Model scheduling method and device based on navigation signal and storage medium

Abstract

The invention relates to the field of neural network models, and discloses a model scheduling method and device based on navigation signals and a storage medium. The method provided by the embodiment of the invention comprises the following steps: based on a pre-acquired navigation signal, determining the signal complexity of the navigation signal, the signal complexity being determined by comparing the numerical value of the navigation signal with a preset numerical value range; generating a scheduling instruction corresponding to the signal complexity based on the signal complexity; in response to the scheduling instruction, calling a corresponding convolutional layer in the signal processing model to process the navigation signal at a corresponding precision level, the number of the called convolutional layers and the precision level being in positive correlation with the signal complexity; and mapping the processed navigation signal according to the corresponding precision grade to generate an output signal. According to the method provided by the embodiment of the invention, the corresponding convolution layer and the precision level in the signal processing module can be scheduled according to the complexity of the navigation signal to process the navigation signal, so that the calculation amount and the power consumption during navigation signal processing are reduced.

Description

Technical Field

[0001] The present invention relates to the field of neural network models, and in particular to a model scheduling method, device and storage medium based on navigation signals. Background Art

[0002] With the development of satellite navigation technology, convolutional neural networks (CNNs) have been gradually integrated into satellite navigation chips to enable them to classify, detect, and predict trends in data such as spectrograms, pseudo-code residual maps, and channel energy maps. The current mainstream approach is to load static CNN models into the chip and run them to perform various data processing tasks. However, in practical applications, this "fixed structure + fixed accuracy" strategy has the following problems: 1. Due to the fixed model structure, computing resources cannot be scheduled on demand when executing tasks of different complexities, resulting in lightweight tasks being "redundantly executed with high precision", causing the chip to generate a large amount of unnecessary power consumption.

[0003] 2. Some data processing tasks do not require high computing accuracy (such as coarse classification and multipath counting), while some data processing tasks require high-precision computing (such as signal loss warning). The fixed CNN model cannot meet the requirements of multiple tasks with different computing accuracy due to the unified setting of computing accuracy. Summary of the Invention

[0004] Embodiments of the present invention provide a navigation signal-based model scheduling method, device, and storage medium to solve at least one of the above problems.

[0005] In a first aspect, an embodiment of the present invention provides a model scheduling method based on a navigation signal, comprising: generating a corresponding scheduling instruction based on a pre-acquired navigation signal; in response to the scheduling instruction, calling a corresponding convolutional layer in a signal processing model to process the navigation signal with a corresponding accuracy level; and mapping the processed navigation signal with a corresponding accuracy level to generate an output signal.

[0006] The navigation signal-based model scheduling method provided in an embodiment of the present invention can generate corresponding scheduling instructions based on the acquired navigation signal, thereby scheduling different convolutional layers in the signal processing model to process the navigation signal with corresponding levels of accuracy, so that the signal processing model can call corresponding levels of computing power and computing accuracy for processing according to the complexity of the navigation signal, avoiding wasting computing power and generating unnecessary power consumption when processing low-power tasks.

[0007] Optionally, the step of generating a corresponding scheduling instruction based on the pre-acquired navigation signal includes: determining a signal complexity of the navigation signal based on the pre-acquired navigation signal; and generating a scheduling instruction corresponding to the signal complexity based on the signal complexity.

[0008] Optionally, when the navigation signal is in the first numerical range, the signal complexity is the first level; when the navigation signal is in the second numerical range, the signal complexity is the second level; when the navigation signal is in the third numerical range, the signal complexity is the third level; wherein the first numerical range is greater than the second numerical range, and the second numerical range is greater than the third numerical range.

[0009] Optionally, in response to a scheduling instruction, the step of calling a corresponding convolutional layer in a signal processing model to process a navigation signal with a corresponding level of accuracy includes: in response to the scheduling instruction, calling at least one convolutional layer, or at least two convolutional layers, or at least three convolutional layers in the signal processing model; and processing the navigation signal with a corresponding level of accuracy based on the number of called convolutional layers.

