A perception method and apparatus

Abstract

A sensing method and apparatus. A sensing node sends first information, the first information being determined by the sensing node according to a first sensing result and a second sensing result corresponding to a first sensing target, the first information being used to instruct a sensing network element to adjust a weight of at least one neuron in a first AI network and / or add at least one detection layer in the first AI network; or the first information includes an adjustment result of the first AI network. Embodiments of the present application can realize adjustment of the first AI network, thereby improving the performance of the first AI network.

Description

Technical Field

[0001] This application relates to the field of communication technology, and in particular to a sensing method and apparatus. Background Technology

[0002] Sensing refers to the detection of parameters of targets in the physical environment, such as their position and velocity. It can also be called detection. The processing flow of sensing signals involves the processing of sensing nodes and sensing network elements. For example, sensing nodes are responsible for sending and / or receiving sensing signals, as well as related baseband processing, ultimately outputting the sensing results to the sensing network elements. The sensing network elements are responsible for processing the sensing results reported by the sensing nodes, such as identifying the sensed targets.

[0003] When processing sensing data, sensing nodes can leverage artificial intelligence (AI) networks. For example, a sensing node can input sensing data into an AI network to obtain information output by the AI ​​network. This information may include parameters of the sensed target, such as its speed and angle. The accuracy of these parameters is determined by the performance of the AI ​​network itself.

[0004] Improving the performance of AI networks is a problem that needs to be solved. Summary of the Invention

[0005] This application provides a sensing method and apparatus to improve the performance of AI networks. The sensing method and apparatus can also be considered as a communication method and apparatus, or an integrated sensing and communication method and apparatus.

[0006] In a first aspect, a first sensing method is provided, which can be applied to a first device. The first device is, for example, a sensing node, a functional module included in a sensing node, or a larger device including a sensing node. The first device is, for example, a network-side device or a terminal-side device. The network-side device is also referred to as a network device. The network device is, for example, a network equipment, or other equipment including network equipment functions, or a circuit, or a chip system (or chip) or other functional module capable of implementing the functions of the network equipment, and the chip system or functional module is, for example, disposed within the network equipment. The network equipment includes, for example, an access network device. The access network device can be a non-ORAN architecture or an ORAN architecture; or, the access network device can be a CU, DU, or RU under an ORAN architecture. The access network device is, for example, located on the ground, or the access network device is, for example, a satellite, or located on a satellite. The terminal-side device is also referred to as a terminal device. The terminal device is, for example, a terminal equipment, or other equipment including terminal equipment functions, or a circuit, or a system-on-a-chip (or a chip, such as a modem chip, also known as a baseband chip, or a system-on-a-chip (SoC) chip or system-in-a-package (SIP) chip containing a modem core) or other functional module, which can realize the functions of the terminal equipment, and which is, for example, disposed in the terminal equipment. The method includes: a sensing node sending first information, the first information being determined by the sensing node based on a first sensing result and a second sensing result corresponding to a first sensing target, the first information being used to instruct a sensing network element to adjust the weights of at least one neuron in a first AI network, and / or, to add at least one detection layer in the first AI network; or, the first information including the adjustment result of the first AI network.

[0007] In this embodiment, the sensing node can send first information to instruct the adjustment of the first AI network, or the sensing node can adjust the first AI network and send the adjustment result to the sensing network element. Therefore, this embodiment can adjust the first AI network, thereby improving its performance. The first information can be determined based on the first and second sensing results. For example, if the sensing node determines, based on the first and second sensing results, that the performance of the first AI network is poor, or that the information output by the first AI network is inaccurate, then the first information can be used to adjust the first AI network, allowing it to be adjusted in a timely manner and improving its performance.

[0008] In an optional implementation, the method further includes: the sensing node obtaining the first sensing result and the second sensing result. For example, if the difference between the first sensing result and the second sensing result is large, the sensing node can send first information to adjust the first AI network, thereby enabling the first AI network to be adjusted in a timely manner.

[0009] In one optional implementation, the sensing node obtains the first sensing result and the second sensing result by: the sensing node receiving a first signal, which is an echo signal reflected, scattered, or diffracted by the first sensing target; the sensing node determining the first sensing result based on the sensing spectrum corresponding to the first signal; and inputting the information of the first signal into the first AI network to obtain the second sensing result output by the first AI network. The sensing node can apply different processing methods to the same signal to obtain multiple sensing results. Since these multiple sensing results correspond to the same signal, the performance of the signal processing method can be reflected through these multiple sensing results, for example, the performance of the first AI network.

[0010] In one optional implementation, the first perception result and / or the second perception result includes one or more of the following information: velocity information of the first perceived target; distance information of the first perceived target; angle information of the first perceived target; contour information of the first perceived target; detection box information of the first perceived target, wherein the detection box information indicates that the first perceived target is included within the detection box; or, point cloud information of the first perceived target. The information included in the first perception result and the second perception result can be completely identical, partially identical, or completely different. If the first perception result and the second perception result include the same information, the value of the same information can be the same or different. For example, both the first perception result and the second perception result can include velocity information of the first perceived target, and the velocity information included in the first perception result and the velocity information included in the second perception result can be the same or different.

[0011] In one optional implementation, both the first sensing result and the second sensing result include N parameters, where N is a positive integer. The first sensing result and the second sensing result satisfy the following conditions: the deviation between the value of the first parameter included in the first sensing result and the value of the first parameter included in the second sensing result is greater than a first threshold, and the first parameter is one of the N parameters; or, the deviation between the values ​​of the N parameters included in the first sensing result and the values ​​of the N parameters included in the second sensing result is greater than N thresholds respectively. For example, if the deviation of the value of any one parameter in the first sensing result and the second sensing result is large, the sensing node can determine or send the first information, ensuring that the first information is sent in a timely manner, and the judgment process of the sensing node is relatively simple. Alternatively, when the values ​​of multiple parameters in the first sensing result and the second sensing result deviate significantly, the sensing node determines or sends the first information, thereby improving the judgment accuracy of the sensing node and reducing the impact of signal fluctuations.

[0012] In one optional implementation, the first information is used to instruct the adjustment of the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network. The first information further includes perceptual data collected in the current environment, which is used by the perceptual network element to adjust the first AI network. If the first AI network is adjusted by the perceptual network element, the perceptual node can send the perceptual data collected in the current environment to the perceptual network element, which can then adjust the first AI network based on this perceptual data. Since the perceptual data is collected in the current environment, the adjusted first AI network can be adapted to the current environmental state, thereby improving the generalization ability of the first AI network.

[0013] In one optional implementation, the first information is used to instruct the adjustment of the weights of at least one neuron in the first AI network, and / or the addition of at least one detection layer in the first AI network. The method further includes: the sensing node receiving the adjustment result of the first AI network, the adjustment result including weight information of at least one neuron in the first AI network, or including weight information of at least one newly added detection layer in the first AI network; the sensing node updating the first AI network according to the adjustment result. If the first AI network is adjusted by a sensing element, the sensing element can send the adjustment result to the sensing node, and the sensing node updates the first AI network accordingly, so that the updated first AI network can be used to perform sensing.

[0014] In one optional implementation, the first information includes the adjustment result of the first AI network, wherein the method further includes: the sensing node adjusting the first AI network based on the deviation between the first sensing result and the second sensing result to obtain the adjustment result. If the first AI network is adjusted by the sensing node, the sensing node can adjust the first AI network based on the deviation, for example, by correcting the first AI network to reduce the deviation, so that the error of the sensing result output by the adjusted first AI network is smaller.

[0015] In one optional implementation, the sensing node adjusts the first AI network based on the deviation between the first sensing result and the second sensing result to obtain the adjustment result. This includes: the sensing node training the first AI network based on the deviation and sensing data collected in the current environment to obtain the adjustment result. If the sensing node adjusts the first AI network, it can do so based on the sensing data collected in the current environment. Since the sensing data is collected in the current environment, the adjusted first AI network can be adapted to the current environmental state, thereby improving the generalization ability of the first AI network.

[0016] Secondly, a second sensing method is provided, which can be applied to a second device. The second device is, for example, a sensing network element, a functional module included in a sensing network element, or a larger device including a sensing network element. The second device is, for example, a network-side device. This network-side device is also referred to as a network device. The network device is, for example, a network equipment, or other equipment including network equipment functions, or a circuit, or a chip system (or chip) or other functional module capable of implementing the functions of the network equipment, and is, for example, disposed within the network equipment. The network equipment includes, for example, core network equipment and / or access network equipment, or includes equipment located between the core network and the access network. The access network equipment can be a non-ORAN architecture or an ORAN architecture; or, the access network equipment can be a CU, DU, or RU under an ORAN architecture. The access network equipment is, for example, located on the ground, or the access network equipment is, for example, a satellite, or located on a satellite. The core network equipment is, for example, a sensing network element or an LMF (Light Filter Function) or other network element capable of implementing sensing functions. The method includes: a sensing network element receiving first information, the first information being determined by the sensing node based on a first sensing result and a second sensing result corresponding to a first sensing target; the first information being used to instruct the sensing network element to adjust the weights of at least one neuron in a first AI network, and / or to add at least one detection layer in the first AI network; or, the first information including the adjustment result of the first AI network.

[0017] In one alternative implementation, the first information is used to instruct the adjustment of the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network, wherein the first information further includes perception data collected in the current environment, the perception data being used to adjust the first AI network.

[0018] In an optional implementation, the method further includes: the sensing network element training the first AI network according to the first information and the sensing data to obtain the adjustment result.

[0019] In one optional implementation, the first information includes the adjustment result of the first AI network, wherein the method further includes: the sensing network element updating the first AI network according to the adjustment result.

[0020] For information on the technical effects of the second aspect or some of the optional implementation methods, please refer to the description of the technical effects of the first aspect or corresponding implementation methods.

