An outdoor communication information collection method and system

Abstract

The present application relates to the technical field of wireless communication network, and provides an outdoor communication information collection method and system. In order to solve the problem of how to collect the outdoor state information of a target terminal in a wireless communication network without relying on the active uplink signal transmission of the target terminal, the method transmits a non-communication dedicated downlink signal with sensing function, receives the echo reflected by the terminal, separates the target reflection signal component from the echo based on the inherent passive scattering characteristics of the terminal, extracts the physical layer characteristics of the target reflection signal component, maps the extracted characteristics to the outdoor state information of the terminal, generates a communication information collection record based on the state information, and finally completes the information collection, thereby realizing zero-power state information collection without relying on the active signal transmission of the terminal, and fundamentally breaking the contradiction between the endurance capability of the outdoor terminal and the timeliness of information collection.

Description

Technical Field

[0001] This invention relates to the field of wireless communication network technology, and particularly to information collection and terminal management technology in wireless communication networks. Specifically, it relates to an outdoor communication information collection method and system, which uses wireless signals from network-side nodes to perform target detection and status identification in order to collect communication information from outdoor terminals. Background Technology

[0002] Outdoor communication information collection refers to the automated process of acquiring and monitoring the status information of mobile terminals located in outdoor environments through wireless communication networks. It is one of the fundamental functions in the field of wireless communication network terminal management. Taking a field team collaboration scenario based on outdoor smart terminals (such as outdoor smartphones, outdoor smartwatches, handheld GPS devices, etc., used in outdoor and wilderness scenarios) as an example: A field exploration team is operating in a mountainous area without cellular network coverage. Team members carry smartphones or handheld terminals with satellite communication capabilities (i.e., outdoor smart terminals). The command center at the rear needs to monitor the current location and movement status of each member in real time to assess the team's safety situation and respond promptly to emergencies. During this process, the location and status information of each member's terminal needs to be continuously collected and transmitted back. However, continuous active communication in a network-free environment will rapidly consume its battery power, and charging conditions in the wilderness are extremely limited. There is an irreconcilable contradiction between the terminal's battery life and the real-time nature of information collection.

[0003] In traditional wireless communication networks, the collection of communication information from outdoor terminals primarily relies on proactive reporting from the terminal itself. Specifically, the terminal's built-in positioning module periodically calculates its own location and then proactively sends a measurement report containing location and status information to the network via cellular uplink channels or satellite communication links. In areas with weak or no coverage, the terminal continuously performs cell search and random network access, attempting to establish a connection with the network to complete the information reporting. The network plays a passive receiving role throughout the entire information collection process, only parsing and aggregating the data reported by the terminal.

[0004] However, in the aforementioned traditional methods of collecting communication information from outdoor terminals, an irreconcilable contradiction exists between the timeliness of information collection and the terminal's battery life. To improve the timeliness of information collection, the terminal must frequently initiate uplink communications, and each uplink transmission consumes a large amount of power, severely shortening the terminal's continuous working time in outdoor environments. Conversely, reducing the reporting frequency to extend battery life results in a significant lag in the status information obtained by the network side, failing to meet real-time monitoring requirements.

[0005] The root cause of this problem lies in the fact that the underlying signaling architecture of traditional wireless communication networks constructs a centralized "request-response" information acquisition model. Under this model, the network side can only decode and process uplink signals that are actively transmitted by authenticated and synchronized terminals on specific time-frequency resources allocated by the network, conforming to a preset communication protocol format. Any information about the terminal's status is considered to be payload data that must be actively encoded and injected into the network by the terminal itself as the information source. As a result, information acquisition in traditional communication networks cannot escape its strong dependence on uplink signals actively transmitted by terminals, leading to an irreconcilable technical contradiction between the battery life of outdoor terminals and the timeliness of information acquisition.

[0006] Therefore, how to break free from the dependence on the target terminal to actively transmit uplink signals in wireless communication networks and realize the collection of its outdoor status information has become an urgent technical problem to be solved in the field of outdoor communication information collection. Summary of the Invention

[0007] To address the aforementioned technical problems, this invention provides an outdoor communication information collection method and system. By transmitting sensing signals through network-side nodes and performing echo separation and state interpretation based on the inherent passive scattering characteristics of the terminal, a multi-dimensional passive information collection mechanism is constructed. This differs from the traditional "request-response" collection mode that relies on the terminal to actively report or respond to network paging. It enables the collection of identifiable status information with zero power consumption and zero communication involvement of the terminal.

[0008] To address the aforementioned technical problems, in a first aspect, the present invention provides the following technical solution: an outdoor communication information collection method applied to a network-side node, comprising: broadcasting a downlink sensing signal, wherein the downlink sensing signal is a wireless signal with sensing function that is not solely used for data communication; receiving an echo signal formed by the downlink sensing signal reflected by at least one target terminal, and separating a target reflection signal component corresponding to the target terminal from the echo signal based on the inherent passive scattering characteristics of the target terminal; performing physical layer feature extraction on the target reflection signal component, and mapping the extracted features to outdoor state information of the target terminal; and generating a communication information collection record corresponding to the target terminal based on the outdoor state information to complete the information collection.

[0009] Secondly, the present invention also provides an outdoor communication information acquisition system applied to a network-side node, comprising: a first broadcast module for broadcasting a downlink sensing signal, wherein the downlink sensing signal is a wireless signal with sensing function that is not only used for data communication; a first separation module for receiving an echo signal formed by the downlink sensing signal reflected by at least one target terminal, and separating a target reflection signal component corresponding to the target terminal from the echo signal based on the inherent passive scattering characteristics of the target terminal; a first extraction module for performing physical layer feature extraction on the target reflection signal component, and mapping the extracted features to outdoor status information of the target terminal; and a first generation module for generating a communication information acquisition record corresponding to the target terminal based on the outdoor status information, so as to complete the information acquisition.

[0010] The beneficial effects of this invention are as follows: In the above-mentioned outdoor communication information acquisition method and system, the method first establishes an independent excitation source that does not rely on the terminal actively transmitting uplink signals by broadcasting a non-communication-dedicated downlink signal that is not only used for data communication but also has sensing functions, thus providing an electromagnetic illumination basis for subsequent passive sensing. Based on this, by receiving the echo signal reflected by the terminal and separating the target reflected signal component from it according to the inherent passive scattering characteristics of the terminal, the identity of different terminals is distinguished, which is different from the acquisition method in traditional wireless communication networks that relies on the terminal actively reporting its identity. This upgrades target identification from "passively receiving identified information" to "actively identifying the identity of silent terminals". On this basis, by extracting physical layer features from the target reflected signal component and mapping the extracted features to outdoor state information, cross-domain interpretation from physical layer echo parameters to state data that the communication system can understand is realized. Finally, by generating communication information acquisition records corresponding to the target terminal based on the outdoor state information, the sensing results are encapsulated into information units that the communication network can recognize, thus completing the complete closed loop of information acquisition.

[0011] By combining the aforementioned interconnected effects, a multi-dimensional passive information sensing mechanism was constructed, consisting of "network-side active excitation - terminal passive scattering response - identity feature separation - physical layer state interpretation - data acquisition and record generation".

[0012] Thus, unlike the traditional "request-response" acquisition mode that relies on the terminal to actively report or respond to network paging, this invention treats the terminal as a physical entity that passively changes the electromagnetic environment. It utilizes the network-side sensing signal excitation and the network side autonomously completes echo reception, identity recognition, and status interpretation. This transforms the traditional "request-response" mode, where the terminal actively consumes power, into an "excitation-sensing" mode where the network side consumes zero terminal power. Ultimately, it achieves outdoor status information acquisition without relying on the target terminal to actively transmit uplink signals, fundamentally breaking the irreconcilable contradiction between the timeliness of information acquisition and the terminal's battery life. Attached Figure Description

[0013] Figure 1 A flowchart illustrating the outdoor communication information collection method provided in an embodiment of the present invention;

[0014] Figure 2 This is a schematic diagram of the first sub-process of the outdoor communication information collection method provided in an embodiment of the present invention;

[0015] Figure 3 This is a schematic block diagram of an outdoor communication information collection system provided in an embodiment of the present invention. Detailed Implementation

[0016] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0017] The outdoor communication information collection method and system provided by this invention can be a data processing engine carried by a network-side node, which can directly drive technical applications including but not limited to the following multiple dimensions:

[0018] First, in team safety management scenarios such as wilderness exploration, geological surveys, and outdoor competitions, team members are equipped with rugged smart terminals with satellite communication capabilities. In work areas without cellular network coverage, network-side nodes deployed in accompanying emergency communication vehicles, drone relay platforms, or portable sensing base stations continuously broadcast downlink sensing signals and receive reflected echoes from each member's terminal. Based on the inherent passive scattering characteristics of each terminal, identity separation and status interpretation are performed to generate real-time location and motion status information for each member. When a member is detected to be stationary for an extended period or to have abnormal status parameters, an alarm is automatically triggered and the last known location is output. This provides passive information collection support with zero terminal power consumption for wilderness search and rescue, fundamentally resolving the contradiction between terminal battery life and continuous location tracking in long-term outdoor operations.

[0019] Secondly, in the scenario of outdoor IoT device asset inspection and status monitoring, by deploying network side nodes on fixed sites or mobile inspection equipment, passive status data can be collected from outdoor IoT terminals that are widely distributed and have high manual inspection costs. The existence, location and operating status information can be obtained without the terminals actively reporting, providing an efficient technical means for asset inventory and anomaly detection.