[0010] Optionally, the step of processing the navigation signal with a corresponding accuracy level based on the number of called convolutional layers includes: configuring a corresponding number of output channels in the corresponding convolutional layer based on the number of called convolutional layers; obtaining an accuracy level corresponding to the number of called convolutional layers; and processing the navigation signal based on the called convolutional layers, the output channels configured for each convolutional layer, and the corresponding accuracy level.

[0011] Optionally, in the step of configuring a corresponding number of output channels in the corresponding convolutional layer, the model scheduling method based on the navigation signal includes: obtaining mask field information corresponding to the output channel; and configuring a corresponding number of output channels in the corresponding convolutional layer based on the mask field information.

[0012] Optionally, the processed navigation signal is mapped with a corresponding accuracy level, and the step of generating an output signal includes: based on the accuracy level used to process the navigation signal, mapping the numerical range of the processed navigation signal to a numerical range corresponding to the accuracy level to obtain a mapped navigation signal; and generating an output signal based on the mapped navigation signal.

[0013] Optionally, after generating the output signal, the model scheduling method based on the navigation signal also includes: generating quantization range information corresponding to the numerical range of the output signal based on the numerical range of the output signal; and updating the quantization parameters of the model used to receive the output signal based on the quantization range information.

[0014] Optionally, the navigation signal includes a signal-to-noise ratio signal and / or a residual signal and / or a phase-locked signal and / or an interference signal.

[0015] In a second aspect, an embodiment of the present invention provides a model scheduling device, comprising: a processor and a memory, wherein instructions are stored in the memory; the processor calls the instructions in the memory so that the processor executes the navigation signal-based model scheduling method of any of the aforementioned embodiments of the first aspect of the present invention.

[0016] The processor of the model scheduling device provided by the embodiment of the present invention executes the model scheduling method based on the navigation signal of any of the aforementioned embodiments of the first aspect of the present invention by calling the instructions in the memory, and can generate corresponding scheduling instructions according to the acquired navigation signal, thereby scheduling different convolutional layers in the signal processing model to process the navigation signal with corresponding accuracy levels, so that the signal processing model can call the corresponding degree of computing power and computing accuracy for processing according to the complexity of the navigation signal, thereby avoiding wasting computing power when processing low-power tasks and generating unnecessary power consumption.

[0017] In a third aspect, an embodiment of the present invention provides a computer-readable storage medium, in which instructions are stored. When the instructions are executed by a processor, the model scheduling method based on navigation signals of any of the aforementioned embodiments of the first aspect of the present invention is implemented.

[0018] The instructions stored in the computer-readable storage medium provided in an embodiment of the present invention can be called by a processor to execute the model scheduling method based on navigation signals of any of the aforementioned embodiments of the first aspect of the present invention, and corresponding scheduling instructions are generated according to the acquired navigation signals, thereby scheduling different convolutional layers in the signal processing model to process the navigation signals with corresponding accuracy levels, so that the signal processing model can call corresponding levels of computing power and computing accuracy for processing according to the complexity of the navigation signals, thereby avoiding wasting computing power and generating unnecessary power consumption when processing low-power tasks. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0020] Figure 1 Flowchart of a first embodiment of a model scheduling method based on navigation signals of the present invention; Figure 2 This is a flowchart of step S110 in the first embodiment of the model scheduling method based on navigation signals of the present invention; Figure 3 This is a flowchart of step S120 in the first embodiment of the model scheduling method based on navigation signals of the present invention; Figure 4 This is a flowchart of step S122 in the first embodiment of the model scheduling method based on navigation signals of the present invention; Figure 5 This is a flowchart of step S130 in the first embodiment of the model scheduling method based on navigation signals of the present invention; Figure 6 Flowchart of a second embodiment of the model scheduling method based on navigation signals of the present invention; Figure 7 A structural block diagram of a first embodiment of a satellite navigation device for executing the model scheduling method based on navigation signals of the present invention; Figure 8 A scheduling flow chart of a satellite navigation device for executing the model scheduling method based on navigation signals of the present invention; Figure 9 FIG2 is a schematic diagram showing the structural configuration of a first embodiment of a model scheduling method based on navigation signals of the present invention; Figure 10 This is a structural block diagram of the switching accuracy level of the first embodiment of the model scheduling method based on navigation signals of the present invention; Figure 11 This is a structural block diagram of an embodiment of the model scheduling device of the present invention. DETAILED DESCRIPTION