[0021] Thirdly, an apparatus is provided. The apparatus can be the first apparatus described in the first aspect above. The apparatus possesses the functions of the first apparatus described above. For example, the apparatus is capable of implementing the functions described in the first aspect above. For instance, the apparatus includes modules, units, or means corresponding to performing the operations involved in the first aspect above. These modules, units, or means can be implemented through software, hardware, or a combination of software and hardware. The first apparatus is, for example, a terminal device, or other device including terminal device functions, or a chip system (or chip or circuit) or other functional module capable of implementing the functions of a terminal device, and is, for example, disposed in a terminal device. Alternatively, the first apparatus is, for example, a network device, or other device including network device functions, or a chip system (or chip or circuit) or other functional module capable of implementing the functions of a network device, and is, for example, disposed in a network device. The network device includes, for example, an access network device. In an optional implementation, the apparatus includes a baseband device and a radio frequency device. In another optional implementation, the device includes a processing unit (sometimes also called a processing module) and a transceiver unit (sometimes also called a transceiver module). The transceiver unit is capable of both sending and receiving functions. When the transceiver unit performs the sending function, it can be called a sending unit (sometimes also called a sending module); when the transceiver unit performs the receiving function, it can be called a receiving unit (sometimes also called a receiving module). The sending unit and the receiving unit can be the same functional module, which is called the transceiver unit and can perform both sending and receiving functions; alternatively, the sending unit and the receiving unit can be different functional modules, and the transceiver unit is a collective term for these functional modules.

[0022] In one optional implementation, the transceiver unit (or the sending unit) is configured to send first information, which is determined by the sensing node based on the first and second sensing results corresponding to the first sensing target. The first information is used to instruct the sensing network element to adjust the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network; or, the first information includes the adjustment results of the first AI network.

[0023] In an alternative embodiment, the device further includes a storage unit (sometimes also called a storage module), the processing unit being coupled to the storage unit and executing programs or instructions in the storage unit to enable the device to perform the functions of the first device described in the first aspect above.

[0024] Fourthly, an apparatus is provided. The apparatus can be the second apparatus described in the second aspect above. The apparatus possesses the functions of the second apparatus described above. For example, the apparatus is capable of implementing the functions described in the second aspect above; for example, the apparatus includes modules, units, or means corresponding to performing the operations involved in the second aspect above. These modules, units, or means can be implemented through software, hardware, or a combination of software and hardware. The second apparatus is, for example, a network device, or other device including network device functions, or a chip system (or chip or circuit) or other functional module capable of implementing the functions of a network device, and is, for example, disposed within a network device. The network device includes, for example, core network equipment and / or access network equipment. In one optional implementation, the apparatus includes a baseband device and a radio frequency device. In another optional implementation, the apparatus includes a processing unit (sometimes also called a processing module) and a transceiver unit (sometimes also called a transceiver module). For details on the implementation of the transceiver unit, please refer to the relevant description in the third aspect.

[0025] In one optional implementation, the transceiver unit (or the receiving unit) is configured to receive first information, which is determined by the sensing node based on a first sensing result and a second sensing result corresponding to a first sensing target. The first information is used to instruct the sensing network element to adjust the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network; or, the first information includes the adjustment result of the first AI network.

[0026] In an alternative embodiment, the device further includes a storage unit (sometimes also called a storage module), the processing unit being coupled to the storage unit and executing programs or instructions in the storage unit to enable the device to perform the functions of the second device described in the second aspect above.

[0027] Fifthly, an apparatus is provided, the apparatus comprising a memory and one or more processors. The memory is used to store part or all of a computer program or instructions necessary for implementing the functions involved in the first or second aspect described above. The one or more processors are executable to carry out the computer program or instructions, such that, when executed, the apparatus implements the methods in any possible design or implementation of the first or second aspect described above.

[0028] In one possible design, the device may further include interface circuitry, wherein the processor is configured to communicate with other devices or components via the interface circuitry.

[0029] In one possible design, the device may also include the memory.

[0030] The aforementioned device may be a network device, a communication module in a network device, or a chip in a network device that is responsible for communication functions, such as a modem chip (also known as a baseband chip) or a SoC or SIP chip that contains a modem module.

[0031] A sixth aspect provides an apparatus comprising a memory and one or more processors. The memory is used to store part or all of a computer program or instructions necessary for implementing the functions described in the first aspect. The one or more processors are executable to carry out the computer program or instructions, such that, when executed, the apparatus implements the methods in any possible design or implementation of the first aspect.

[0032] In one possible design, the device may further include interface circuitry, wherein the processor is configured to communicate with other devices or components via the interface circuitry.

[0033] In one possible design, the device may also include the memory.

[0034] The aforementioned device may be a terminal device, a communication module in a terminal device, or a chip in a terminal device that is responsible for communication functions, such as a modem chip (also known as a baseband chip) or a SoC or SIP chip that contains a modem module.

[0035] A seventh aspect provides a sensing system including a sensing network element. The sensing network element is used to perform the method described in the second aspect, which is executed by the second device. For example, the sensing network element can be implemented using the device described in the fourth or fifth aspect.

[0036] Optionally, the sensing system may further include sensing nodes. These sensing nodes are used to perform the method described in the first aspect, which is executed by the first device. The sensing node is, for example, a terminal device or a network device (e.g., an access network device). For example, the terminal device may be implemented using the device described in the third or sixth aspect; the network device may be implemented using the device described in the third or fifth aspect.

[0037] Eighthly, a computer-readable storage medium is provided for storing a computer program or instructions that, when executed, cause the method performed by the first or second means in the preceding aspects to be implemented.

[0038] Ninthly, a computer program product containing instructions is provided, which, when the computer program or instructions are run on a computer, causes the methods described in the above aspects to be implemented.

[0039] In a tenth aspect, a chip system is provided, including a processor and an interface, the processor being configured to call and execute instructions from the interface to enable the chip system to implement the methods described above. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the access network equipment structure under the ORAN architecture.

[0041] Figure 2 This is a schematic diagram of a RAN chip structure.

[0042] Figure 3 and Figure 4 These are schematic diagrams of single-station sensing mode and dual-station sensing mode, respectively.

[0043] Figure 5 and Figure 6 This is a schematic diagram of two network architectures used in the embodiments of this application;

[0044] Figure 7 A flowchart of a sensing method provided in an embodiment of this application;

[0045] Figure 8 This is an example of a sensing node obtaining a first sensing result in an embodiment of this application;

[0046] Figure 9 This is an example of a network structure for the first AI network in the embodiments of this application;

[0047] Figure 10 This is an example of adjusting the weight of at least one neuron in the first AI network in an embodiment of this application;

[0048] Figure 11 A schematic diagram of an apparatus provided in an embodiment of this application;

[0049] Figure 12 This is a schematic diagram of another device provided in an embodiment of this application. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the embodiments of this application will be further described in detail below with reference to the accompanying drawings.

[0051] In this application embodiment, the number of nouns, unless otherwise specified, refers to "singular nouns or plural nouns," that is, "one or more." "At least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist simultaneously, or B exists alone, where A and B can be singular or plural. The character " / " generally indicates that the related objects before and after are in an "or" relationship. For example, A / B means: A or B. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c means: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0052] The ordinal numbers such as "first" and "second" mentioned in the embodiments of this application are used to distinguish multiple objects, and are not used to limit the size, content, order, timing, priority, or importance of the multiple objects. Furthermore, the numbering of steps in the various embodiments described in this application is only to distinguish different steps and is not used to limit the order in which the steps are performed.

[0053] The following explanations of some terms or concepts used in the embodiments of this application are provided to facilitate understanding by those skilled in the art.

[0054] In this embodiment of the application, the terminal device is a device with wireless transceiver function, which may be a fixed device, a mobile device, a handheld device (e.g., a mobile phone), a wearable device, an in-vehicle device, or a wireless device (e.g., a communication module, a modem, or a chip system, etc.) built into the above devices. The terminal devices are used to connect people, objects, and machines, and can be widely used in various scenarios, including but not limited to the following: sensing scenarios, cellular communication, device-to-device (D2D) communication, vehicle-to-everything (V2X) communication, machine-to-machine / machine-type (M2M / MTC) communication, Internet of Things (IoT), virtual reality (VR), augmented reality (AR), industrial control, self-driving, remote medical care, smart grid, smart furniture, smart office, smart wearables, smart transportation, smart city, drones, robots, and terminal devices in indoor commercial scenarios (such as mobile phone screen mirroring, file sharing, and mobile phone to VR glasses). When the terminal equipment is applied to V2X, it can also be called a V2X device, such as a smart car, digital car, unmanned car, driverless car, pilotless car, or automobile, self-driving car, or autonomous car, pure electric vehicle (EV), hybrid electric vehicle (HEV), range-extended electric vehicle (REEV), plug-in hybrid electric vehicle (PHEV), new energy vehicle, or roadside unit (RSU). The terminal equipment can also be a device used in D2D communication, such as an electricity meter or water meter.

[0055] Furthermore, in this embodiment of the application, the terminal device can also be a terminal device in an Internet of Things (IoT) system. IoT is an important component of the future development of information technology. Its main technical feature is to connect objects to the network through communication technology, thereby realizing an intelligent network of human-machine interconnection and object-to-object interconnection.

[0056] The various terminal devices described above, if located in a vehicle (e.g., placed inside or installed inside a vehicle), can all be considered in-vehicle terminal devices, also known as on-board units (OBUs). The terminal device of this application can also be an in-vehicle module, in-vehicle component, in-vehicle chip, or in-vehicle unit built into a vehicle as one or more components or units. The vehicle can implement the methods of this application through the built-in in-vehicle module, in-vehicle component, in-vehicle chip, or in-vehicle unit.

[0057] The terminal equipment may sometimes be referred to as user equipment (UE), terminal, access station, UE station, remote station, wireless communication equipment, or user device, etc.