[0020] Third, in the scenario of monitoring outdoor public facilities in smart cities, passive status monitoring of communication terminals deployed in urban public spaces can be carried out through roadside sensing nodes or mobile sensing devices, supporting the automatic identification of abnormal movement or stationary states and providing data support for urban emergency management.

[0021] The following detailed description of some embodiments of the present invention is provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0022] Example 1: An outdoor communication information collection method.

[0023] In this embodiment, please refer to Figure 1 , Figure 1 This is a flowchart illustrating an outdoor communication information collection method provided in an embodiment of the present invention. Figure 1 As shown, in this embodiment, the method is applied to a network-side node and includes the following steps S11-S14:

[0024] S11: Broadcast downlink sensing signal, which is a wireless signal with sensing function that is not only used for data communication.

[0025] Downlink sensing signals are excitation signals actively emitted by network-side nodes, using radio electromagnetic waves as the carrier, to illuminate the spatial area where a target terminal is located, thereby stimulating the target terminal to generate a passive scattering response that can be used for information acquisition. Unlike downlink signals in traditional communication networks that are only used to carry user data (such as Synchronization Signal Block (SSB), Channel State Information Reference Signal (CSI-RS), Physical Downlink Shared Channel (PDSCH), etc.), the core design goal of downlink sensing signals is to provide a high-quality electromagnetic illumination source (i.e., an active detection signal source) for subsequent passive target detection and state identification, rather than transmitting communication data. Specifically, its "sensing function" is manifested in the fact that the signal possesses good autocorrelation and cross-correlation characteristics in the time, frequency, spatial, or code domains, enabling network-side nodes to accurately estimate physical layer parameters such as signal propagation delay, Doppler shift, and angle of arrival from the received reflected echoes.

[0026] The implementation methods of downlink sensing signals include, but are not limited to: 1) linear frequency modulation (Chirp) fusion waveform based on orthogonal frequency division multiplexing (OFDM). This waveform combines the high spectral efficiency of OFDM with the large time-bandwidth product characteristics of Chirp signals, enabling high-resolution time delay estimation in frequency-selective fading channels; 2) spread spectrum sensing signals modulated by pseudo-random sequences (such as m-sequences, Gold sequences, Zadoff-Chu sequences, etc.). This waveform utilizes the sharp autocorrelation peaks of the pseudo-random sequences to achieve accurate time of arrival measurement. The specific waveform chosen depends on the system's comprehensive requirements for sensing resolution, anti-interference capability, and spectral compatibility. In this embodiment of the invention, the OFDM-Chirp fusion waveform can be selected as the preferred implementation of the downlink sensing signal based on the following technical considerations: First, the waveform can directly utilize the existing OFDM baseband processing architecture of the communication transceiver, and the base station can be given sensing function through software upgrade, with low hardware modification cost; Second, the pulse compression characteristics of the Chirp signal enable it to obtain high processing gain at low transmit power, which is suitable for the outdoor deployment needs of power-constrained network side nodes (such as UAV relay platforms).

[0027] The specific implementation of this step is as follows:

[0028] First, the sensing signal generation unit of the network-side node determines the time-frequency resource location, bandwidth, duration, and transmission period of the downlink sensing signal based on the pre-configured sensing resource block parameters. The sensing resource block parameters can be statically or semi-statically configured through the core network or Operation and Maintenance Management (OAM) entity. The configuration can be: bandwidth approximately 20MHz to 100MHz, duration approximately 1ms to 10ms, and transmission period approximately 100ms to 1s. Furthermore, it can be dynamically adjusted according to the expected movement speed of the target terminal—a shorter transmission period can be used for high-speed moving targets to improve the state update rate, while a longer transmission period can be used for low-speed or stationary targets to reduce the power consumption of the network-side node.

[0029] Secondly, the sensing signal generation unit generates a baseband sensing signal within the allocated sensing resource block according to the selected waveform type (e.g., an OFDM-Chirp fusion waveform). Specifically, for an OFDM symbol containing N subcarriers, where N is the total number of subcarriers within the sensing resource block, typically ranging from 512 to 4096, with a preferred value of 2048, the modulation symbol X(k) on the k-th subcarrier can be generated according to the following formula:

[0030] X(k)=A(k)·exp(jπ·(k-K0)² / B);

[0031] Where A(k) is the amplitude weighting coefficient of the k-th subcarrier, used to control the power distribution of the downlink sensing signal on each subcarrier to achieve spectrum shaping. The typical value range of A(k) is 0 to 1, with a preferred value of 1, i.e., each subcarrier transmits with equal amplitude; K0 is the center subcarrier index, taking the integer part of N / 2; B is the frequency modulation parameter, with a typical value range of 10¹² Hz / s to 10¹ 4 The preferred value is 5 × 10¹³ Hz / s, and the selection criteria are as follows: the modulation frequency within this range can provide sufficient pulse compression ratio to achieve sub-meter distance resolution while maintaining compatibility with the subcarrier spacing of the existing 5G NR standard; j is the imaginary unit, and π is pi. After the modulation symbol undergoes inverse fast Fourier transform (IFFT) and the addition of a cyclic prefix (CP), a time-domain OFDM-Chirp signal is formed.

[0032] Then, the sensing signal generation unit upconverts the generated baseband sensing signal to a preset carrier frequency (e.g., 3.5 GHz or 28 GHz band) via a radio frequency front-end, amplifies it by a power amplifier, and broadcasts it to the spatial area where the target terminal is located via an antenna array. The typical range of the transmit power is 23 dBm to 43 dBm, and it can be adjusted according to the coverage distance requirements.

[0033] It should be noted that the waveform selection for the aforementioned downlink sensing signal is not limited to OFDM-Chirp fusion waveforms. In one example, a sensing signal based on Zadoff-Chu sequence modulation can be used, leveraging the constant envelope zero autocorrelation (CAZAC) characteristic of the Zadoff-Chu sequence, which is suitable for scenarios requiring a low peak-to-average power ratio (PAPR). In another example, a pulse train waveform based on phase coding can be used, improving the signal-to-noise ratio of distant target detection through the coherent accumulation of multiple pulses, suitable for wide-area sensing scenarios with large coverage. Any signal form capable of providing high-resolution physical layer parameter estimation capabilities for subsequent passive target detection falls within the scope of protection of this invention.

[0034] Furthermore, the deployment method of the aforementioned network-side sensing nodes is not limited to fixed sites. In one example, the network-side sensing node can be deployed on a vehicle-mounted platform, activating and operating upon arrival at the target area to provide sensing coverage of the surrounding area; in another example, the network-side sensing node can be mounted on a drone platform, providing accompanying coverage of the ground area by hovering or stationary. Any technical means capable of broadcasting downlink sensing signals to illuminate the spatial area where the target terminal is located falls within the protection scope of this invention.

[0035] For example, consider a scenario involving the safety management of a wilderness expedition team: A hiking expedition team of approximately 10 members is operating in a mountainous area without cellular network coverage. A network-side sensing node is deployed on the accompanying emergency communication vehicle. Its transmitting antenna is mounted on the roof. Under typical parameter configurations (e.g., transmit power approximately 30dBm, antenna gain approximately 10dBi, signal bandwidth approximately 40MHz, processing gain corresponding to the time-bandwidth product approximately 46dB, receiver sensitivity approximately -90dBm, etc.), it can form an effective sensing radius of approximately 1km for the rugged smart terminals with satellite communication capabilities held by team members. The coverage area can be further extended by adjusting the transmit power, antenna gain, and signal processing parameters. When the team rests at the campsite, the network-side sensing node continuously broadcasts downlink sensing signals. The signal uses an OFDM-Chirp fusion waveform with a bandwidth of approximately 40MHz, a center frequency of approximately 3.5GHz, and a transmission period of approximately 200ms, illuminating the target spatial area centered on the campsite, providing a stable electromagnetic excitation source for subsequent steps. As the team moves along the predetermined route, the sensing coverage can be extended to the entire team's activity area by pre-deploying portable sensing beacons at key nodes along the way, or by having accompanying drone relay platforms carry sensing nodes for accompanying area coverage.

[0036] In step S11, an active excitation source, independent of the terminal's active transmission signal, is constructed by broadcasting a downlink sensing signal with sensing capabilities. The transmission platform for the downlink sensing signal is not limited to fixed sites; it can also be mounted on mobile vehicles or portable devices to achieve area coverage in a task-driven manner. This transforms the traditional 'request-response' mode of passively waiting for terminal reports in communication networks into an excitation mode where the network actively conducts electromagnetic illumination. This provides the physical basis for subsequent steps to achieve passive state sensing with zero terminal power consumption and solves the problem of passive sensing failing to start due to the lack of an independent excitation source.

[0037] S12: Receive the echo signal formed by the downlink sensing signal reflected by at least one target terminal, and based on the passive scattering characteristics inherent in the target terminal, separate the target reflection signal component corresponding to the target terminal from the echo signal.

[0038] Echo signal: refers to the superposition of electromagnetic wave signals that return to the receiving antenna of the network-side node after the downlink sensing signal encounters the target terminal and other objects in its surrounding environment during its propagation in space, undergoing physical processes such as reflection, scattering, and diffraction. The echo signal contains rich physical layer information, such as the target terminal's distance, speed, and material characteristics, but it is also mixed with clutter interference from other scattering objects such as the ground, vegetation, and buildings. The signal strength of the echo signal is usually much lower than that of the transmitted signal, and its fading characteristics are affected by multipath effects and the terminal's motion state.