[0021] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0022] It should be noted that all directional indications such as up, down, left, right, front, back, etc. in the embodiments of the present invention are only used to explain the relative position relationship, movement status, etc. between the various components in a certain specific posture as shown in the accompanying drawings. If the specific posture changes, the directional indication will also change accordingly.

[0023] In addition, the descriptions of "first", "second", etc. in the present invention are for descriptive purposes only and should not be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" or "second" may explicitly or implicitly include at least one of such features. In addition, the technical solutions between the various embodiments can be combined with each other, but this must be based on the fact that they can be implemented by ordinary technicians in this field. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0024] For ease of understanding, the model scheduling method based on navigation signals according to an embodiment of the present invention is described below. Figure 1 As shown, the navigation signal-based model scheduling method in the embodiment of the present invention includes steps S110 to S130.

[0025] In step S110 , a corresponding scheduling instruction is generated based on the pre-acquired navigation signal.

[0026] like Figure 2 As shown, in some optional embodiments, step S110 includes steps S111 to S112.

[0027] In step S111 , the signal complexity of the navigation signal is determined based on the pre-acquired navigation signal.

[0028] In step S112 , based on the signal complexity, a scheduling instruction corresponding to the signal complexity is generated.

[0029] In this embodiment, the navigation signal includes a signal-to-noise ratio signal and / or a residual signal and / or a phase-locked signal and / or an interference signal.

[0030] Specifically, when the navigation signal is within a first numerical range, the signal complexity is at level 1. When the navigation signal is within a second numerical range, the signal complexity is at level 2. When the navigation signal is within a third numerical range, the signal complexity is at level 3.

[0031] The first numerical range is greater than the second numerical range, and the second numerical range is greater than the third numerical range.

[0032] When the signal-to-noise ratio (SNR) is high, the navigation signal strength is high and interference is low. When the SNR is low, the navigation signal strength is low and interference is high. The lower the residual value of the residual signal, the more stable the navigation signal. The phase-locked signal is used to determine whether carrier phase lock has been achieved during navigation. A stable phase indicates carrier phase lock. Phase jumps, fading, or interference indicate phase lock is not achieved. The interference signal is the interference strength of broadband, narrowband, or pulse signals generated by the external environment. The higher the interference signal strength, the greater the interference to the navigation signal.

[0033] In this embodiment, when the navigation signal is a signal-to-noise ratio signal, the first value range is greater than or equal to 40 dB-Hz, the second value range is 30 dB-Hz to 40 dB-Hz, and the third value range is less than 30 dB-Hz.

[0034] When the navigation signal also includes a residual signal, a phase-locked signal, and an interference signal, if the signal complexity is at the first level, that is, when the signal-to-noise ratio is greater than or equal to 40 dB-Hz, the residual value of the residual signal is the lowest, the phase of the phase-locked signal is stable, the carrier phase is locked, and there is no interference signal.

[0035] If the signal complexity is the second level, that is, when the signal-to-noise ratio signal is between 30 dB-Hz and 40 dB-Hz, the residual value of the residual signal is larger than the residual value at the first level but smaller than the residual value at the third level, the phase-locked signal is unlocked, and the interference signal is of medium intensity.

[0036] If the signal complexity is the third level, that is, when the signal-to-noise ratio is less than 30 dB-Hz, the residual value of the residual signal is larger than that at the second level, the phase-locked signal frequently loses lock, and the intensity of the interference signal is the largest.