[0058] In this application embodiment, the device for implementing the terminal device function can be a terminal device, which can be a terminal device or a device capable of supporting the terminal device in implementing the function, such as a chip system. This device can be installed in the terminal device. In the technical solutions provided in this application embodiment, the example of a terminal device being used to implement the terminal device function is used to describe the technical solutions provided in this application embodiment.

[0059] The network devices in this application embodiment include, for example, access network devices (or access network elements) and / or core network devices (or core network elements). The access network devices are devices with wireless transceiver capabilities, used to communicate with the terminal devices. The access network devices include, but are not limited to, base stations (base transceiver stations (BTS), Node B, evolved Node B (eNodeB) / eNB, or the next generation Node B (gNodeB) / gNB), transmission reception points (TRPs), base stations evolved from the 3rd generation partnership project (3GPP), access nodes in wireless fidelity (Wi-Fi) systems, wireless relay nodes, wireless backhaul nodes, etc. The base stations can be: macro base stations, micro base stations, pico base stations, small cells, relay stations, etc. Multiple base stations can support networks using the same access technology or networks using different access technologies. A base station can contain one or more co-located or non-co-located transmission and reception points. The access network equipment can also be a radio controller, centralized unit (CU), and / or distributed unit (DU) in a cloud radioaccess network (CRAN) scenario. The access network equipment can also be a server, etc. For example, the network equipment in V2X technology can be a roadside unit (RSU). The following description uses a base station as an example to illustrate the access network equipment. A base station can communicate with a terminal device, or it can communicate with a terminal device through a relay station. A terminal device can communicate with multiple base stations in different access technologies. The core network equipment is used to implement functions such as mobility management, data processing, session management, policy and billing. The names of the equipment implementing core network functions may differ in systems using different access technologies; this application does not limit this.Taking the 5th generation (5G) mobile communication technology system as an example, the core network equipment includes, for example, access and mobility management function (AMF), session management function (SMF), policy control function (PCF), or user plane function (UPF), etc.

[0060] In a CU-DU architecture, or in an open RAN (ORAN) system, access network equipment may include one or more logical network elements such as CU, DU, CU-control plane (CP), CU-user plane (UP), or radio unit (RU). CU and DU can be separate entities or included in the same network element, such as a baseband unit (BBU). RU may be included in radio equipment or radio units, such as a remote radio unit (RRU), active antenna unit (AAU), or remote radio head (RRH). One possible structure for access network equipment can be found in [reference needed]. Figure 1 Among them, core network equipment and access network equipment can communicate through backhaul links; within access network equipment, CU and DU can communicate through midhaul links, and DU and RU can communicate through fronthaul links.

[0061] Alternatively, another architecture for the access network equipment can be referenced. Figure 2 , Figure 2Taking access network equipment implemented through chips as an example, such as a RAN chip, the RAN chip may include a CU, DU, and RU. The CU can perform L2 and L3 functions, etc.; the DU can perform L1 functions and some L2 functions, etc.; and the RU can perform L1 computing and radio frequency (RF) digital functions, etc. The CU communicates with the core network equipment through a backhaul interface, which carries the traffic between the CU and the core network equipment. The CU may include a central processing unit (CPU) based on x86 or ARM architecture, as well as field programmable gate arrays (FPGAs), graphics processing units (GPUs), or other accelerators, etc. The CPU and the FPGA, GPU, or other accelerators can communicate through a peripheral component interconnect express (PCIe) interface.

[0062] The CU and DU communicate via a midhaul interface, which carries the traffic between the CU and DU. The DU may include an x86 or ARM architecture CPU, as well as FPGAs, GPUs, or other accelerators, which can communicate with the FPGA, GPU, or other accelerators via a PCIe interface.

[0063] The DU and RU communicate via a fronthaul interface, which carries the traffic between the DU and RU. If the access network equipment uses an integrated DU, the integrated DU can include the functions of both the DU and RU, and the RAN may no longer need to include a separate RU. The RU may include a RAN fronthaul processing unit, a digital processing unit, and an RF processing unit. The RAN fronthaul processing unit is implemented, for example, using an FPGA or an application-specific integrated circuit (ASIC). The digital processing unit is implemented, for example, using an FPGA or an ASIC.

[0064] The RU can be connected to an antenna to communicate with the UE via the antenna.

[0065] In different systems, CU (or CU-CP and CU-UP), DU, or RU may have different names, but those skilled in the art will understand their meaning. For example, in an ORAN system, CU can also be called an open CU (O-CU), DU can also be called an open DU (O-DU), CU-CP can also be called an open CU-CP (O-CU-CP), CU-UP can also be called an open CU-UP (O-CU-CP), and RU can also be called an open RU (O-RU). For ease of description, this application uses CU, CU-CP, CU-UP, DU, and RU as examples in its embodiments. Any of the units among CU (or CU-CP, CU-UP), DU, and RU in the embodiments of this application can be implemented through software modules, hardware modules, or a combination of software modules and hardware modules.

[0066] The CU and DU can be configured according to the protocol layer functions of the wireless network they implement. For example, the CU can be configured to implement the functions of the Packet Data Convergence Protocol (PDCP) layer and above (such as the Radio Resource Control (RRC) layer and / or the Service Data Adaptation Protocol (SDAP) layer); the DU can be configured to implement the functions of protocol layers below the PDCP layer (such as one or more of the Radio Link Control (RLC) layer, Media Access Control (MAC) layer, or Physical (PHY) layer). Alternatively, the CU can be configured to implement the functions of protocol layers above the PDCP layer (such as the RRC and / or SDAP layers), and the DU can be configured to implement the functions of protocol layers below the PDCP layer (such as one or more of the RLC, MAC, or PHY layers).

[0067] The above CU and DU configurations are merely examples; the functions of the CU and DU can be configured as needed. For instance, the CU or DU can be configured to have more protocol layer functions, or only some protocol layer processing functions. For example, some RLC layer functions and protocol layer functions above the RLC layer can be placed in the CU, while the remaining RLC layer functions and protocol layer functions below the RLC layer can be placed in the DU. Furthermore, the functions of the CU or DU can be divided according to service type or other system requirements, such as by latency. Functions that require low latency can be placed in the DU, while functions that do not require low latency can be placed in the CU.

[0068] DU and RU can cooperate to implement the functions of the PHY layer. A DU can be connected to one or more RUs. The functions of DU and RU can be configured in various ways depending on the design. For example, a DU can be configured to implement baseband functions, and an RU can be configured to implement mid-RF functions. Another example is that a DU can be configured to implement higher-level functions in the PHY layer, and an RU can be configured to implement lower-level functions in the PHY layer, or to implement both lower-level and RF functions. Higher-level functions in the physical layer can include a portion of the physical layer's functions that are closer to the MAC layer, while lower-level functions in the physical layer can include another portion of the physical layer's functions that are closer to the mid-RF side.

[0069] In this application embodiment, the apparatus for implementing the functions of a network device can be referred to as a network apparatus. This network apparatus can be a network element, a network device, or an apparatus capable of supporting the network device or network element in implementing the function, such as a chip system. This apparatus can be installed within the network device. In the technical solutions provided in this application embodiment, the apparatus for implementing the functions of a network device is described as a network apparatus (for example, an apparatus for implementing the functions of an access network apparatus is an access network apparatus, and an apparatus for implementing the functions of a core network apparatus is a core network apparatus).

[0070] A sensing signal is a signal used to sense (or detect) a target (or object). Sensing signals can also be called detection signals, linear frequency modulated signals, radar signals, radar sensing signals, radar detection signals, or environmental sensing signals, etc. Sensing signals can be pulse signals or signals from wireless communication systems. For example, a sensing signal can be an orthogonal frequency division multiplexing (OFDM) signal obtained by modulating a specific sequence on a subcarrier. This specific sequence can be any of the following sequences: Zadoff-Chu sequence (ZC sequence), pseudo-random sequence, or predefined sequence. Pseudo-random sequences include any of the following sequences: longest linear feedback shift register sequence (m-sequence) or Gold sequence. Predefined sequences can be, for example, random data symbols, such as random data symbols modulated by quadrature phase shift keying (QPSK) or quadrature amplitude modulation (QAM).

[0071] An echo signal is a signal generated by the reflection, scattering, or diffraction of a sensing signal by a target. Both the echo signal and the sensing signal can reflect target parameters. For example, the time delay of the echo signal relative to the sensing signal can reflect the distance of the target relative to the transmitter, and the Doppler shift of the echo signal relative to the sensing signal can reflect the target's velocity.

[0072] Communication-sensing fusion signals, also known as synthetic-sensing fusion signals, synthetic signals, or integrated synthetic-sensing signals, are signals used for both communication and sensing. When used for communication, the fusion signal carries the communication data or reference signal sequence that needs to be transmitted between communication devices. When used for sensing, the fusion signal can be understood as being used to sense (or detect) targets.

[0073] For example, the signals used for sensing described herein may include sensing signals and / or synesthetic fusion signals, etc.

[0074] The target can be any tangible object in the environment that can reflect electromagnetic waves, such as mountains, forests, or buildings, and can also include mobile objects such as vehicles, drones, pedestrians, and terminal devices. The target can also be referred to as a sensing target, detection target, perceived target, detected target, perceived object, detected object, or sensed device, etc., and the embodiments of this application do not limit it.

[0075] For echo signals from multiple carriers, antennas, and symbols, a sensing spectrum can be generated by applying techniques such as joint Fourier transform (FT) in the time, frequency, and spatial domains or beamforming to convert the echo signal into a sensing spectrum in the beam domain. This sensing spectrum contains filtered signals from all velocity / delay / angle domain sampling points (which can be converted to Cartesian coordinates) within a specified sensing area. In other words, this sensing spectrum can represent the scattered echo signal of each target point at its corresponding velocity / delay / angle position.

[0076] A neural network is an artificial intelligence method that consists of interconnected nodes or artificial neurons arranged in a hierarchical structure and transmitting and processing data through weighted connections.