[0039] Passive scattering characteristics: These refer to the unique backscattering response of a target terminal to incident downlink sensing signals when it is in a passive state without actively emitting any signals or consuming its own power, due to its inherent physical structure and radio frequency circuit characteristics. This backscattering response manifests as measurable and distinguishable physical layer characteristics in terms of electromagnetic wave parameters, which is the physical basis for distinguishing different individual terminals. Passive scattering characteristics are "inherent" to the target terminal, originating from its own unchangeable physical properties rather than externally assigned identifiers. Their formation mechanism includes three levels: First, the determinacy of geometry—the physical shape, size, and antenna layout of the terminal determine the induced current distribution and re-radiation mode of electromagnetic waves on its surface (i.e., the radiation pattern of the radar cross section RCS); second, the determinacy of material parameters—the dielectric constant and permeability of the terminal's casing, circuit board, and components determine the amplitude and phase changes of electromagnetic waves during penetration, reflection, and absorption; third, the impedance specificity of the RF front-end—the RF front-end, composed of the terminal's receiving antenna, matching network, and low-noise amplifier (LNA), presents a complex, frequency-dependent passive impedance network even in non-operating states. This passive impedance network remodulates (changes its amplitude and phase) the captured sensing signal and retransmits it, constituting the terminal's "RF fingerprint." It should be noted that the physical basis of passive scattering characteristics is solid and verifiable in engineering: from the perspective of classical electromagnetics, any object will produce backscattering under the illumination of incident electromagnetic waves. Its backscattering response is a function of frequency, determined by the object's geometry, constituent materials, and the impedance characteristics of the RF front-end. Even among terminals of the same model and batch, due to differences in component manufacturing tolerances, microscopic traces on printed circuit boards, and solder joint morphology, their passive scattering characteristics are physically distinguishable, sufficient to support subsequent identification signal separation. Passive scattering characteristics can be quantitatively obtained in a laboratory environment using standard RF measurement equipment. The measurement methods and data formats can refer to the well-known S-parameter measurement techniques of vector network analyzers, which will not be elaborated here. How to obtain and characterize passive scattering characteristics will be detailed in subsequent Examples 2 and 3.

[0040] The specific implementation of this step is as follows:

[0041] First, the sensing receiver array of the network-side node (which can be co-located with or separate from the sensing transmit antenna) simultaneously acquires composite scattered echoes in the radio frequency environment while broadcasting downlink sensing signals. The sensing receiver array can contain M spatial diversity antenna elements (M≥2), each of which can independently receive echo signals. The received echo signals are amplified by a low-noise amplifier (LNA), down-converted, and converted to digital by analog-to-digital converter (ADC) to form a digital baseband echo signal matrix.

[0042] Secondly, the baseband processing unit of the network-side node preprocesses the digital baseband echo signal, including but not limited to DC offset correction, IQ imbalance compensation, bandpass filtering, etc., to eliminate interference introduced by transceiver hardware non-ideals.

[0043] Then, the baseband processing unit can call a pre-built passive scattering feature template library for the target terminal. This library stores the mapping relationship between the terminal identifier of each target terminal to be monitored and its corresponding passive scattering feature template. Using the passive scattering feature template as a priori constraint, the baseband processing unit performs an identity-based signal separation algorithm on the pre-processed echo signal. Specifically, the separation process involves transforming the echo signal from the time domain to the frequency domain, calculating the correlation between the complex value of the echo signal and the complex reflection coefficient of the corresponding frequency point in the passive scattering feature template at each frequency point, determining the distribution range of the reflected signal corresponding to the target terminal on the time delay axis based on the correlation peak position, and extracting the signal components belonging to this distribution range from the echo signal. The specific template matching and separation algorithm will be detailed in the subsequent embodiment 2.

[0044] It should be noted that the signal separation methods based on passive scattering characteristics described above are not limited to frequency domain correlation methods. In one example, a time-domain matched filtering method can be used, which involves convolving the received signal with a time-domain reference signal reconstructed from the feature template. This method is suitable for scenarios where the time-domain characteristics of the terminal feature template are significant. In another example, a subspace decomposition-based algorithm (such as the MUSIC multi-signal classification algorithm) can be used. This method separates the signal subspace from the noise subspace by performing feature decomposition on the echo signal covariance matrix, and then searches using the projection of the feature template vector onto the noise subspace. This method is suitable for high-resolution separation scenarios where the echo signals from multiple terminals are closely spaced in the time delay domain. Any technical means that can separate specific terminal signal components from composite echoes using pre-stored inherent terminal characteristics is within the scope of protection of this invention.

[0045] For example, following the previous example, the sensing and receiving array on the emergency communication vehicle continuously receives composite echo signals from the terminals of expedition members and their surrounding environment. For a certain type of rugged smart terminal belonging to member A, its passive scattering feature template had been obtained through laboratory calibration and stored in the terminal passive scattering feature template library before departure. The baseband processing unit retrieves the passive scattering feature template corresponding to member A's terminal identifier from the terminal passive scattering feature template library and performs frequency domain correlation matching between the received echo signal and the passive scattering feature template. Due to the unique scattering characteristics generated by member A's terminal's metal body and specific antenna layout, the correlation operation shows a significant peak at the time delay position corresponding to the actual distance of the terminal. Based on this, the target reflection signal component corresponding to member A's terminal is successfully separated from the mixed signal containing echoes from approximately 10 member terminals and environmental clutter.

[0046] In step S12, by receiving the echo signal and performing identity signal separation based on the inherent passive scattering characteristics of the target terminal, the anonymous echo signal that can only detect "whether there is an object" in traditional technology is transformed into an identity echo signal component that can identify "which terminal it is". This provides a dedicated signal source for subsequent steps to extract the physical layer features and state information of a specific terminal, and solves the key problem that passive sensing cannot be associated with a specific terminal due to the lack of terminal identification.

[0047] S13: Perform physical layer feature extraction on the target reflected signal components, and map the extracted features to the outdoor status information of the target terminal.

[0048] Physical layer features refer to quantifiable parameters extracted from the reflected signal components of the target, directly reflecting the changes the signal undergoes during transmission at the physical layer. This concept differs from features extracted from higher layers of the communication protocol stack (such as the MAC layer and application layer). In this invention, physical layer features specifically refer to signal parameters that have a deterministic mapping relationship with the terminal's spatial location and motion state. These mainly include, but are not limited to, time delay parameters (reflecting the propagation time of the signal from transmission to reception), Doppler parameters (reflecting the signal frequency shift caused by terminal motion), and their derived statistics.

[0049] Outdoor status information refers to a structured data set that describes the real-time status of a target terminal in an outdoor environment, derived from extracted physical layer features through a pre-defined mapping relationship. Its content may include, but is not limited to, the terminal's radial distance relative to network-side nodes, radial velocity, and motion state type (e.g., stationary, walking, fast movement). Outdoor status information serves as the direct basis for subsequent generation of communication information collection records.

[0050] The specific implementation of this step is as follows:

[0051] First, the feature extraction unit of the network-side node receives the target reflected signal component from step S12 and performs a time-delay-Doppler two-dimensional Fourier transform on the target reflected signal component. Mathematically, this can be represented as follows: after mixing the time-domain received signal s(t) with the conjugate of the transmitted signal, Fourier transforms are performed on the fast time dimension (reflecting time delay) and the slow time dimension (reflecting Doppler) respectively to obtain the two-dimensional spectrum S(τ, f_d), where τ is the time delay axis and f_d is the Doppler frequency axis. Specific transformation formulas and parameter configurations will be detailed in subsequent embodiment 5.

[0052] Secondly, the feature extraction unit automatically detects and extracts a set of feature parameters related to the target terminal from the two-dimensional spectrum S(τ, f_d), including: the position τ0 of the main peak on the time delay axis (corresponding to the time delay value where the maximum scattering energy is located), the position f_d0 of the main peak on the Doppler axis, the 3dB width of the main peak (reflecting time delay spread and Doppler spread), and the sidelobe distribution pattern. The detection of the main peak position can be achieved using the constant false alarm rate (CFAR) algorithm, the specific implementation method and parameter configuration of which will be detailed in the subsequent embodiment 5.

[0053] Then, the state mapping unit performs a deterministic mapping from physical layer parameters to outdoor state information based on the set of feature parameters output by the feature extraction unit. The specific mapping relationship can be as follows:

[0054] Based on the peak delay position τ0 and the electromagnetic wave propagation speed c, the radial distance of the terminal relative to the network-side node is calculated using the formula d=c·τ0 / 2.

[0055] Based on the Doppler position of the main peak f_d0, the carrier frequency f_c, and the electromagnetic wave propagation speed c, the radial velocity of the terminal is calculated using the formula v=f_d0·c / (2f_c);

[0056] Based on the statistical characteristics of the Doppler spread parameters (such as the spectral width and shape of the Doppler spectrum), combined with preset classification rules or classification models, the motion state type of the terminal is determined. The specific motion state classification method will be detailed in the subsequent embodiment 6.

[0057] It should be noted that the implementation methods of the above-mentioned physical layer feature extraction and state mapping are not limited to time-delay-Doppler two-dimensional Fourier transform. In one example, a sparse reconstruction algorithm based on compressed sensing can be used to achieve high-resolution time-delay-Doppler estimation under the condition that the sampling rate is lower than the Nyquist rate, which is suitable for scenarios where the sensing signal bandwidth is limited or storage resources are scarce. In another example, an end-to-end mapping model based on deep learning can be used to directly take the original target reflection signal component as input and output outdoor state information, which is suitable for scenarios that pursue high accuracy and low latency after training with a large amount of data. Any technical means that can extract physical layer features related to the terminal state from the target reflection signal component and map them into state information are within the protection scope of this invention.