[0037] Among them, the residual value of the residual signal, the frequency at which the phase-locked signal loses lock, and the strength of the interference signal can be preset and are not limited in this application.

[0038] After determining the signal complexity, a corresponding scheduling instruction is generated according to the level of signal complexity, so that the signal processing model can process the corresponding navigation signal according to the scheduling instruction. The specific process of calling the signal processing model is as follows.

[0039] In step S120 , in response to the scheduling instruction, the corresponding convolutional layer in the signal processing model is called to process the navigation signal with the corresponding accuracy level.

[0040] like Figure 3 As shown, in some optional embodiments, step S120 includes steps S121 to S122.

[0041] In step S121 , in response to the scheduling instruction, at least one convolutional layer, or at least two convolutional layers, or at least three convolutional layers in the signal processing model are called.

[0042] In step S122 , the navigation signal is processed with a corresponding level of accuracy based on the number of convolutional layers called.

[0043] In this embodiment, the signal processing model is a convolutional neural network (CNN) model. In response to a scheduling instruction, when the signal complexity is at the first level (i.e., the lowest complexity), at least one convolutional layer in the signal processing model is invoked to process the navigation signal. When the signal complexity is at the second level (i.e., the medium complexity), at least two convolutional layers in the signal processing model are invoked to process the navigation signal. When the signal complexity is at the third level (i.e., the highest complexity), at least three convolutional layers in the signal processing model are invoked to process the navigation signal.

[0044] Specifically, when processing a navigation signal with a first level of signal complexity, the signal processing model of this embodiment calls the convolution layer of the first layer of the signal processing model to process and implement rapid classification and simple judgment of the navigation signal. When processing a navigation signal with a second level of signal complexity, the convolution layers of the first and second layers of the signal processing model are called for multipath identification of the navigation signal. When processing a navigation signal with a third level of signal complexity, the convolution layers of the first, second, and third layers of the signal processing model are called to identify and calculate the pseudorange oscillation and sliding window period of the navigation signal.

[0045] That is, the navigation signal-based model scheduling method of the embodiment of the present invention increases the number of convolutional layers and output channels used to process the navigation signal and improves the accuracy level only when the navigation signal is unstable and interference occurs. When the received navigation signal is in a stable state and is not interfered with, the signal processing model defaults to processing the navigation signal at the first level of signal complexity.

[0046] like Figure 4 As shown, further, step S122 includes steps S1221 to S1223.

[0047] In step S1221, based on the number of called convolutional layers, a corresponding number of output channels is configured in the corresponding convolutional layer.

[0048] In step S1222, the accuracy level corresponding to the number of called convolutional layers is obtained.

[0049] Specifically, in step S1221, the step of configuring a corresponding number of output channels in the corresponding convolutional layer includes: obtaining mask field information corresponding to the output channels, and configuring a corresponding number of output channels in the corresponding convolutional layer based on the mask field information.

[0050] In step S1223, the navigation signal is processed based on the called convolutional layers, the output channels configured for each convolutional layer, and the corresponding accuracy level.

[0051] In this embodiment, when processing navigation signals with a signal complexity of level 1, the signal processing model invokes the convolutional layers of the first layer and processes the navigation signals with an accuracy level of INT2. In this case, the number of output channels configured is 4 to 8. When processing navigation signals with a signal complexity of level 2, the signal processing model invokes the convolutional layers of the first and second layers and processes the navigation signals with an accuracy level of INT4. In this case, the number of output channels configured is 8 to 16. When processing navigation signals with a signal complexity of level 3, the signal processing model invokes the convolutional layers of the first, second, and third layers and processes the navigation signals with an accuracy level of INT8. In this case, the number of output channels configured is 16 to 32.

[0052] When calling the corresponding convolutional layers and output channels, the uncalled convolutional layers and output channels skip the data processing of the navigation signal, thereby reducing unnecessary calculations and lowering the power consumption generated when the signal processing model is running.