[0077] Generalization ability is a key metric for evaluating the performance of neural networks. A neural network with good generalization ability can adapt to data changes in various real-world scenarios, thus playing a reliable role in practical applications. For example, in image recognition, a well-trained model should not only be able to accurately identify images in the training set, but also correctly classify newly captured images from different angles and under different lighting conditions.

[0078] For a long time, wireless sensing has been an independently developed technology. Sensing services are provided by various specialized sensing devices, such as conventional radar, lidar, computed tomography (CT), and magnetic resonance imaging (MRI). In 5G and earlier communication systems, positioning was the sensing service that mobile communication systems could provide. In future mobile communication systems, general sensing services other than positioning will be integrated into the communication system, becoming a completely new function, thereby opening up entirely new services, such as high-precision positioning, environmental reconstruction, and gesture and motion recognition.

[0079] Among them, sensing network elements can reconstruct the environment of a target area through means such as lasers, radars, or base stations. For example, sensing network elements can reconstruct the real physical environment based on the measurement results reported by devices such as lasers, radars, or base stations. For example, based on the measurement results, sensing network elements can reconstruct environmental information using methods such as scattering polygons to characterize various scattering objects in the environment, such as walls and furniture (also known as sensing targets, targets, or target objects).

[0080] In sensing, based on the different sender and receiver of the sensing signal, sensing modes can be divided into two types: single-station sensing and dual-station sensing. Single-station sensing mode is also called self-transmitting and self-receiving mode, single-end sensing mode, or monocentric sensing mode, etc. This means that the device transmitting the sensing signal and the device receiving the echo signal reflected from the target are the same device, such as... Figure 3As shown, both the device transmitting the sensing signal and the device receiving the echo signal are device 1; the dual-station sensing mode, also known as A-transmit B-receive mode or self-transmit and other-receive mode, refers to a mode where the device transmitting the sensing signal and the device receiving the echo signal reflected from the target are different devices, such as... Figure 4 As shown, the device that sends the sensing signal is device 2, and the device that receives the echo signal is device 3. Figure 3 and Figure 4 The example used is a vehicle. Figure 3 In this context, device 1 is a base station or UE. In single-site sensing mode, device 1 transmits a sensing signal, and device 1 receives the sensing signal through sensing targets in the environment (e.g., Figure 3 The echo signals generated by reflection, scattering, or diffraction of vehicles (in the environment) are used for environmental perception. For example, Figure 4 In this dual-site sensing mode, device 2 is a base station and device 3 is a UE, or device 2 is a base station and device 3 is a UE, or device 2 and device 3 are two different UEs, or device 2 and device 3 are two different base stations, etc. In this dual-site sensing mode, device 1 sends a sensing signal, and device 2 receives the sensing signal through scattering objects in the environment (e.g., [missing information]). Figure 4 The system uses echo signals generated by reflection, scattering, or diffraction of vehicles (in the environment) to perform environmental perception. In the dual-site perception mode, if device 1 is a UE and device 2 is a network device, or device 1 is a network device and device 2 is a UE, then this dual-site perception mode can also be called the site-side joint perception mode.

[0081] For reference Figure 5 This is a schematic diagram of a potential perceptual network architecture. Figure 5 It is based on the 5G core network (5G core, 5GC). Figure 5 The network architecture shown can also be an application scenario of the embodiments of this application.

[0082] exist Figure 5 The architecture shown includes a new sensing function (SF) network element, also known simply as a sensing network element. This SF can be a device or component that provides sensing functionality to the network; it can also be called a sensing management function (SMF), or have other names. This SF can be deployed on the core network side or the RAN side. Figure 5 Taking deployment in the core network as an example. Figure 5In the network architecture shown, the SF can reuse the interfaces between the location management function (LMF) and other 5GC network elements such as AMF, network exposure function (NEF), unified data management (UDM), network data analytics function (NWDAF), and PCF for sensing interaction. Sensing signaling between the SF and the radio access network (RAN) or UE can be transmitted through the AMF; sensing measurement data acquired by the RAN or UE can be transmitted to the SF via the control plane, for example, by reusing the Long Term Evolution (LTE) positioning protocol (LPP) or the New Radio (NR) Positioning Protocol Annex (NRPPa) protocol, or it can be transmitted via the user plane, forwarded to the SF through the UPF, or directly transmitted to the SF.

[0083] The newly added SF in this network architecture can realize basic sensing functions, such as sensing authorization, sensing control, sensing measurement data processing, or result output. Specifically, interfaces are set up and interaction is established between the SF and 5GC network elements such as AMF, NEF, UDM, NWDAF, PCF, LMF, and UPF, as defined below.

[0084] NS1: A new interface between SF and AMF, which can transmit sensing and control signaling. Additionally, this interface can also transmit sensing measurement data in scenarios where sensing measurement data is uploaded to the control plane.

[0085] NS2: A new interface between SF and NEF. This interface can transmit signaling messages between sensing network elements relayed through NEF and application functions (AF) on the service side, and at the same time open the sensing results to the AF.

[0086] NS3: A new interface between SF and UDM. This interface can be used for authentication or authorization, and to obtain UE-aware subscription information, service AMF information, or other information.

[0087] NS4: A new interface between SF and NWDAF. Through this interface, SF and NWDAF can jointly complete artificial intelligence (AI) processing related to perception services.

[0088] NS5: A new interface between SF and PCF. Through this interface, SF can transmit information such as sensing requirements, quality of service (QoS) requirements, or sensing results of sensing services to PCF. PCF can then make decisions to generate policy control and charging (PCC) policies related to sensing services.

[0089] NS6: A new interface between SF and LMF. Through this interface, SF can obtain location-related information, such as the sensing area, the RAN information of the sensing target, and the location information of the sensed UE.

[0090] NS7: A new interface between SF and UPF. Sensing measurement data can be directly transmitted from (R)AN to SF via UPF, or indirectly forwarded to SF via UPF. In scenarios where (R)AN performs sensing, forwarding via UPF can improve the functionality of UPF to support data transmission at the (R)AN granularity.

[0091] In addition to the newly added interfaces mentioned above, existing interfaces (such as N1, N2, N5, N8, N33, etc.) can also support the transmission of information related to sensing services, such as authentication information, sensing service type, sensing service quality requirements, sensing measurement data, or sensing results, etc.

[0092] Figure 5 Taking the SF (Side Array) as an independent device as an example; alternatively, the SF and LMF (Location Management Array) can be co-located, meaning the network element used for handling sensing services and the network element used for handling positioning services can be the same network element; or the SF can be co-located with other core network elements, such as the AMF (Location Management Array). The LMF is the core network element in 5GC that provides control plane positioning, capable of calculating and feeding back location information in the 5G network, providing functions such as positioning process management, UE capability acquisition, auxiliary data provision, and UE location estimation. Optionally, if the SF and LMF are co-located, the LMF and the gateway mobile location center (GMLC) can be functionally enhanced to support basic sensing functions. The GMLC can be the first network element within the operator's network to process sensing requests, performing privacy checks or authorization functions, routing sensing requests to the AMF, or performing LMF selection, etc.

[0093] For example, if the SF and LMF are co-located, an additional interface can be added between the LMF and GMLC to transmit information related to awareness services, such as adding an NL9 interface. Additionally, interfaces related to the LMF and GMLC (such as one or more of the following: NL1 interface between AMF and LMF, NL2 interface between AMF and GMLC, NL5 interface between NEF and GMLC, or NL6 interface between UDM and GMLC) can also support the transmission of information related to awareness services, as detailed below.

[0094] N33: The interface between AF and NEF, through which information such as the type of sensing business, business requirements, and sensing results can be transmitted.

[0095] NL5: The interface between NEF and GMLC, through which information such as the type of sensing business, business requirements, and sensing results can be transmitted.

[0096] NL6: The interface between GMLC and UDM, through which privacy inspection data can be transmitted.

[0097] NL2: The interface between NEF and AMF, through which information such as the perceived business type, business requirements, and perceived results can be transmitted.

[0098] NL1: The interface between AMF and LMF, through which information such as perceived business type, business requirements, and perceived results can be transmitted.

[0099] NL9: A new interface between GMLC and LMF, through which information such as the perceived service type, service requirements, and perceived results can be transmitted.

[0100] Can be referenced again Figure 6 This is a schematic diagram of another potential perceptual network architecture. Figure 6 It is based on 5GC. Figure 6 The network architecture shown can also be another application scenario of the embodiments of this application.

[0101] exist Figure 6In the network architecture shown, the SF (Sensitive Detection) is relatively independent of the existing core network elements. The SF requires little or no interaction with the core network elements. For scenarios where sensing needs exist only in a specific area, or where sensing is the only requirement, this network architecture can provide sensing services without 5GC control or with only a few network elements involved in control. Furthermore, localized deployment of the SF ensures that sensing measurement data or results do not leave the campus, thus meeting enterprises' security and privacy requirements for sensing measurement data or results, and reducing sensing latency. This network architecture is relatively simple, flexible, efficient, has few transmission nodes, and is easy to deploy. Optionally, this network architecture can support UE-related sensing needs, and implementation schemes for functions such as authorization, mobility management, and billing can be considered as needed.

[0102] In this network architecture, the SF can directly establish a connection with the RAN node. Control plane sensing signaling and user plane sensing measurement data can be transmitted via the newly defined interface NS1. When the UE participates in sensing, control plane signaling can be forwarded to the SF via the AMF, and sensing measurement data can be transmitted via NS1. Furthermore, there can also be an interface between the SF and 5GC network elements (such as AMF, NEF, or NWDAF) to control the AF to provide sensing service requirements to the SF through core network functions. The interface between the SF and 5GC network elements is described below.

[0103] NS1: A new interface between the SF and (R)AN, which can transmit sensing control signaling or sensing measurement data. In one implementation, the SF can also be deployed on the RAN side; for example, the SF can be co-located with access network equipment (e.g., a base station), or the SF can be a standalone device within the access network.