[0058] For example, continuing from the previous example, the network-side node extracts the time-delay-Doppler spectrum from the target reflection signal component of member A, and detects that the main peak time delay τ0 is approximately 2.0 μs, corresponding to a radial distance of approximately 300 m; the main peak Doppler frequency shift f_d0 is approximately -30 Hz, corresponding to a radial velocity of approximately -1.3 m / s (the negative sign indicates distance from the sensing node); the Doppler spread parameter is approximately 15 Hz. After comparison with a preset threshold range, the motion state type is determined to be "walking". Thus, the outdoor state information of member A terminal is generated as: {Terminal identifier: member A, radial distance: approximately 300 m, radial velocity: approximately -1.3 m / s, motion state type: walking}.

[0059] In step S13, physical layer feature extraction and state mapping are performed on the target reflected signal components to transform the original echo physical parameters into structured outdoor state information that the communication system can understand. This realizes cross-domain interpretation from "physical layer perception" to "communication layer information" and provides a directly usable state data source for the terminal management function of the communication network.

[0060] S14: Based on the outdoor status information, generate a communication information collection record corresponding to the target terminal to complete the information collection.

[0061] The specific implementation of this step is as follows:

[0062] First, the information encapsulation unit of the network-side node receives the outdoor status information of the target terminal output by S13. The outdoor status information includes at least one status parameter (such as radial distance, radial velocity, motion state type, etc.).

[0063] Secondly, the information encapsulation unit acquires auxiliary metadata for this information collection, including but not limited to: collection timestamp (which can be provided by the clock module of the network-side node with millisecond-level accuracy), node identifier and / or location identifier of the network-side node, terminal identifier of the target terminal, and configuration parameter index of the downlink sensing signal used in this collection.

[0064] Then, the information encapsulation unit encapsulates the outdoor status information and auxiliary metadata according to a preset data structure, generating a communication information collection record corresponding to the target terminal. The preset data structure can adopt key-value pairs, JSON, or a custom binary TLV (type-length-value) format. An exemplary communication information collection record may contain the following fields: record serial number, terminal identifier, collection timestamp, node location identifier, radial distance, radial velocity, motion state type, signal-to-noise ratio estimate, collection record check code, etc.

[0065] Finally, the information encapsulation unit reports the generated communication information collection records to the core network or a designated application server via a backhaul network (such as wired Ethernet, wireless backhaul link, or satellite backhaul link) to complete a full outdoor communication information collection process. The reporting cycle can be synchronized with the broadcast cycle of the downlink sensing signal, or it can be triggered on demand according to application requirements.

[0066] It should be noted that the generation and reporting methods of the aforementioned communication information collection records are not limited to real-time, line-by-line reporting. In one example, a batch reporting method can be used, aggregating multiple collection records generated within a preset time window into a single data packet before unified reporting, suitable for scenarios with limited backhaul network bandwidth. In another example, an edge computing unit can be configured locally on the network-side node to perform preliminary trajectory fitting or anomaly detection on multiple consecutive collection records, and then only the processing results are reported, suitable for scenarios with strict limitations on the amount of data transmitted. Any technical means capable of structurally encapsulating and reporting outdoor status information to complete information collection falls within the protection scope of this invention.

[0067] For example, following the previous example, after the information encapsulation unit of the network-side node receives the outdoor status information of Member A's terminal, it obtains the current collection timestamp T3 (e.g., 2025-07-05 10:30:00.200), combines it with the node location identifier of the emergency communication vehicle (e.g., GPS coordinates: N30.12345°, E120.67890°), and encapsulates it to generate a communication information collection record: {Record Serial Number: REC-001, Terminal Identifier: Member A, Collection Timestamp: T3, Node Location: N30.12345° E120.67890°, Radial Distance: Approximately 300m, Radial Velocity: Approximately -1.3m / s, Motion Status Type: Walking, Estimated Signal-to-Noise Ratio: Approximately 15dB}. This record is reported in real time to the application server of the rear command center via the satellite backhaul link, and the situation display screen of the command center updates the real-time location and motion status icon of Member A accordingly.

[0068] In step S14, structured communication information collection records are generated based on outdoor status information and reported. The results of physical layer perception are encapsulated into standard information units that can be identified, transmitted, and stored by the communication network, completing a complete closed loop from "signal transmission" to "information storage" and providing applicable terminal status data services for upper-layer applications.

[0069] In this embodiment, through sequential processing of S11-S14, an active excitation source is constructed by broadcast sensing signals, identity-identified echoes are separated based on the inherent scattering characteristics of the terminal, physical layer features are extracted and mapped to outdoor state information, and communication information collection records are generated to complete the information loop. This fully realizes the outdoor communication information collection method of claim 1, and provides a zero-power passive sensing basic scheme that does not rely on the active transmission of signals by the terminal for the field of outdoor communication information collection.

[0070] Example 2: Target reflection signal component separation method based on feature template matching.

[0071] In this embodiment, based on the inherent passive scattering characteristics of the target terminal, the target reflected signal component corresponding to the target terminal is separated from the echo signal, including:

[0072] Obtain a preset passive scattering feature template corresponding to the target terminal. The preset passive scattering feature template contains a sequence of complex reflection coefficients of the radio frequency front end of the target terminal in a preset frequency band. Each complex reflection coefficient in the sequence of complex reflection coefficients contains an amplitude response value and a phase offset value.

[0073] The complex received values ​​of the echo signal at each frequency point within the preset frequency band are multiplied by the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and an inverse Fourier transform is performed based on the result of the conjugate multiplication at each frequency point to obtain the frequency domain cross-correlation spectrum between the echo signal and the preset passive scattering feature template.

[0074] The peak position and peak amplitude of the frequency domain cross-correlation spectrum are determined, and the signal components within the time delay interval corresponding to the peak position where the peak amplitude exceeds a preset matching threshold are identified as target reflected signal components.

[0075] in:

[0076] A pre-defined passive scattering feature template refers to a standardized data set acquired in advance through laboratory measurements or on-site calibration and stored in the template library of network-side nodes. This template characterizes the scattering response of a specific target terminal to incident electromagnetic waves in a passive state. The pre-defined passive scattering feature template can be statically pre-calibrated in a one-time process or dynamically calibrated periodically as needed to ensure it matches the current state of the corresponding target terminal. Its core component is the sequence of complex reflection coefficients of the target terminal's RF front-end within a pre-defined frequency band. The function of the passive scattering feature template is to provide a reference for matched filtering during signal separation, enabling network-side nodes to identify the signal of a specific target terminal from complex composite echoes.

[0077] Complex received value: This refers to the signal value obtained at each frequency point k within a preset frequency band after the echo signal received by the network-side node is transformed from the time domain to the frequency domain using a Fast Fourier Transform (FFT), and is usually denoted as Y(k). The magnitude |Y(k)| of each complex received value Y(k) represents the amplitude of the received signal at that frequency point, and its argument ∠Y(k) represents the phase of the received signal at that frequency point. The complex received value is composed of the vector superposition of the scattering responses of all scatterers (including target terminals, environmental clutter, etc.) in the echo signal at that frequency point. Its complex form preserves the phase information of the signal and is the basis for subsequent conjugate multiplication with the passive scattering feature template to achieve phase alignment and coherent accumulation. In engineering implementation, the complex received value can be obtained as follows: the preprocessed digital baseband echo signal in S12 is transformed to the frequency domain using FFT, and the frequency domain samples at each subcarrier or frequency point are extracted to obtain the sequence of complex received values ​​for each frequency point. The specific calculation process can be found in the well-known FFT algorithm in this field, and will not be elaborated here.

[0078] Complex reflection coefficient: This refers to the ratio of the complex amplitude of the reflected signal to the incident signal when irradiated by an incident electromagnetic wave of a specific frequency. It is a complex number, with its magnitude representing the amplitude change of the reflected signal relative to the incident signal, and its phase angle representing the phase shift of the reflected signal relative to the incident signal. The expression can be:

[0079] Γ(k)=|Γ(k)|·exp(j·∠Γ(k));

[0080] Where Γ(k) is the complex reflection coefficient at the k-th frequency point, |Γ(k)| is the amplitude response value, ∠Γ(k) is the phase offset value, and j is the imaginary unit, where:

[0081] Amplitude response value: refers to the magnitude of the complex reflection coefficient |Γ(k)|, which reflects the target terminal's ability to reflect incident signal energy at this frequency. The value is affected by the radiation efficiency of the target terminal's antenna, the degree of impedance matching, and the structural scattering cross section.

[0082] Phase offset value: refers to the phase angle ∠Γ(k) of the complex reflection coefficient, reflecting the rotation effect of the target terminal's RF front-end impedance network on the phase of the incident signal. It takes an angle value between -π and π and is a key parameter constituting the individual identification characteristics of the target terminal. Its internal working principle is as follows: the lumped parameter elements (such as inductors and capacitors) and distributed parameter structures (such as transmission lines and antenna traces) of the target terminal's RF front-end constitute a complex impedance Z_L(f). When this complex impedance is mismatched with the antenna's characteristic impedance Z0, reflection occurs at the connection point. The phase of the reflection coefficient is determined by the nature (inductive or capacitive) and degree of impedance mismatch.

