communication device does not expect or the AI network does not allow at least one of the following to occur: Adjacent inputs are missing consecutively; There are more than K1 consecutive missing inputs; There are more than K2 missing inputs in total; K1 and K2 are integers greater than or equal to 2.
20. The first communication device according to claim 13, characterized in that: The processing module is further configured to: if at least one of the following occurs, no longer use the AI network: Adjacent inputs are missing consecutively; There are more than K1 consecutive missing inputs; There are more than K2 missing inputs in total; K1 and K2 are integers greater than or equal to 2.
21. The first communication device according to claim 13, characterized in that: The AI network allows the missing inputs to satisfy a specific pattern or regularity.
22. The first communication device according to claim 13, characterized in that: The processing module is further configured to: if the AI network has missing input, no longer use the AI network; Wherein, the first communication device does not expect or the AI network does not allow missing input.
23. The first communication device according to claim 13, characterized in that: The first communication device further includes a sending module, configured to send at least one of the following to the second communication device: an identification of the missing input; the number of said missing inputs; The pattern or regularity of the missing input; Using an AI algorithm or a non-AI algorithm outside of the AI network to obtain the output; The performance loss caused by the missing input.
24. The first communication device according to claim 13, characterized in that: The first communication device further includes a sending module, configured to send the output of the AI network to a second communication device; The first communication device is a terminal, and the second communication device is a network side device; or The first communication device is a network side device, and the second communication device is a terminal; or The first communication device is a terminal, and the second communication device is a terminal; or The first communication device is a network side device, and the second communication device is a network side device.
25. A terminal, characterized in that: The method comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the method for processing missing partial input of an AI network as claimed in any one of claims 1 to 12.
26. A network side device, characterized in that: The method comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein when the program or instruction is executed by the processor, the method for processing missing partial input of an AI network as claimed in any one of claims 1 to 12 is implemented.
27. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the method for processing missing partial input of the AI network as described in any one of claims 1 to 12 is implemented.