[0104] NS2: A new interface that may be added between SF and AMF. This interface can receive awareness service requirements from UE, or transmit signaling between SF and other network elements in the core network, such as transmitting interaction messages between SF and UDM.

[0105] NS3: A potential new interface between SF and NEF. This interface can transmit signaling between SF and the service-side AF via NEF, and can also expose the sensing results to the AF. The interaction between SF and AF may not go through NEF. In actual deployment, NS2 and NS3 may be chosen as one of the two options. That is, the AF can send sensing service requests indirectly to SF or directly to SF (without NEF) via NS2 (NEF); or, the AF can send sensing service requests to SF via N33 (NEF) and NS2 (AMF).

[0106] NS4: A potential new interface between SF and NWDAF, through which SF and NWDAF can jointly perform intelligent analysis and prediction to generate perception results.

[0107] Figure 5 or Figure 6 These examples all use the network including SF as an example. Alternatively, the network may not include SF, but rather other network elements implement the sensing-related functions, such as AMF and / or LMF.

[0108] The technical solutions provided in this application can be applied to fourth-generation (4G) mobile communication systems, such as LTE systems, or to 5G systems, such as NR systems, or to next-generation mobile communication systems or other similar communication systems, such as future communication systems, or to existing satellite mobile communication technology systems; no specific limitations are imposed. For example... Figure 5 and Figure 6 All are based on 5GC. In addition, SF can also be deployed in other networks, such as 6G networks or other future communication networks.

[0109] The embodiments of this application can be applied to Figure 3 , Figure 4 , Figure 5 or Figure 6 The scenario shown can also be used in other scenarios, such as any scenario involving sensing services.

[0110] The method provided in the embodiments of this application is described below with reference to the accompanying drawings. In various embodiments of this application, the signal used for sensing is transmitted through reflection, scattering, transmission, or diffraction. The sensing node (e.g., a network device or UE) can determine the relevant characteristics of the sensing target based on the received signal. For example, it can estimate the time delay, Doppler, or angular spectrum information based on the received signal to determine the distance, angle, or velocity of the sensing target. In addition, the sensing node can also send information, such as first information, to the sensing network element (e.g., SF). In the accompanying drawings corresponding to the various embodiments of this application, all steps indicated by dashed lines are optional steps.

[0111] The various embodiments described herein can be applied to Figure 3 , Figure 4 , Figure 5 or Figure 6 The network architecture shown. For example, the sensing node described in the various embodiments of this document can be... Figure 3 The device 1 mentioned above may be, or may be Figure 4 The device shown as 3, or it could be Figure 5 or Figure 6 The (R)AN shown, or could be Figure 5 or Figure 6 The UE shown. The sensing network elements described in the various embodiments of this document can be... Figure 5 or Figure 6The SF shown.

[0112] This application provides a sensing method, please refer to the embodiments therein. Figure 7 Here is a flowchart of the method.

[0113] S701, the sensing node obtains the first sensing result and the second sensing result.

[0114] Taking a sensing node as an example where the sensing node is an access network device and the access network device is not an ORAN architecture, S701 can be executed by the access network device as a whole. Alternatively, taking a sensing node as an example where the access network device is an access network device and the access network device is an ORAN architecture, S701 can be executed by the CU or DU included in the access network device.

[0115] Both the first and second sensing results correspond to the first sensing target. Alternatively, it can be understood that the sensing node can obtain two sensing results for the first sensing target. Optionally, the first and second sensing results can be obtained based on the same sensing data. For example, the sensing node performs a sensing process (e.g., using a single-station sensing mode or a dual-station sensing mode), obtains sensing data, and the sensing node can obtain two sensing results based on this data, namely the first sensing result and the second sensing result. Optionally, the sensing data may include one or more of the following: a first signal received by the sensing node, a sensing spectrum, or point cloud information. The first signal is, for example, an echo signal, which can be a signal used for sensing that has been reflected, scattered, transmitted, or diffracted by the first sensing target and arrives at the sensing node. The sensing spectrum can be obtained by the sensing node processing the echo signal; for example, the sensing spectrum may include a power delay profile (PDP), or other sensing spectra. The point cloud information can be obtained by the sensing node processing the echo signal; for example, the sensing node processes the echo signal to obtain a sensing spectrum, and then obtains the point cloud information based on the sensing spectrum. As an optional implementation method for a sensing node to obtain point cloud information, the sensing node can determine the information of one or more points based on the sensing spectrum. If the information of a certain point indicates that the energy corresponding to that point is greater than a threshold, the sensing node can add the information of that point to the point cloud information; otherwise, the sensing node will not add the information of that point to the point cloud information, thereby obtaining the point cloud information. The energy corresponding to a point is characterized, for example, by the signal-to-noise ratio (SNR) or other parameters of that point.

[0116] The first sensing result is determined by the sensing node based on the sensing spectrum corresponding to the first signal. For example, if a sensing node receives a first signal, it can determine the first sensing result based on the sensing spectrum corresponding to the first signal. (See reference...) Figure 8This is one example of how a sensing node obtains its first sensing result. For instance, a sensing node can acquire the multidimensional sensing spectrum corresponding to the first signal based on the inverse discrete fourier transform (IDFT) in the frequency domain, the discrete fourier transform (DFT) in the slow time domain, and the two-dimensional (2D) spatial domain DFT. Then, it can apply techniques such as constant false alarm rate (CFAR) detection and peak search to determine the position of the target parameter on this sensing spectrum, thereby estimating the first sensing result.

[0117] The second perception result is determined by the perception node based on the first signal and the first AI network. For example, the perception node receives the first signal and can input the information of the first signal into the first AI network to obtain the perception result output by the first AI network, which is the second perception result. The information of the first signal may include one or more of the following: the first signal itself, the perception spectrum corresponding to the first signal, or the point cloud information corresponding to the first signal.

[0118] This can be understood as the sensing nodes processing the same sensing data in two different ways, thus obtaining two sets of sensing results.

[0119] The first AI network can be an AI network deployed at a sensing node. Optionally, the first AI network can be, for example, an AI model. For instance, a sensing network element can train the first AI network and send its information to the sensing node, thereby enabling the sensing node to deploy the first AI network. Alternatively, a third-party network element can train the first AI network and send its information to the sensing node, thus enabling the sensing node to deploy the first AI network.

[0120] Regardless of which network element trains the first AI network, optionally, the sensing network element can also maintain the first AI network. For example, if the sensing network element trains the first AI network, it can not only send information about the first AI network to the sensing node, but also deploy the first AI network itself. Similarly, if a third-party network element trains the first AI network, it can not only send information about the first AI network to the sensing node, but also send information about the first AI network to the sensing network element, enabling the sensing network element to deploy the first AI network as well. Optionally, the first AI network deployed on the sensing network element side and the first AI network deployed on the sensing node side can be the same or synchronous. For example, the structures of the first AI networks deployed on both sides can be completely identical. Therefore, the "first AI network" in this embodiment can refer to either the first AI network deployed on the sensing node side or the first AI network deployed on the sensing network element side. Alternatively, it can be understood that the sensing network element and the sensing node in this embodiment deploy the same AI network.

[0121] Alternatively, for distinction, the first AI network deployed by the sensing nodes can be called AI network A, and the first AI network deployed by the sensing elements can be called AI network B. However, AI network A and AI network B have completely identical structures, and the weights of corresponding neurons in AI network A and AI network B are exactly the same. For example, AI network A and AI network B have identical structures, both AI network A and AI network B include neuron 'a', and the weights of neuron 'a' in AI network A are the same as the weights of neuron 'a' in AI network B.

[0122] The following text will mainly use the example of referring to both sides of the AI ​​network as the first AI network.

[0123] This application does not limit the structure of the first AI network. For example, the first AI network may include one or more of the following network structures: fully connected network, convolutional network, residual network, Transformer network, hybrid attention mechanism network, or reinforcement learning network. See also... Figure 9 This is an example of a first-generation AI network. Figure 9 Taking the first AI network, which includes a fully connected module and a hybrid attention mechanism module, as an example, the actual structure of the first AI network is not limited to this. For example... Figure 9 In the first AI network, the input component is a hybrid attention mechanism module, and the output component is a fully connected module. Perceptual data passes through the hybrid attention mechanism module (…). Figure 9 In the example of the hybrid attention mechanism module (which includes a channel attention layer and a spatial attention layer), the data features are obtained after processing (for example, the data features are implemented in the form of a feature spectrum). The data features are then processed by the fully connected module to finally output the second perception result. Figure 9Taking the hybrid attention mechanism module as an example, which includes a channel attention layer and a spatial attention layer, the actual number is not limited to this; Figure 9 Taking the fully connected module as an example, which includes two neural layers, the actual number is not limited to this.

[0124] Optionally, the first perception result may include one or more of the following information: velocity information of the first perceived target, distance information of the first perceived target, angle information of the first perceived target, contour information of the first perceived target, detection box information of the first perceived target, or point cloud information of the first perceived target. The detection box information indicates that the detection box may include the first perceived target.

[0125] Optionally, the second perception result may include one or more of the following information: velocity information of the first perceived target, distance information of the first perceived target, angle information of the first perceived target, contour information of the first perceived target, detection box information of the first perceived target, or point cloud information of the first perceived target.

[0126] The parameters included in the first perception result and the second perception result can be completely identical, partially identical, or completely different. For example, the first perception result includes the velocity and angle information of the first perceived target, and the second perception result also includes the velocity and angle information of the first perceived target; that is, the parameters included in the first and second perception results are completely identical. As another example, the first perception result includes the velocity and distance information of the first perceived target, and the second perception result includes the velocity and contour information of the first perceived target; that is, the parameters included in the first and second perception results are partially identical. Yet another example, the first perception result includes the velocity information of the first perceived target, and the second perception result includes the angle and contour information of the first perceived target; that is, the parameters included in the first and second perception results are completely different.