[0083] Frequency-domain cross-correlation spectrum: This refers to a two-dimensional spectrum obtained by performing frequency-by-frequency conjugate multiplication and inverse Fourier transform on the received echo signal and a preset passive scattering feature template in the frequency domain. The spectrum is plotted with time delay on the horizontal axis and correlation amplitude on the vertical axis. The peak position of the frequency-domain cross-correlation spectrum indicates the relative time delay (i.e., the distance to the target terminal) of the signal component in the echo signal that matches the corresponding preset passive scattering feature template, while the peak amplitude indicates the confidence level of the match. Mathematically, it is an efficient implementation of the cross-correlation function between the received echo signal and the signal corresponding to the preset passive scattering feature template in the frequency domain. Compared to time-domain convolution, the computational complexity of the frequency-domain cross-correlation spectrum can be reduced from O(N²) to O(N·logN), making it suitable for high-bandwidth, high-resolution scenarios.

[0084] Peak amplitude: refers to the amplitude of a local maximum point in the frequency domain cross-correlation spectrum that exceeds a certain noise floor level. The closer the peak amplitude is to 1, the higher the matching degree between the signal component at that time delay position in the received echo signal and the preset passive scattering feature template.

[0085] The specific implementation of this embodiment is as follows:

[0086] First, the template management unit of the network-side node receives a terminal monitoring list from the upper-layer application or core network. Based on the target terminal identifiers included in the list, it retrieves and loads the corresponding passive scattering feature template from a preset passive scattering feature template library. Furthermore, each passive scattering feature template can be stored in vector form, with vector elements being complex numbers corresponding to complex reflection coefficients at various frequency points within a preset frequency band. The typical range of the preset frequency band is 1MHz to 200MHz, covering a continuous frequency range near the system's operating carrier frequency. The typical value of the frequency point spacing is 15kHz to 100kHz, compatible with OFDM subcarrier spacing or frequency sweep steps. The method for constructing the passive scattering feature template library will be detailed in subsequent embodiment 3.

[0087] Secondly, the baseband processing unit transforms the preprocessed digital baseband echo signal in S12 from the time domain to the frequency domain using Fast Fourier Transform (FFT) to obtain the complex received value Y(k) at each frequency point k. Simultaneously, it extracts the complex reflection coefficient Γ(k) at the corresponding frequency point k from the corresponding passive scattering feature template.

[0088] Then, for each frequency point k, the baseband processing unit performs a frequency-point-by-frequency conjugate multiplication operation between the complex received value and the corresponding complex reflection coefficient, i.e., calculating C(k) = Y(k)·Γ*(k), where Γ*(k) represents the complex conjugate of Γ(k). In other words, C(k) is calculated as the product of the complex conjugate of Y(k) and Γ(k), meaning the complex conjugate of Γ(k) is first taken and then multiplied with Y(k). The purpose of conjugate multiplication is to eliminate the phase shift introduced by the passive scattering feature template, ensuring that the signal components from the target terminal achieve coherent superposition after the subsequent inverse Fourier transform, while the signal components from other target terminals or clutter remain incoherent due to phase mismatch.

[0089] Next, the baseband processing unit performs an inverse fast Fourier transform (IFFT) on the conjugate multiplication result sequence C(k), transforming the frequency domain correlation result to the time delay domain to obtain the frequency domain cross-correlation spectrum R(τ), where τ is the time delay variable. The computational complexity of R(τ) is O(N·logN), where N is the total number of frequency points.

[0090] Then, the baseband processing unit executes a peak detection algorithm on the frequency domain cross-correlation spectrum R(τ). Peak detection can employ an adaptive threshold method based on constant false alarm rate (CFAR) or set a fixed threshold. A typical fixed matching threshold ranges from 0.1 to 0.5 (normalized amplitude), with a preferred value of 0.2. The selection criteria are: this threshold ensures a detection probability (typically greater than 95%) while keeping the false alarm probability within an acceptable range (typically less than 1%). For peak values ​​whose amplitude exceeds the preset matching threshold, the corresponding time delay position τ_p and peak amplitude A_p are recorded.

[0091] Finally, the baseband processing unit determines the signal components within the time delay interval [τ_p-Δτ / 2, τ_p+Δτ / 2] as the target reflection signal components corresponding to the target terminal, where Δτ is the time delay resolution (Δτ=1 / B, B is the signal bandwidth). If there are multiple peaks exceeding the threshold, the main component can be selected based on the peak amplitude, or all components can be retained to cope with multipath scenarios.

[0092] It should be noted that the above-mentioned calculation method for the frequency domain cross-correlation spectrum is not limited to frequency-by-frequency point conjugate multiplication and cumulative summation IFFT. In one example, a segmented frequency domain correlation method can be used, dividing the preset frequency band into multiple sub-bands, calculating the sub-band correlation spectrum in each sub-band, and then performing incoherent accumulation, which is suitable for frequency-selective fading channels. In another example, a generalized cross-correlation (GCC) method can be used, pre-whitening filtering (such as PHAT weighting) is performed on the received signal and the template before conjugate multiplication to sharpen the correlation peak and improve the time delay estimation accuracy. Any technical means to match the received signal with the feature template and generate the correlation spectrum through frequency domain operations is within the protection scope of this invention.

[0093] For example, continuing from the previous example, for the target terminal of member A, its preset passive scattering feature template may contain complex reflection coefficients at approximately 2048 frequency points within a preset frequency band (approximately 3.49 GHz to 3.51 GHz, bandwidth approximately 20 MHz). The amplitude response value of each complex reflection coefficient may be between approximately 0.05 and 0.3 (reflecting the small radar cross section of the target terminal), and the phase offset value may be distributed between -π / 2 and π / 2. After performing FFT on the received echo signal, the network-side node performs Y(k)·Γ*(k) operation at each frequency point, and obtains the frequency domain cross-correlation spectrum through IFFT. In the frequency domain cross-correlation spectrum, a normalized peak of approximately 0.8 appears at a time delay of approximately 2.0 μs, exceeding the preset matching threshold of approximately 0.2. The signal components within this time delay interval are determined to be the target reflection signal components of the target terminal of member A.

[0094] In this embodiment, by introducing a preset passive scattering feature template as a matching benchmark and using frequency domain cross-correlation spectrum calculation as the core operation for signal separation, the traditional signal separation is transformed from a "blind separation" mode that relies on signal strength or statistical characteristics into an "identified matching separation" mode based on prior terminal features. Thus, based on the identity-based signal separation achieved in Embodiment 1 above, this embodiment further solves the problem of how to specifically implement signal separation based on passive scattering features, significantly improving the accuracy of separation and anti-interference capability.

[0095] Example 3: Method for constructing passive scattering feature templates.

[0096] In this embodiment, obtaining the preset passive scattering feature template corresponding to the target terminal includes:

[0097] When the target terminal is in a non-working state, a preset frequency sweep excitation signal is broadcast to the space where the target terminal is located. The frequency range of the preset frequency sweep excitation signal covers the preset frequency band.

[0098] The system receives the reflection response signal generated by the target terminal under the illumination of the preset frequency sweep excitation signal, and separates the scattering component related to the radio frequency front-end impedance characteristics of the target terminal from the reflection response signal.

[0099] The scattering components are demodulated in complex form to extract the amplitude response value and phase offset value at each frequency sweep point, construct the complex reflection coefficient sequence, and associate the complex reflection coefficient sequence with the terminal identifier of the target terminal to form the preset passive scattering feature template.

[0100] in:

[0101] The preset frequency sweep excitation signal refers to a continuous wave or stepped-frequency signal transmitted by a calibration device (which can be a network-side node or a dedicated calibration instrument) to the target terminal during the passive scattering feature template construction stage. Its frequency varies linearly or stepwise within a preset frequency band over time. Its function is to excite the target terminal's RF front-end to generate a measurable scattering response at all frequencies of interest. Frequency sweeping methods can include stepped-frequency continuous wave (SFCW) or linear frequency modulated continuous wave (FMCW). SFCW measures the response point-by-point by transmitting single-frequency signals of different frequencies in different time slots; FMCW covers the entire frequency band in a single measurement by transmitting a linearly varying continuous wave. The typical range of the sweep step interval is 10kHz to 100kHz, and the typical range of the total number of frequency points N is 500 to 5000.

[0102] Scattered component: This refers to the portion of the scattered signal extracted from the reflected response signal of the target terminal after processing such as clutter suppression and denoising. It is primarily determined by the impedance network of the target terminal's RF front-end. The amplitude and phase of this scattered signal vary with frequency, reflecting the frequency response characteristics of the target terminal's antenna port reflection coefficient. Its separation can be achieved using the background cancellation method: first, measure the ambient background scattering without the target terminal; then, measure the total scattering after the target terminal is placed in the array; finally, subtract the two measurements vectorwise to obtain the individual scattered component from the target terminal.

[0103] Complex demodulation refers to the process of down-converting the received RF or IF scattered signal to baseband and extracting the in-phase (I) and quadrature (Q) components to reconstruct the complex signal. Specifically, the received reflected response signal is mixed with the in-phase carrier cos(2πf_LO t) and the quadrature carrier sin(2πf_LO t) of the local oscillator. After low-pass filtering, two baseband signals, I(t) and Q(t), are obtained. The complex baseband signal is then represented as I(t) + jQ(t). The amplitude √(I² + Q²) of this complex signal corresponds to the amplitude response value, and the phase arctan(Q / I) corresponds to the phase offset value.