[0053] In step S130 , the processed navigation signal is mapped with a corresponding accuracy level to generate an output signal.

[0054] like Figure 5 As shown, further, step S130 includes steps S131 to S132.

[0055] In step S131 , based on the accuracy level used to process the navigation signal, the numerical range of the processed navigation signal is mapped to a numerical range corresponding to the accuracy level to obtain a mapped navigation signal.

[0056] In step S132 , an output signal is generated based on the mapped navigation signal.

[0057] In this embodiment, based on the accuracy level when processing the navigation signal, the processed navigation signal is mapped according to the accuracy level, so that the numerical range of the processed navigation signal is the same as the numerical range corresponding to the accuracy level, thereby achieving numerical synchronization, thereby obtaining an output signal, and the output signal is used for subsequent further data processing flow.

[0058] Figure 6 This is a flow chart of a second embodiment of the model scheduling method based on navigation signals of the present invention. Some steps of the second embodiment are the same as those of the first embodiment. The differences between the two will be described below, and the similarities will not be detailed again.

[0059] like Figure 6 As shown, in the second embodiment of the present invention, after generating the output signal, the model scheduling method based on the navigation signal further includes steps S240 to S250. The steps of the model scheduling method based on the navigation signal in the second embodiment of the present invention are as follows: In step S210, a corresponding scheduling instruction is generated based on the pre-acquired navigation signal.

[0060] In step S220 , in response to the scheduling instruction, the corresponding convolutional layer in the signal processing model is called to process the navigation signal with the corresponding accuracy level.

[0061] In step S230 , the processed navigation signal is mapped with a corresponding accuracy level to generate an output signal.

[0062] In step S240 , based on the numerical range of the output signal, quantization range information corresponding to the numerical range of the output signal is generated.

[0063] In step S250 , the quantization parameter of the model for receiving the output signal is updated based on the quantization range information.

[0064] In this embodiment, the output signal is used for subsequent further signal processing. For example, the output signal is used for subsequent transmission to an LSTM (long short-term memory network) model, so that the LSTM model can match the accuracy level used for calculation according to the output signal, thereby updating the quantization parameter based on the quantization range information, thereby improving the accuracy and efficiency of the LSTM model when using the output signal for sequence modeling.

[0065] The model scheduling method based on navigation signals provided by the embodiment of the present invention can be applied to Figure 7 In the satellite navigation device shown, the satellite navigation device includes a receiving module 201 , a determining module 202 , a scheduling module 203 and an executing module 204 .

[0066] The receiving module 201 is used to receive the navigation signal, the judging module 202 is used to generate a corresponding scheduling instruction based on the pre-acquired navigation signal, the scheduling module 203 is used to call the corresponding convolution layer in the signal processing model to process the navigation signal with the corresponding accuracy level according to the scheduling instruction, and the executing module 204 is used to map the processed navigation signal with the corresponding accuracy level to generate an output signal.

[0067] like Figure 8 As shown, the receiving module 201 is the front end of the satellite navigation device. After the front end of the satellite navigation device processes the received navigation signal, it generates a corresponding spectrum diagram and / or pseudo-code residual diagram and transmits it to the judgment module 202. The judgment module 202 performs task level judgment and determines the complexity of the signal. Then, through the scheduling module 203 and the execution module 204, the signal processing model is scheduled and the corresponding accuracy level is selected. Finally, an output signal is generated and transmitted to the main control CPU or LSTM inference module (i.e., LSTM model) for further data processing.

[0068] like Figure 9 and Figure 10 As shown, Figure 9 2 shows that when the scheduling module 203 of the satellite navigation device calls the signal processing model according to the obtained scheduling instruction, it reads the calling method through the stored structure configuration table to call the corresponding convolution layer and output channel and shield the uncalled convolution layer and output channel. Figure 10 The scheduling module 203 of the satellite navigation device is Figure 10 The channel selection path of a single PE (Processing Element) shown selects the corresponding accuracy level, thereby completing the data processing of the navigation signal with the corresponding convolution layer, output channel and calculation accuracy level.