[0127] If the first perception result and the second perception result include the same parameter, then the value of that parameter included in the first perception result and the value of that parameter included in the second perception result may be the same or different. For example, if both the first perception result and the second perception result include the speed information of the first perceived target, the value of that speed information included in the first perception result and the value of that speed information included in the second perception result may be the same or different.

[0128] Optionally, the sensing node may also send the first information without performing S701, therefore S701 is an optional step.

[0129] S702, the sensing node sends the first information. Correspondingly, the sensing network element receives the first information.

[0130] Taking a sensing node as an example where the sensing node is an access network device and the access network device is not an ORAN architecture, S702 can be executed by the access network device as a whole. Alternatively, taking a sensing node as an example where the access network device is an access network device and the access network device is an ORAN architecture, S702 can be executed by the RU or DU included in the access network device.

[0131] Optionally, the sensing node can determine the first information based on the first sensing result and the second sensing result, or send the first information. Alternatively, even without the first sensing result and / or the second sensing result, the sensing node can still determine the first information or send the first information; there is no limitation on this.

[0132] Take, for example, a sensing node determining (or sending) first information based on a first sensing result and a second sensing result. For instance, the first sensing result includes N parameters, and the second sensing result also includes these N parameters. Alternatively, it can be understood that the parameters included in the first and second sensing results are partially the same or completely identical. These N parameters are the same parameters included in both the first and second sensing results, where N is a positive integer. For example, if the first and second sensing results satisfy one or more of the following relationships, the sensing node can determine or send the first information: the deviation between the value of the first parameter included in the first sensing result and the value of the first parameter included in the second sensing result is greater than a first threshold, and the first parameter is one of the N parameters; within a first time period, the deviation between the value of the first parameter included in the first sensing result and the value of the first parameter included in the second sensing result is greater than the first threshold; the deviation between the values ​​of the N parameters included in the first sensing result and the values ​​of the N parameters included in the second sensing result is greater than N thresholds respectively; within a second time period, the deviation between the values ​​of the N parameters included in the first sensing result and the values ​​of the N parameters included in the second sensing result is greater than N thresholds respectively.

[0133] The deviation between the value of the first parameter in the first perception result and the value of the first parameter in the second perception result is greater than a first threshold. This can be understood as follows: as long as the deviation between the value of the same parameter in the first and second perception results is greater than the first threshold, the perception node can determine or send the first information. For example, the first perception result includes the velocity, distance, and angle information of the first perceived target, and the second perception result includes the velocity, angle, contour, and point cloud information of the first perceived target. Then, the N parameters can include the velocity and angle information of the first perceived target. If the deviation between the velocity information value in the first and second perception results is greater than the first threshold, and / or the deviation between the angle information value in the first and second perception results is greater than the first threshold, then the perception node can determine or send the first information. The first threshold can be different for different parameters. For example, the first threshold for velocity information can be a velocity threshold, and the first threshold for angle information can be an angle threshold. In this approach, the perception node only needs to determine the value of one parameter to determine or send the first information, which is relatively simple.

[0134] The deviations between the values ​​of the N parameters included in the first perception result and the values ​​of the N parameters included in the second perception result are each greater than N thresholds. This can also be understood as follows: if the deviation between the values ​​of all identical parameters included in the first and second perception results is greater than the corresponding threshold, the perception node can determine or send the first information; otherwise, the perception node may not determine or send the first information. For example, the first perception result includes the velocity, distance, and angle information of the first perceived target, and the second perception result includes the velocity, angle, contour, and point cloud information of the first perceived target. The N parameters can include the velocity and angle information of the first perceived target. If the deviation between the velocity information value included in the first and second perception results is greater than a velocity threshold, and the deviation between the angle information value included in the first and second perception results is greater than an angle threshold, the perception node can determine or send the first information. The deviation of a parameter value can be compared with the threshold corresponding to that parameter. Different parameters can correspond to different thresholds; for example, velocity information can correspond to a velocity threshold, distance information to a distance threshold, and angle information to an angle threshold, etc.

[0135] Alternatively, the sensing node can also combine time into its judgment, for example, determining whether the deviation between parameter values ​​within a certain period (a first duration or a second duration) exceeds a corresponding threshold, thereby determining or sending the first information. Within this period, the sensing node can acquire the first and second sensing results once or multiple times. The echo signals corresponding to different first and second sensing results can be different. For example, if the sensing node continuously performs the sensing process within this period, it can acquire echo signals multiple times and further obtain the first and second sensing results. The first sensing results and the second sensing results acquired in different instances can be the same or different. By combining the parameter values ​​within a certain period, the sensing node can reduce the impact of parameter value fluctuations, making the judgment more accurate.

[0136] If the deviation between the values ​​of the same parameter in the first and second perception results is large, it indicates a decrease in the accuracy of the first AI network or insufficient generalization ability of the first AI network. Therefore, in this case, the perception node can determine or send first information to adjust the first AI network, thereby improving its accuracy and generalization ability.

[0137] The first piece of information can indicate different content, as illustrated in the following examples.

[0138] As an optional implementation of the first information, the first information may instruct the sensing network element to adjust the weights of at least one neuron in the first AI network, and / or instruct the addition of at least one detection layer in the first AI network. It can also be understood that the first AI network can have multiple adjustment methods, and the first information may indicate the specific adjustment method. For example, the adjustment methods of the first AI network may include adjusting the weights of at least one neuron in the first AI network, and / or including adjusting the addition of at least one detection layer in the first AI network. In this implementation, the sensing node decides the adjustment method, and the sensing network element adjusts accordingly, without the sensing network element having to decide the adjustment method itself, thus reducing the burden on the sensing network element.

[0139] Optionally, any one of the at least one detection layer may be, for example, a neural layer in a fully connected module, or a spatial attention layer or channel attention layer in a hybrid attention mechanism module, or it may be other types of neural network layers, such as a convolutional layer. The different detection layers in the at least one detection layer may be of the same type, for example, all being neural layers in a fully connected module; or the different detection layers may be of different types, for example, detection layer 1 in the at least one detection layer may be a spatial attention layer, and detection layer 2 in the at least one detection layer may be a convolutional layer. Any one of the at least one detection layers may be used, for example, to improve the first AI network's ability to detect features of perceived targets, thereby improving the performance of the first AI network.

[0140] The first information sent by the sensing node indicates the adjustment method of the first AI network. This first AI network can be understood as either the first AI network deployed on the sensing node side or the first AI network deployed on the sensing network element side. If the first information indicates an adjustment method, it means that the sensing network element will adjust the first AI network. After receiving the first information, the sensing network element can adjust the first AI network deployed on its side. Specifically, if the first AI network indicated by the first information is understood as the first AI network deployed on the sensing network element side, then the sensing network element can adjust the first AI network deployed on its side according to the adjustment method of the first AI network on its side. Alternatively, if the first AI network indicated by the first information is understood as the first AI network deployed on the sensing node side, then the sensing network element can adjust the first AI network deployed on its side according to the adjustment method of the first AI network on the sensing node side.

[0141] The first AI network may include one or more neurons, and the at least one neuron indicated by the first information may be some or all of the neurons within the first AI network. For example, the first AI network may include one or more neural layers, wherein each neural layer may include one or more neurons, and different neural layers may include the same or different numbers of neurons. The at least one neuron indicated by the first information may belong to some or all of the neural layers included in the first AI network. Optionally, the at least one neuron may include deep neurons within the first AI network. For example, in the first AI network, neurons included in neural layers near the output site may be called deep neurons; neurons included in neural layers near the input site may be called superficial neurons. By adjusting the deep neurons of the first AI network, the processing of more fine-grained information by the first AI network can be enhanced, and the accuracy of the first AI network can be improved.

[0142] The first AI network may include one or more neural layers, and the first information may also indicate the addition of at least one detection layer within the first AI network. The location of the at least one detection layer within the first AI network is not limited; for example, the location of the at least one detection layer within the first AI network may be indicated by the first information. Optionally, the at least one detection layer may include a detection layer located at the input site of the first AI network, and / or include a detection layer located at the output site of the first AI network. For example, one of the at least one detection layers may be located at the input site of the first AI network. Figure 9 Taking the network structure shown as an example, a detection layer can be added to the input part of the first AI network. For example, the perceived data can first be input into this detection layer and then reach the channel attention module. Optionally, the detection layer added to the input part of the first AI network can also be called the target perception feature detection head neural layer, or it can have other names.

[0143] For example, one of the at least one detection layer may be located at the output of the first AI network. Figure 9 Taking the network structure shown as an example, a detection layer can be added to the output of the first AI network, for instance, by adding this detection layer to a fully connected network. Optionally, the detection layer added to the fully connected network of the first AI network can also be called the target perception parameter output fully connected neural layer, or it can have other names.

[0144] The adjustment method indicated by the first information can be determined randomly, or it can be determined based on the deviation between the first and second perception results (e.g., the deviation of the first parameter, or the deviation of N parameters). Taking the adjustment method indicated by the first information based on this deviation as an example: For instance, if the deviation is large, an adjustment method involving adding a detection layer can be used. Adding a detection layer represents a relatively large adjustment to the first AI network. If the deviation is large, it indicates that the accuracy of the first AI network is low, or its performance has deteriorated significantly, so this adjustment method can be used to improve the performance of the first AI network. Conversely, if the deviation is small (i.e., although the deviation is greater than the corresponding threshold, it is relatively small), an adjustment method involving adjusting neurons can be used. Adjusting neurons represents a relatively small adjustment to the first AI network. If the deviation is large, it indicates that the performance of the first AI network has deteriorated less, so this adjustment method can be used to improve the performance of the first AI network.