[0104] The specific implementation of this embodiment is as follows:

[0105] First, the calibration operator places the target terminal in a non-operating state (power off or in flight mode) at a predetermined location in the calibration environment (such as a microwave anechoic chamber or an open area without strong reflection interference). The position and orientation of the calibration equipment relative to the geometric relationship with the target terminal should be accurately recorded to ensure geometric consistency between the passive scattering feature template and the actual deployment scenario.

[0106] Secondly, the signal source of the calibration equipment generates a preset frequency sweep excitation signal. Taking SFCW as an example, the signal source starts from the starting frequency f_start and increases by Δf point by point to the ending frequency f_end, with the frequency range [f_start, f_end] covering the preset frequency band. At each frequency point, the transmit power remains constant, typically ranging from 0dBm to 20dBm, and the transmit duration can be approximately 100μs to 1ms per frequency point to ensure that the receiver has sufficient integration time to obtain stable measurement values.

[0107] Then, the receiving end of the calibration equipment synchronously acquires the reflection response signal at each swept frequency point. After the received reflection response signal is amplified by an LNA and bandpass filtered, it is demodulated with a copy of the transmitted signal (I / Q demodulation) to extract the complex baseband signal value of that frequency point, including the amplitude response value |S(f_k)| and the phase offset value ∠S(f_k).

[0108] Next, the processing unit of the calibration equipment performs background cancellation and clutter suppression processing on the acquired raw reflection response signal. The specific operation of background cancellation can be as follows: The environmental background scattering data B(f_k) without a target terminal is pre-acquired and stored, and then subtracted from the total scattering data T(f_k) with a target terminal to obtain the target terminal's individual scattering component S(f_k) = T(f_k) - B(f_k). Further clutter suppression may include, but is not limited to, time-domain windowing and frequency-domain filtering to reduce multipath reflections and noise interference.

[0109] Then, the processing unit constructs the complex reflection coefficient sequence Γ(k) = S(f_k) / A(f_k) of the target terminal based on the complex values ​​of the scattering component S(f_k) at each frequency point f_k, where A(f_k) is the known complex amplitude of the transmitted signal at that frequency point (which can be obtained through calibration measurement). This results in a complex reflection coefficient sequence containing N elements, where each element consists of a triplet of {frequency f_k, amplitude response value |Γ(k)|, and phase offset value ∠Γ(k)}.

[0110] Finally, the processing unit associates the complex reflection coefficient sequence with the unique terminal identifier (such as IMEI, MEID, or custom terminal number) of the target terminal, packages and stores it as a preset passive scattering feature template file, and uploads it to the template library of the network-side node. The template library management functions can include adding, deleting, modifying, querying templates and version control, and can be deployed in the local storage of the network-side node or a centralized template management server in the core network.

[0111] It should be noted that the above template construction method is not limited to laboratory frequency sweep calibration. In one example, an online self-calibration method can be used, that is, after system deployment, the known signal generated by the target terminal briefly powered on in a known location and known state is used as a reference to calculate its scattering characteristics in reverse, which is suitable for scenarios where laboratory calibration cannot be performed in advance. In another example, a virtual calibration method based on electromagnetic simulation software can be used, by importing the terminal's CAD model and material parameters, and using full-wave electromagnetic simulation (such as the finite integral method FIT, the method of moments MoM) to calculate its scattering response at various frequency points, which is suitable for the early development stage when the terminal physical prototype is not yet ready. Any technical means that can acquire and store the scattering response characteristics of the terminal to the incident signal in a passive state are within the protection scope of this invention.

[0112] For example, following the previous example, before the expedition team departed, a certain model of rugged smart terminal belonging to member A was calibrated in the laboratory. The calibration equipment used a stepped-frequency continuous wave method, sweeping frequencies from approximately 3.49 GHz to approximately 3.51 GHz in steps of approximately 50 kHz, for a total of approximately 401 frequency points. The rugged smart terminal was in a powered-off state and placed on the calibration bracket. Reflection response signals of approximately 500 μs were collected at each frequency point. After background cancellation and I / Q demodulation, a complex reflection coefficient sequence was obtained. At approximately 3.5 GHz, the amplitude response value was approximately 0.12, and the phase offset value was approximately -0.34 radians. This complex reflection coefficient sequence was associated with member A's terminal identifier "Member_A_Device_01" to generate a feature template file "template_Member_A.bin", which was then uploaded to the template library of the emergency communication vehicle.

[0113] In this embodiment, by defining a complete template construction process of frequency sweep excitation, scattering separation, complex demodulation and associated storage, an engineering-implementable means of obtaining the preset passive scattering feature template used in Embodiment 2 above is provided, thereby solving the fundamental technical problem of "where does the passive scattering feature template come from" and providing a solid data foundation for the deployability and scalability of the entire passive identity perception system.

[0114] Example 4: Frequency domain cross-correlation spectrum refinement method based on frame segmentation processing.

[0115] In this embodiment, the complex received values ​​of the echo signal at each frequency point within the preset frequency band are multiplied by the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and then an inverse Fourier transform is performed based on the result of the conjugate multiplication at each frequency point to obtain the frequency domain cross-correlation spectrum between the echo signal and the preset passive scattering feature template, including:

[0116] The echo signal is divided into continuous time-domain signal frames according to a preset time frame length, and adjacent time-domain signal frames have a preset overlap interval.

[0117] Perform a Fourier transform on each of the time-domain signal frames to obtain the frequency-domain receiving vector corresponding to each time-domain signal frame;

[0118] Each frequency domain receiving vector is multiplied element-wise by the frequency domain template vector formed by the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and the result of the element-wise conjugate multiplication is subjected to inverse Fourier transform to obtain the time delay domain cross-correlation spectrum corresponding to each time domain signal frame.

[0119] The time-delay domain cross-correlation spectra corresponding to each time-domain signal frame are spliced ​​together in chronological order to form a time-delay-time two-dimensional cross-correlation distribution map, which is a frame-by-frame refined representation of the frequency domain cross-correlation spectrum.

[0120] in:

[0121] Delay-domain cross-correlation spectrum: This refers to the one-dimensional cross-correlation spectrum obtained by performing frequency-domain cross-correlation calculations on a single time-domain signal frame, with time delay as the horizontal axis. Its physical meaning corresponds to the time delay distribution of the reflected signal components from the target terminal within that frame. Since each frame is relatively short (usually on the order of milliseconds), the single-frame delay-domain cross-correlation spectrum can approximately reflect the distance information of the target terminal at that instant, but the signal-to-noise ratio may be limited.

[0122] A time-delay two-dimensional cross-correlation distribution map (TD-SCDMA) is a two-dimensional grayscale or color image formed by stacking the time-delay domain cross-correlation spectra corresponding to multiple consecutive time-domain signal frames on the vertical axis (time axis) in chronological order. The horizontal axis of the TD-SCDMA represents time delay (corresponding distance), and the vertical axis represents time (corresponding frame number or absolute time). The brightness or color of each pixel represents the cross-correlation amplitude at that time-delay coordinate. Compared to the frequency domain cross-correlation spectrum obtained from a single calculation in Example 2, the TD-SCDMA adds a time dimension, visually displaying the trajectory of the target terminal's distance over time. Simultaneously, it improves the detection signal-to-noise ratio through multi-frame incoherent accumulation. Essentially, the TD-SCDMA is a refinement and expansion of the frequency domain cross-correlation spectrum in the time dimension; it is a frame-by-frame refined representation of the frequency domain cross-correlation spectrum.

[0123] The specific implementation of this embodiment is as follows:

[0124] First, the baseband processing unit of the network-side node performs frame segmentation processing on the preprocessed digital baseband echo signal in S12. The typical value of the time frame length T_frame can range from 1ms to 10ms, with a preferred value of 5ms. The selection criteria are: this frame length is sufficient to distinguish distance changes caused by walking speed (approximately 1 to 2 m / s) (approximately 5mm to 10mm per frame), while ensuring that each frame contains a sufficient number of sampling points (for a 100MHz sampling rate, a 5ms frame contains approximately 500k sampling points) to meet the frequency resolution requirements of FFT. The typical value of the overlap ratio between adjacent frames can range from 25% to 75%, with a preferred value of 50%. This is to avoid the loss of time continuity caused by frame boundary truncation effects, while also considering processing throughput.

[0125] Secondly, the baseband processing unit performs an FFT on each time-domain signal frame, transforming it from the time domain to the frequency domain to obtain the frequency-domain received vector Y_m(k) corresponding to each frame, where m is the frame number and k is the frequency index. The number of FFT points N_FFT is determined by the number of sampling points within the frame, with typical values ​​ranging from 512 to 4096.

[0126] Then, the baseband processing unit loads the corresponding preset passive scattering feature template from the preset passive scattering feature template library, and arranges the complex reflection coefficient sequence in the passive scattering feature template in the order of the same frequency point to form the frequency domain template vector Γ(k). For each frame m, element-wise conjugate multiplication C_m(k) = Y_m(k)·Γ*(k) is performed, and inverse fast Fourier transform (IFFT) is performed on C_m(k) to obtain the time delay domain cross-correlation spectrum R_m(τ) corresponding to the frame. The specific implementation process of element-wise conjugate multiplication and IFFT is the same as in Example 2, and will not be repeated here.