[0069] The navigation signal-based model scheduling method provided in an embodiment of the present invention includes: generating a corresponding scheduling instruction based on a pre-acquired navigation signal; in response to the scheduling instruction, calling a corresponding convolutional layer in a signal processing model to process the navigation signal with a corresponding accuracy level; and mapping the processed navigation signal with the corresponding accuracy level to generate an output signal.

[0070] The navigation signal-based model scheduling method provided in an embodiment of the present invention can generate corresponding scheduling instructions based on the acquired navigation signal, thereby scheduling different convolutional layers in the signal processing model to process the navigation signal with corresponding levels of accuracy, so that the signal processing model can call corresponding levels of computing power and computing accuracy for processing according to the complexity of the navigation signal, avoiding wasting computing power and generating unnecessary power consumption when processing low-power tasks.

[0071] For the above method embodiment, the embodiment of the present invention also provides the following Figure 11 A model scheduling device shown includes: a processor 301 and a memory 302, wherein the memory 302 stores instructions; the processor 301 calls the instructions in the memory 302 so that the processor 301 executes the model scheduling method based on navigation signals of any of the aforementioned embodiments of the present invention.

[0072] The navigation signal-based model scheduling method of the above-mentioned embodiment of the present invention includes: generating a corresponding scheduling instruction based on the pre-acquired navigation signal; in response to the scheduling instruction, calling the corresponding convolution layer in the signal processing model to process the navigation signal with a corresponding accuracy level; mapping the processed navigation signal with the corresponding accuracy level to generate an output signal.

[0073] The model scheduling device provided by an embodiment of the present invention can generate corresponding scheduling instructions based on the acquired navigation signal by implementing the above-mentioned model scheduling method based on the navigation signal, thereby scheduling different convolutional layers in the signal processing model to process the navigation signal with corresponding accuracy levels, so that the signal processing model can call the corresponding degree of computing power and computing accuracy for processing according to the complexity of the navigation signal, avoiding wasting computing power when processing low-power tasks and generating unnecessary power consumption.

[0074] Furthermore, the model scheduling device provided by the embodiment of the present invention may further include a communication interface 303 and a bus 304 , and the processor 301 , the memory 302 and the communication interface 303 are electrically connected via the bus 304 .

[0075] Memory 302 may include high-speed random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage. Communication between the system network element and at least one other network element is achieved through at least one communication interface 303 (which may be wired or wireless), and may use the Internet, a wide area network, a local area network, a metropolitan area network, etc. Bus 304 may be an ISA bus, a PCI bus, or an EISA bus. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 11 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0076] The processor 301 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by hardware integrated logic circuits in the processor 301 or by software instructions. The above processor 301 may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the various methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in conjunction with the embodiments of the present invention can be directly implemented and executed by a hardware decoding processor, or by a combination of hardware and software modules in the decoding processor. The software module can be located in a storage medium well-known in the art, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, etc. The storage medium is located in memory 302, and processor 301 reads information in memory 302 and, in conjunction with its hardware, completes the steps of the method of the aforementioned embodiment.

[0077] An embodiment of the present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to execute the steps of the above-mentioned navigation signal-based model scheduling method.

[0078] The computer-readable storage medium provided by an embodiment of the present invention stores data and computer-executable instructions of the above-mentioned model scheduling method based on navigation signals. The above-mentioned model scheduling method based on navigation signals includes: generating corresponding scheduling instructions based on pre-acquired navigation signals; in response to the scheduling instructions, calling the corresponding convolutional layer in the signal processing model to process the navigation signal with a corresponding accuracy level; and mapping the processed navigation signal with the corresponding accuracy level to generate an output signal.