[0145] As another optional implementation of the first information, the first information can indicate the deviation between the first perception result and the second perception result, for example, indicating the deviation of the first parameter, or indicating the deviation of each of the N parameters. Upon receiving the first information, the sensing network element can independently decide on the adjustment method, thereby adjusting the first AI network deployed on the sensing network element side using that adjustment method. In this implementation, the sensing network element can decide on the adjustment method, and the sensing node only needs to report the deviation of the perception result, which simplifies the implementation of the sensing node and allows the solution of this application embodiment to be applied to low-capacity sensing nodes.

[0146] As another optional implementation of the first information, the first information can instruct the adjustment of the first AI network. That is, the first information only indicates that the first AI network needs to be adjusted, which is equivalent to indicating the adjustment requirement, but does not indicate the adjustment method. Upon receiving the first information, the sensing network element can decide on the adjustment method itself, thereby adjusting the first AI network deployed on the sensing network element side using that method. In this implementation, the sensing network element can decide the adjustment method, which simplifies the implementation of the sensing node and allows the solution of this application embodiment to be applied to low-capacity sensing nodes.

[0147] As another optional implementation of the first information, the first information may include the adjustment result of the first AI network. In this implementation, the first AI network deployed on the sensing node side can be adjusted by the sensing node, for example, the sensing node can adjust the first AI network deployed on the sensing node side using the aforementioned adjustment method. After the adjustment is completed, the sensing node can obtain the adjustment result and then send the adjustment result to the sensing network element. The sensing network element receives the adjustment result and can update the first AI network deployed on the sensing network element side according to the adjustment result. For example, if the sensing node uses an adjustment method that adjusts the weights of at least one neuron in the first AI network, then the adjustment result may include the adjusted weight information of the at least one neuron, and / or the position information of the at least one neuron within the first AI network. As another example, if the sensing node uses an adjustment method that adds at least one detection layer to the first AI network, then the adjustment result may include information about the added at least one detection layer, such as the weight information of the at least one detection layer and / or the position information of the at least one detection layer within the first AI network. The weight information of a detection layer may include, for example, the weight information of some or all of the neurons in the detection layer. In this implementation, the sensing node can adjust the first AI network itself, which simplifies the implementation of the sensing network element. As can be seen, the adjustment process in this application embodiment can be performed on the sensing node side or on the sensing network element side, which is quite flexible.

[0148] In addition to the above-mentioned implementation methods, the first information may also have other implementation methods, and there are no restrictions on them.

[0149] If the first information includes the adjustment result of the first AI network, the sensing node can adjust the first AI network deployed on the sensing node side to obtain the adjustment result. Optionally, the sensing node can adjust the first AI network based on the deviation between the first and second sensing results to obtain the adjustment result. For example, the sensing node can determine the adjustment method based on the deviation, and then adjust the first AI network according to the adjustment method. In the embodiments of this application, the adjustment of the first AI network can be achieved by training the first AI network. That is, adjusting the first AI network can include training the first AI network. For example, the sensing node trains the first AI network by adjusting the weights of at least one neuron in the first AI network. Optionally, the sensing node can freeze the remaining neurons in the first AI network other than the at least one neuron. Freezing can be understood as not changing the weights of the remaining neurons during training, thereby achieving training of the at least one neuron by freezing the remaining neurons. For example, refer to Figure 10 This is a training example for adjusting the weights of at least one neuron within the first AI network for a sensing node. Figure 10 Taking the example where at least one neuron is a deep neuron in the first AI network, and the remaining neurons are shallow neurons in the first AI network. From Figure 10 As can be seen, the sensory node freezes the shallow neurons but not the deep neurons, thus enabling the training of the deep neurons.

[0150] For example, if the perception node adopts the adjustment method of adding at least one detection layer in the first AI network, then the perception node can first add the at least one detection layer in the first AI network, and then train the first AI network after adding the detection layer.

[0151] Additionally, training the first AI network requires one or more training samples. For example, a perception node can perform perception in the current scene (e.g., if the perception node determines that the first AI network needs adjustment, or if it determines that there is a deviation between the first perception result and the second perception result as described above, then it can perform perception), obtain perception data, and then train the first AI network based on the perception data. Optionally, the perception data may include one or more of the following: echo signals, the perception spectrum corresponding to the echo signals, or point cloud information corresponding to the echo signals.

[0152] If the first information indicates the adjustment method, or the deviation between the first and second perception results, or the adjustment of the first AI network, the first information may optionally include the perception data collected by the perception node (e.g., the perception data collected by the perception node in the current scene), so that the perception network element can train the first AI network deployed on the perception node side based on the perception data.

[0153] It is evident that regardless of whether the first AI network is adjusted by sensing nodes or sensing network elements, it can be trained based on the sensing data collected by the sensing nodes in the current scene. Since the sensing data used to train the first AI network is collected by the sensing nodes in the current scene and reflects the characteristics of the current scene, training the first AI network based on this sensing data can make the first AI network adapt to the current scene, thereby improving the generalization ability of the first AI network.

[0154] Optionally, the method may also include S703, whereby the sensing network element adjusts or updates the first AI network based on the first information.

[0155] For example, the first information may indicate the adjustment method, the deviation between the first and second perception results, or the adjustment of the first AI network. The sensing network element can then adjust the first AI network deployed on its side based on the first information, such as training the first AI network. Optionally, the first information may also include the perception data collected by the sensing nodes, which the sensing network element can use to train the first AI network. The method by which the sensing network element trains the first AI network is similar to the method by which the sensing nodes train the first AI network; please refer to the relevant description in S702.

[0156] For example, if the first information indicates the adjustment result, the sensing network element can update the first AI network deployed on the sensing network element side according to the adjustment result, so that the first AI network deployed on the sensing network element side is synchronized with the first AI network deployed on the sensing node side. In this way, the sensing network element does not need to retrain the first AI network.

[0157] If the first AI network is trained by a perception network element, optionally, the perception network element can also send the adjustment results of the first AI network to the perception node, so that the perception node can update the first AI network deployed on the perception node side according to the adjustment results. The adjustment results are similar to the adjustment results obtained by the perception node during training, as described above.

[0158] If the first AI network deployed on the sensing node side and the sensing network element side is updated, the sensing nodes and sensing network elements can continue to perform sensing using their respective deployed first AI networks. Since the solution implemented in this application improves the performance of the first AI network, it can improve sensing accuracy.

[0159] Optionally, the solution in this application embodiment can be continuously executed on the sensing node and sensing network element side, or it can be executed periodically, without limiting the execution conditions.

[0160] This application embodiment enables adjustments to the first AI network deployed on the sensing nodes and sensing network elements, thereby improving the performance of the first AI network. The first information can be determined based on the first and second sensing results. For example, if the sensing node determines, based on the first and second sensing results, that the performance of the first AI network is poor, or that the information output by the first AI network is inaccurate, then the first information can be used to adjust the first AI network, allowing it to be adjusted in a timely manner and improving its performance.

[0161] Figure 11 A schematic diagram of a device provided in an embodiment of this application is given. The device 1100 may be... Figure 7 The sensing network element or its circuit system described in the illustrated embodiment is used to implement the method corresponding to the sensing network element in the above method embodiments. Alternatively, the device 1100 may be... Figure 7 The sensing node or its circuit system described in the illustrated embodiment is used to implement the method corresponding to the sensing node in the above method embodiments. For example, one type of circuit system is a chip system.

[0162] Since the device 1100 in the embodiments of this application can implement the sensing method, the device 1100 can also be called a sensing device. In implementation, the device 1100 may have sensing function but no communication function, or it may have both sensing function and communication function. If the device 1100 has communication function, it may also be called a communication device, etc., without limitation.

[0163] The device 1100 includes at least one processor 1101. The processor 1101 can be used for internal processing within the device to implement certain control processing functions. Optionally, the processor 1101 includes instructions. Optionally, the processor 1101 can store data. Optionally, different processors can be independent devices, located in different physical locations, or located on different integrated circuits. Optionally, different processors can be integrated into one or more processors, for example, integrated on one or more integrated circuits.

[0164] Optionally, the device 1100 includes one or more memories 1103 for storing instructions. Optionally, the memories 1103 may also store data. The processor and the memories may be separate or integrated together.

[0165] Optionally, the device 1100 includes a communication line 1102 and at least one communication interface 1104. Since the memory 1103, communication line 1102, and communication interface 1104 are all optional, therefore... Figure 11 All are represented by dashed lines.

[0166] Optionally, device 1100 may further include a transceiver and / or an antenna. The transceiver can be used to send information to or receive information from other devices. The transceiver may be referred to as a transceiver unit, transceiver circuit, input / output interface, etc., and is used to realize the transmission and reception functions of device 1100 via the antenna. Optionally, the transceiver includes a transmitter and a receiver. For example, the transmitter can be used to generate a radio frequency (RF) signal from a baseband signal, and the receiver can be used to convert the RF signal back into a baseband signal.

[0167] Processor 1101 may include a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of programs according to the present application.

[0168] Communication line 1102 may include a path for transmitting information between the aforementioned components.

[0169] Communication interface 1104 uses any transceiver-like device for communicating with other devices or communication networks, such as Ethernet, radio access network (RAN), wireless local area network (WLAN), wired access network, etc.

[0170] The memory 1103 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 1103 may exist independently and be connected to the processor 1101 via communication line 1102. Alternatively, the memory 1103 may be integrated with the processor 1101.

[0171] The memory 1103 stores computer execution instructions for implementing the scheme of this application, and its execution is controlled by the processor 1101. The processor 1101 executes the computer execution instructions stored in the memory 1103, thereby realizing... Figure 7 The steps performed by the sensing network element or sensing node in the illustrated embodiment.

[0172] Optionally, the computer execution instructions in the embodiments of this application may also be referred to as application code, and the embodiments of this application do not specifically limit this.

[0173] In a specific implementation, as one embodiment, the processor 1101 may include one or more CPUs, for example... Figure 11 CPU0 and CPU1 in the CPU.