[0127] Next, the baseband processing unit arranges the time-delay domain cross-correlation spectra R_m(τ) corresponding to each frame on the time axis in ascending order of frame number m, forming a two-dimensional time-delay cross-correlation distribution map P(τ, m). When there is overlap between frames, the step on the time axis is T_frame × (1 - overlap ratio), instead of T_frame. Further time-dimensional low-pass filtering can be performed between adjacent frames to smooth noise.

[0128] Finally, the baseband processing unit can replace the single-frequency domain cross-correlation spectrum R(τ) in Example 2 with the time-delay-time two-dimensional cross-correlation distribution map P(τ, m) for subsequent peak detection and signal separation. The advantage of using the two-dimensional distribution map for peak detection is that two-dimensional CFAR detection can be performed on the time-delay-time plane, and incoherent accumulation can be performed along the time axis (such as taking the average or maximum value of the correlation amplitude of each frame), which significantly improves the detection probability under low signal-to-noise ratio conditions.

[0129] It should be noted that the above-mentioned frame segmentation and stitching implementation methods are not limited to fixed frame lengths and fixed overlap ratios. In one example, an adaptive frame length adjustment method can be used, dynamically adjusting the frame length according to the detected terminal motion speed—using shorter frame lengths for high-speed targets to reduce intra-frame motion blur, and using longer frame lengths for low-speed targets to improve frequency resolution. In another example, a weighted overlap summation (WOLA) method can be used, applying a window function (such as a Hamming window or Blackman window) to each frame during framing, and weighting the overlapping intervals during stitching to further suppress spectral leakage caused by frame boundary truncation effects. Any technical means of obtaining a two-dimensional correlation distribution map by framing the echo signal and performing time-series stitching of the correlation results of each frame is within the protection scope of this invention.

[0130] For example, following the previous example, the network-side node processes the received echo signal into frames of approximately 5ms each with an overlap of approximately 50%. For approximately 1 second of continuous echo data, approximately 400 frames are obtained. After performing frequency domain cross-correlation calculations on each frame, approximately 400 time-delay domain cross-correlation spectra are obtained. These time-delay domain cross-correlation spectra are then stitched together in chronological order to form a time-delay-time two-dimensional cross-correlation distribution map. In this map, a persistent peak trajectory can be clearly observed, with its time delay position slowly increasing from approximately 2.0μs (corresponding to a distance of approximately 300m) to approximately 2.1μs (corresponding to a distance of approximately 315m), reflecting the process of the target terminal (member A) gradually moving away from the sensing node at a walking speed. By performing incoherent accumulation of the correlation amplitude along the time axis for approximately 10 frames (approximately 50ms), the peak signal-to-noise ratio is improved by approximately 10dB compared to a single frame, significantly enhancing detection reliability.

[0131] In this embodiment, by introducing a pipeline of frame-by-frame processing, frame-by-frame cross-correlation calculation, and time-dimensional stitching, the single frequency domain cross-correlation spectrum calculation in Embodiment 2 above is refined into a time-delay-time two-dimensional cross-correlation distribution map that can reveal the trajectory of target distance evolution over time. Thus, while maintaining the frequency domain cross-correlation spectrum as the basic calculation framework, the problem of decreased single correlation spectrum detection performance due to non-stationarity of echo signals caused by target terminal motion is further solved, significantly improving the detection signal-to-noise ratio and distance tracking continuity of dynamic targets.

[0132] Example 5: A state mapping method based on time delay-Doppler joint analysis.

[0133] In this embodiment, please refer to Figure 2 , Figure 2 This is a schematic diagram of the first sub-process of the outdoor communication information collection method provided in an embodiment of the present invention. Figure 2 As shown, in this embodiment, physical layer feature extraction is performed on the target reflected signal components, and the extracted features are mapped to the outdoor state information of the target terminal, including:

[0134] S21. Perform a time-delay-Doppler two-dimensional Fourier transform on the target reflected signal component to obtain the time-delay-Doppler spectrum of the target reflected signal component;

[0135] S22. Extract the main peak position, main peak width, and Doppler expansion parameters corresponding to the target terminal from the time-delay-Doppler spectrum;

[0136] S23. Determine the radial distance of the target terminal relative to the network-side node based on the main peak position, determine the radial velocity of the target terminal based on the time change rate of the main peak position, and determine the motion state type of the target terminal based on the Doppler expansion parameters;

[0137] S24. The radial distance, the radial velocity, and the motion state type are used as the outdoor state information of the target terminal.

[0138] in:

[0139] The time-delay-Doppler spectrum refers to the two-dimensional spectrum obtained by performing Fourier transforms on the reflected signal components of the target in both the fast time dimension (corresponding to a single pulse or OFDM symbol, reflecting time delay) and the slow time dimension (corresponding to multiple pulses or multiple OFDM symbols, reflecting Doppler frequency shift). The horizontal axis represents the time delay τ, and the vertical axis represents the Doppler frequency f_d. The amplitude of each pixel represents the scattered energy at the corresponding time delay and Doppler frequency. The time-delay-Doppler spectrum is a standard tool in the field of radar and communication sensing integration. Its general processing flow can refer to the pulse Doppler processing method known in the art, and will not be elaborated here. In this invention, the specific calculation configuration of the time-delay-Doppler spectrum has been customized for outdoor terminal sensing scenarios, including the selection of the number of pulse accumulations, the number of Fourier transform points, and the constant false alarm rate detection threshold, etc., as detailed in the subsequent steps of this embodiment and the relevant description in Embodiment 6.

[0140] Doppler spread parameter: refers to the degree of spectral broadening of the reflected signal component in the Doppler dimension, usually quantified by the 3dB bandwidth or root-mean-square bandwidth of the Doppler spectrum. Its physical cause is that slight differences in relative velocity between different scattering parts of the target terminal, or micro-motion modulation caused by attitude changes during the overall movement of the target terminal, result in the reflected signal containing a Doppler frequency range, rather than a single frequency value. The Doppler spread parameter is closely related to the motion state of the target terminal—the spread of a stationary target terminal is close to zero (contributed only by environmental micro-vibrations); the spread of a walking target terminal is approximately 10Hz to 30Hz due to the swaying of limbs and body; and the spread of a target terminal in a moving vehicle is typically greater than 50Hz due to vehicle vibration and rapid multipath changes.

[0141] Radial distance: refers to the straight-line distance between the target terminal and the receiving antenna of the network-side node, determined by the electromagnetic wave round-trip propagation time τ0 and the speed of light in free space c, and calculated using the formula d = c·τ0 / 2. Here, the factor 1 / 2 originates from the signal round-trip path, and the value of c is 2.99792458 × 10⁻⁶. 8 m / s.

[0142] Radial velocity: refers to the velocity component of the target terminal along the radial direction of the network-side node (i.e., the direction of the line connecting the two). Positive values ​​indicate approaching and negative values ​​indicate moving away. The calculation formula is v=-f_d0·c / (2f_c), where f_d0 is the Doppler frequency shift, f_c is the carrier frequency of the downlink sensing signal, and the factor 2 comes from the double Doppler frequency shift introduced by bidirectional propagation. The sign is negative so that the velocity is positive when moving away (consistent with radar convention).

[0143] The specific implementation of this embodiment is as follows:

[0144] First, the feature extraction unit of the network-side node acquires the target reflection signal component output by S12. The target reflection signal component can be a two-dimensional data matrix, where the fast time dimension (columns) corresponds to the sampling points within a single pulse (reflecting time delay), and the slow time dimension (rows) corresponds to multiple consecutive pulses (reflecting Doppler). The typical range of the number of pulses K is 64 to 512, with an optimal value of 128. The selection criteria are: this number of pulses can provide sufficient Doppler frequency resolution (Δf_d=PRF / K, approximately 1Hz to 5Hz) within a typical sensing signal transmission period (approximately 100ms to 200ms) to distinguish between walking and stationary states.

[0145] Secondly, the feature extraction unit performs a time-delay-Doppler two-dimensional Fourier transform on the two-dimensional data matrix. Specifically, this involves performing a first FFT along the fast time dimension (each column) to convert the time-domain sampling into a range gate; and a second FFT along the slow time dimension (each row) to convert the inter-pulse phase changes into Doppler frequencies. The number of points for the two FFTs are N_τ and N_d, respectively, with typical values ​​ranging from N_τ = 512 to 2048, and N_d = K or extended to 256 by zero padding.

[0146] Then, the feature extraction unit executes a two-dimensional constant false alarm rate (CFAR) detection algorithm on the obtained time-delay-Doppler spectrum to automatically extract the main peak corresponding to the target terminal. A typical implementation of two-dimensional CFAR detection can be as follows: a hollow rectangular window (guard window and reference window) is set with the cell to be detected as the center. The average amplitude of all cells within the reference window is calculated as the noise power estimate. The amplitude of the cell to be detected is compared with this noise power estimate multiplied by a preset threshold factor T. If it exceeds the threshold, it is determined to be the target peak. The threshold factor T can range from 10 to 20 (corresponding to a signal-to-noise ratio threshold of approximately 10 dB to 13 dB), with a preferred value of 13, which can be configured by parameters.

[0147] Next, the feature extraction unit extracts various feature parameters from the detected target peak: the main peak delay position τ0 is directly calculated from the delay index of the target peak; the main peak Doppler position f_d0 is calculated from the Doppler index of the target peak; the main peak width is measured at a point where the Doppler dimension and the target peak amplitude decrease by 3dB; and the Doppler expansion parameter B_d is obtained by calculating the amplitude-weighted root mean square bandwidth centered on the target peak in the Doppler dimension.