[0079] The computer-readable storage medium provided by an embodiment of the present invention can generate corresponding scheduling instructions based on the acquired navigation signal by implementing the above-mentioned model scheduling method based on the navigation signal, thereby scheduling different convolutional layers in the signal processing model to process the navigation signal with corresponding levels of accuracy, so that the signal processing model can call upon corresponding levels of computing power and computing accuracy for processing according to the complexity of the navigation signal, thereby avoiding wasting computing power and generating unnecessary power consumption when processing low-power tasks.

[0080] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0081] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0082] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that the technical solutions described in the above embodiments can still be modified, or some of the technical features thereof can be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A model scheduling method based on navigation signals, characterized in that: The method comprises: determining, based on the pre-acquired navigation signal, a signal complexity of the navigation signal, wherein the navigation signal includes a signal-to-noise ratio signal and / or a residual signal and / or a phase-locked signal and / or an interference signal, and the signal complexity is determined by comparing a value of the navigation signal with a preset value range; Based on the signal complexity, generating a scheduling instruction corresponding to the signal complexity; In response to the scheduling instruction, calling corresponding convolutional layers in a signal processing model to process the navigation signal with a corresponding accuracy level, wherein the number of called convolutional layers and the corresponding accuracy level are positively correlated with the complexity of the signal; The processed navigation signal is mapped according to the corresponding accuracy level to generate an output signal.

2. The navigation signal-based model scheduling method according to claim 1, characterized in that: When the navigation signal is within the first value range, the signal complexity is at the first level; When the navigation signal is within the second value range, the signal complexity is at the second level; When the navigation signal is within a third value range, the signal complexity is at a third level; The first numerical range is larger than the second numerical range, and the second numerical range is larger than the third numerical range.

3. The model scheduling method based on navigation signals according to claim 1, characterized in that: The step of, in response to the scheduling instruction, calling a corresponding convolutional layer in a signal processing model to process the navigation signal with a corresponding accuracy level comprises: In response to the scheduling instruction, calling at least one convolutional layer, or at least two convolutional layers, or at least three convolutional layers in the signal processing model; The navigation signal is processed with the corresponding accuracy level based on the number of convolutional layers invoked.

4. The navigation signal-based model scheduling method according to claim 3, characterized in that: The step of processing the navigation signal with the corresponding accuracy level based on the number of called convolutional layers includes: Based on the number of called convolutional layers, configure the corresponding number of output channels in the corresponding convolutional layer; Obtaining the accuracy level corresponding to the number of convolutional layers called; The navigation signal is processed based on the called convolutional layers, the output channels configured for each convolutional layer, and the corresponding accuracy level.

5. The navigation signal-based model scheduling method according to claim 4, characterized in that: In the step of configuring a corresponding number of output channels in the corresponding convolutional layer, the method includes: Obtaining mask field information corresponding to the output channel; Based on the mask field information, a corresponding number of output channels is configured in the corresponding convolutional layer.

6. The navigation signal-based model scheduling method according to claim 1, characterized in that: The step of mapping the processed navigation signal with the corresponding accuracy level to generate an output signal comprises: mapping, based on the accuracy level for processing the navigation signal, a numerical range of the processed navigation signal to a numerical range corresponding to the accuracy level, to obtain the mapped navigation signal; The output signal is generated based on the mapped navigation signal.

7. The navigation signal-based model scheduling method according to claim 6, characterized in that: After generating the output signal, the method further includes: generating, based on the numerical range of the output signal, quantization range information corresponding to the numerical range of the output signal; Based on the quantization range information, a quantization parameter of a model for receiving the output signal is updated.

8. A model scheduling device, characterized in that: The model scheduling device includes: a processor and a memory, wherein instructions are stored in the memory; The processor calls the instructions in the memory to enable the model scheduling device to implement the navigation signal-based model scheduling method according to any one of claims 1 to 7.

9. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the navigation signal-based model scheduling method according to any one of claims 1 to 7 is implemented.

Images