[0174] In a specific implementation, as one embodiment, device 1100 may include multiple processors, for example... Figure 11 Processors 1101 and 1105 are mentioned. Each of these processors can be a single-core (single-CPU) processor or a multi-core (multi-CPU) processor. A processor here can refer to one or more devices, circuits, and / or processing cores used to process data (e.g., computer program instructions).

[0175] when Figure 11When the device shown is a chip, such as a chip for a sensing network element or a sensing node, the chip includes a processor 1101 (and may also include a processor 1105), a communication line 1102, and a communication interface 1104. Optionally, it may include a memory 1103. Specifically, the communication interface 1104 may be an input interface, pins, or circuits, etc. The memory 1103 may be a register, cache, etc. The processor 1101 and processor 1105 may be a general-purpose CPU, microprocessor, ASIC, or one or more integrated circuits for controlling the execution of a program for the sensing method of any of the above embodiments.

[0176] This application embodiment can divide the device into functional modules according to the above method example. For example, each function can be divided into its own functional modules, or two or more functions can be integrated into one processing module. The integrated modules can be implemented in hardware or as software functional modules. The module division in this application embodiment is illustrative and only represents one logical functional division; in actual implementation, other division methods may be used. For example, in the case of dividing the device into functional modules corresponding to each function... Figure 12 This is a schematic diagram of an apparatus. The apparatus 1200 can be a sensing network element or sensing node involved in the above-described method embodiments, or a chip in a sensing network element or a chip in a sensing node. The apparatus 1200 includes a processing unit 1202 and a transceiver unit 1201. Since the apparatus 1200 in the embodiments of this application can implement a sensing method, the apparatus 1200 can also be called a sensing device. In implementation, the apparatus 1200 may have sensing functionality but no communication functionality, or it may have both sensing and communication functionality. If the apparatus 1200 has communication functionality, it can also be called a communication device, etc., without limitation.

[0177] It should be understood that the device 1200 can be used to implement the steps performed by the sensing network element or sensing node in the sensing method of the embodiments of this application, and the relevant features can be referred to above. Figure 7 The embodiments shown are not described in detail here.

[0178] Optional, Figure 12 The functions / implementation process of the transceiver unit 1201 and the processing unit 1202 can be obtained through Figure 11 The processor 1101 in the memory calls computer execution instructions stored in memory 1103 to implement the function. Alternatively, Figure 12 The function / implementation process of the processing unit 1202 can be achieved through... Figure 11 The processor 1101 in the memory calls computer execution instructions stored in the memory 1103 to implement this. Figure 12 The function / implementation process of the transceiver unit 1201 can be obtained through Figure 11 It is implemented using the communication interface 1104.

[0179] Optionally, when the device 1200 is a chip or circuit, the function / implementation process of the transceiver unit 1201 can also be implemented through pins or circuits. Optionally, the transceiver unit 1201 may include a transmitting unit and / or a receiving unit, wherein the transmitting unit is used to implement the transmitting function and the receiving unit is used to implement the receiving function; or, the transceiver unit 1201 may be an integral module capable of implementing the transmitting and / or receiving functions. Optionally, the transceiver unit 1201 may be implemented using a transceiver.

[0180] This application also provides a computer-readable storage medium storing a computer program or instructions that, when executed, implement the methods performed by the sensing network elements and / or sensing nodes in the aforementioned method embodiments. Thus, the functions described in the above embodiments can be implemented as software functional units and sold or used as independent products. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to it, or a 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 to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0181] This application also provides a computer program product comprising: computer program code, which, when run on a computer, causes the computer to perform the method executed by the sensing network element and / or sensing node in any of the foregoing method embodiments.

[0182] This application also provides a processing device, including a processor and an interface; the processor is used to execute the methods performed by the sensing network elements and / or sensing nodes involved in any of the above method embodiments.

[0183] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., solid-state disk (SSD)).

[0184] The various illustrative logic units and circuits described in the embodiments of this application can be implemented or operate the described functions using a general-purpose processor, digital signal processor (DSP), ASIC, field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof. The general-purpose processor can be a microprocessor; alternatively, it can be any conventional processor, controller, microcontroller, or state machine. The processor can also be implemented using a combination of computing devices, such as a digital signal processor and a microprocessor, multiple microprocessors, one or more microprocessors combined with a digital signal processor core, or any other similar configuration.

[0185] The steps of the methods or algorithms described in the embodiments of this application can be directly embedded in hardware, software units executed by a processor, or a combination of both. The software units can be stored in RAM, flash memory, ROM, erasable programmable read-only memory (EPROM), EEPROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium in the art. Exemplarily, the storage medium can be connected to the processor so that the processor can read information from the storage medium and write information to the storage medium. Optionally, the storage medium can also be integrated into the processor. The processor and storage medium can be disposed in an ASIC, which can be disposed in the terminal device. Optionally, the processor and storage medium can also be disposed in different components of the terminal device.

[0186] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0187] The contents of the various embodiments of this application can be referenced to each other. Unless otherwise specified or there is a logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced to each other. The technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationship.

[0188] It is understood that in the embodiments of this application, the sensing network elements and / or sensing nodes may execute some or all of the steps in the embodiments of this application. These steps or operations are merely examples. In the embodiments of this application, other operations or variations of various operations may also be performed. Furthermore, the steps may be performed in different orders as presented in the embodiments of this application, and it is not necessary to perform all the operations in the embodiments of this application.

Claims

1. A sensing method, characterized in that, The sensing method includes: The sensing node sends first information, which is determined by the sensing node based on the first and second sensing results corresponding to the first sensing target. The first information is used to instruct the sensing network element to adjust the weight of at least one neuron in the first artificial intelligence (AI) network, and / or to add at least one detection layer in the first AI network; or, the first information includes the adjustment result of the first AI network.

2. The method according to claim 1, characterized in that, The method further includes: The sensing node obtains the first sensing result and the second sensing result.

3. The method according to claim 2, characterized in that, The sensing node obtains the first sensing result and the second sensing result, including: The sensing node receives a first signal, which is an echo signal reflected, scattered, or diffracted by the first sensing target. The sensing node determines the first sensing result based on the sensing spectrum corresponding to the first signal, and inputs the information of the first signal into the first AI network to obtain the second sensing result output by the first AI network.

4. The method according to any one of claims 1 to 3, characterized in that, The first perception result and / or the second perception result includes one or more of the following information: The speed information of the first sensed target; Distance information of the first perceived target; The angle information of the first perceived target; The contour information of the first perceived target; The detection box information of the first perceived target, wherein the detection box information indicates that the first perceived target is included within the detection box; or, The point cloud information of the first perceived target.

5. The method according to any one of claims 1 to 4, characterized in that, In one optional implementation, both the first perception result and the second perception result include N parameters, where N is a positive integer, and the first perception result and the second perception result satisfy: The deviation between the value of the first parameter included in the first perception result and the value of the first parameter included in the second perception result is greater than a first threshold, wherein the first parameter is one of the N parameters; or, The deviations between the values ​​of the N parameters included in the first perception result and the values ​​of the N parameters included in the second perception result are each greater than N thresholds.

6. The method according to any one of claims 1 to 5, characterized in that, The first information is used to instruct the adjustment of the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network, wherein, The first information also includes sensing data collected in the current environment, which is used by the sensing network element to adjust the first AI network.

7. The method according to any one of claims 1 to 6, characterized in that, The first information is used to instruct the adjustment of the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network, the method further comprising: The sensing node receives the adjustment result of the first AI network, which includes the weight information of at least one neuron in the first AI network, or the weight information of at least one newly added detection layer in the first AI network. The sensing node updates the first AI network based on the adjustment result.

8. The method according to any one of claims 1 to 5, characterized in that, The first information includes the adjustment results of the first AI network, wherein the method further includes: The sensing node adjusts the first AI network based on the deviation between the first sensing result and the second sensing result to obtain the adjustment result.

9. The method according to claim 8, characterized in that, The sensing node adjusts the first AI network based on the deviation between the first sensing result and the second sensing result to obtain the adjustment result, including: The sensing node trains the first AI network based on the deviation and the sensing data collected in the current environment to obtain the adjustment result.

10. A sensing method, characterized in that, The method includes: The sensing network element receives first information, which is determined by the sensing node based on the first and second sensing results corresponding to the first sensing target. The first information is used to instruct the sensing network element to adjust the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network; or, the first information includes the adjustment results of the first AI network.

11. The method according to claim 10, characterized in that, The first information is used to instruct the adjustment of the weights of at least one neuron in the first AI network, and / or to add at least one detection layer in the first AI network, wherein, The first information also includes perception data collected in the current environment, which is used to adjust the first AI network.

12. The method according to claim 11, characterized in that, The method further includes: The sensing network element trains the first AI network according to the first information and the sensing data to obtain the adjustment result.

13. The method according to claim 10, characterized in that, The first information includes the adjustment results of the first AI network, wherein the method further includes: The sensing element updates the first AI network based on the adjustment result.

14. A sensing system, characterized in that, The sensing system includes sensing network elements and sensing nodes, wherein... The sensing node is used to perform the method as described in any one of claims 1 to 9; The sensing network element is used to perform the method as described in any one of claims 10 to 13.

15. An apparatus, characterized in that, The apparatus includes a module for performing the method as described in any one of claims 1 to 9, or a module for performing the method as described in any one of claims 10 to 13.

16. An apparatus, characterized in that, The apparatus includes a processor for performing the method as described in any one of claims 1 to 9, or the method as described in any one of claims 10 to 13.

17. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a computer program that, when run on a computer, causes the method as described in any one of claims 1 to 9 to be performed, or causes the method as described in any one of claims 10 to 13 to be performed.

18. A computer program product, characterized in that, The computer program product includes a computer program that, when run on a computer, causes the computer to perform the method as described in any one of claims 1 to 9, or causes the computer to perform the method as described in any one of claims 10 to 13.

Images