[0148] Then, the state mapping unit performs a deterministic mapping from physical layer features to outdoor state information. Radial distance d and radial velocity v are calculated according to the formulas described above. The motion state type is determined by matching the extracted Doppler extension parameter B_d with a preset classification rule; the specific classification method will be detailed in subsequent embodiment 6.

[0149] Finally, the state mapping unit combines the calculated radial distance, radial velocity, and motion state type into an outdoor state information, adds the terminal identifier and timestamp, and then transmits it to S14 for subsequent encapsulation.

[0150] It should be noted that the aforementioned time-delay-Doppler analysis method is not limited to two-dimensional FFT. In one example, a method based on sparse time-frequency analysis can be used, leveraging the sparsity of the target in the time-delay-Doppler domain, to achieve super-resolution estimation under conditions of a small number of pulses through compressed sensing reconstruction algorithms. In another example, a method based on space-time adaptive processing (STAP) can be used, jointly processing the spatial domain information and fast and slow time domain information of the array antenna to estimate target parameters while suppressing clutter, suitable for environments with strong ground clutter. Any technical means capable of extracting time delay and Doppler information from the target reflected signal components falls within the protection scope of this invention.

[0151] For example, continuing from the previous example, the feature extraction unit performs a time-delay-Doppler two-dimensional FFT on the target reflection signal component of member A (approximately 128 pulses, sampling rate approximately 100MHz). In the obtained time-delay-Doppler spectrum, CFAR detects that the main peak is located at a time delay of approximately 2.0μs and a Doppler frequency of approximately -30Hz. The 3dB width of the main peak in the Doppler dimension (Doppler spread parameter) is approximately 15Hz. According to the formula, the radial distance = 3 × 10 8 ×2.0×10⁻ 6 / 2≈300m; Radial velocity = -(-30)×3×10 8 / (2×3.5×10 9 The Doppler extension parameter is approximately -1.29 m / s. This parameter is then input into the subsequent classification steps.

[0152] In this embodiment, by employing time-delay-Doppler two-dimensional Fourier transform and CFAR automatic detection, the main peak position, main peak width, and Doppler expansion parameters are systematically extracted from the target reflection signal components. Based on a determined physical formula, these parameters are mapped to radial distance, radial velocity, and motion state type. This transforms the abstract echo signal into structured outdoor state information that can be directly used by the communication system, thus solving the semantic gap between physical layer sensing data and communication layer terminal management requirements.

[0153] Example 6: A motion state classification method with adaptive signal-to-noise ratio.

[0154] In this embodiment, determining the motion state type of the target terminal based on the Doppler extension parameters includes:

[0155] Obtain the signal-to-noise ratio estimate of the target reflected signal component within a preset time period;

[0156] Based on the signal-to-noise ratio (SNR) estimate, a subset of target threshold intervals that matches the SNR estimate is dynamically selected from a preset set of Doppler expansion parameter threshold intervals. The preset set of Doppler expansion parameter threshold intervals includes at least one preset static state threshold interval, at least one preset walking state threshold interval, and at least one preset fast movement state threshold interval. Each threshold interval in the preset set of Doppler expansion parameter threshold intervals has a preset one-to-one correspondence with a preset motion state label.

[0157] The Doppler expansion parameter is compared with the target threshold interval subset to determine the target threshold interval in which the Doppler expansion parameter falls.

[0158] Based on the preset one-to-one correspondence, a preset motion state label corresponding to the target threshold range is determined, and the determined preset motion state label is used as the motion state type of the target terminal.

[0159] in:

[0160] Signal-to-noise ratio (SNR) estimate: This refers to the estimated ratio of signal power to noise power of the reflected signal component within a preset time period, typically expressed in decibels (dB). The SNR estimate can be obtained using various well-known SNR estimation algorithms, such as estimation methods based on second and fourth moments (M2M4), estimation methods based on maximum likelihood (ML), or estimation methods based on eigenvalue decomposition. In this invention, the M2M4 estimation method is preferred. It separates signal power and noise power by calculating the second and fourth moments of the received signal samples, offering advantages such as low computational complexity and no need for training sequences. The typical range of the preset time period is 50ms to 500ms, with a preferred value of 100ms, during which the SNR can be assumed to change slowly.

[0161] The predefined set of Doppler expansion parameter threshold intervals refers to a predefined set of threshold intervals, each corresponding to a motion state type, and encompassing threshold interval boundary values ​​under different signal-to-noise ratio (SNR) conditions. The construction of the Doppler expansion parameter threshold interval set is based on the following physical fact: the estimation accuracy of the Doppler expansion parameter is significantly affected by the SNR—under low SNR conditions, noise will cause the estimated value of Doppler expansion to be too large (noise broadening effect), thus requiring appropriate widening of the threshold interval boundaries; under high SNR conditions, the estimated value is closer to the true physical expansion, and a narrower threshold interval can be used to achieve more refined classification. Therefore, this invention predefines at least two sets of threshold intervals for each motion state type, corresponding to different SNR condition intervals. Exemplary specific threshold interval divisions are described in subsequent steps.

[0162] The specific implementation of this embodiment is as follows:

[0163] First, the signal-to-noise ratio (SNR) estimation unit of the network-side node acquires in-phase quadrature (IQ) sampling data of the target reflection signal components within the most recent preset time period (e.g., 100ms). This data can be processed using the M2M4 estimation algorithm: calculate the second moment M2=E[|r(n)|²] and the fourth moment M4=E[|r(n)|²] of the IQ sampling sequence. 4 ], where r(n) is the nth complex sample value, and E[·] represents the time average; the signal power S and noise power N are estimated according to the formulas S=√(2M2²-M4) and N=M2-S respectively; the signal-to-noise ratio estimate SNR_est=10·log 10 (S / N), in dB.

[0164] Secondly, the adaptive threshold selection unit of the network-side node dynamically selects a matching target threshold interval subset from a preset set of Doppler spread parameter threshold intervals based on the signal-to-noise ratio (SNR) estimate, SNR_est. The specific selection rule can be as follows: the SNR is divided into several conditional intervals, such as a low SNR interval (SNR_est < 10dB), a medium SNR interval (10dB ≤ SNR_est < 20dB), and a high SNR interval (SNR_est ≥ 20dB). Each conditional interval corresponds to a set of predefined motion state threshold intervals. When SNR_est falls into a certain conditional interval, the threshold subset corresponding to that conditional interval is used as the currently used target threshold interval subset.

[0165] An exemplary set of preset Doppler expansion parameter threshold ranges is shown in Table 1 below:

[0166] Table 1

[0167] Signal-to-noise ratio conditional range Static state threshold range Walking state threshold range Fast movement state threshold range High signal-to-noise ratio (≥20dB) B_d < 5Hz 5Hz ≤ B_d < 30Hz B_d≥ 30Hz Signal-to-noise ratio (10 to 20 dB) B_d < 8Hz 8Hz ≤ B_d < 40Hz B_d≥ 40Hz Low signal-to-noise ratio (<10dB) B_d < 12Hz 12Hz≤B_d < 50Hz B_d≥ 50Hz

[0168] The above thresholds are determined as follows: Doppler spread data of the target terminal under different motion states (such as stationary, walking, riding in a vehicle, etc.) are collected in advance, and the mean and variance are statistically analyzed under different signal-to-noise ratio conditions. The points with equal false recognition rates among each category are taken as thresholds. Each threshold can be configured as a parameter and optimized according to the electromagnetic characteristics of the actual deployment environment.

[0169] Then, the Doppler extension parameter B_d extracted in Example 5 is compared with each threshold interval in the target threshold interval subset to determine the target threshold interval in which B_d falls.

[0170] Finally, based on the one-to-one correspondence between the preset threshold range and the motion state label, the motion state label corresponding to the target threshold range (such as "stationary", "walking", "fast movement", etc.) is determined, and the label is used as the motion state type of the target terminal.

[0171] It should be noted that the aforementioned signal-to-noise ratio (SNR) adaptive threshold selection mechanism is not limited to predefined condition intervals and threshold subsets. In one example, a soft classification method based on fuzzy logic can be used, dynamically adjusting the membership function of the Doppler expansion parameters according to the SNR estimate, outputting the confidence score for each state type instead of a hard classification label. In another example, a machine learning-based classifier (such as a Support Vector Machine (SVM) or a lightweight decision tree) can be used, with the Doppler expansion parameters and the SNR estimate as joint input features, outputting motion state classification results, suitable for scenarios requiring more complex motion patterns. Any technical means capable of dynamically adjusting the motion state classification strategy according to the SNR falls within the protection scope of this invention.

[0172] For example, continuing from the previous example, the signal-to-noise ratio (SNR) estimation unit calculates that the SNR estimate of the target reflected signal component of member A's terminal within the most recent approximately 100ms is approximately 15dB, which falls within the medium SNR range. The adaptive threshold selection unit then selects a subset of the target threshold range corresponding to the medium SNR. Member A's Doppler extension parameter B_d is approximately 15Hz, falling within the walking state threshold range [approximately 8Hz, approximately 40Hz] of this subset, corresponding to the motion state label "walking". Therefore, member A's motion state type is determined to be "walking".

[0173] In this embodiment, by introducing a dynamic threshold selection mechanism with adaptive signal-to-noise ratio, the decision boundary of motion state classification is correlated with signal quality, which effectively compensates for the noise broadening effect under low signal-to-noise ratio conditions, solves the problem of the decrease in classification accuracy of fixed threshold classifier when the signal-to-noise ratio changes, and significantly improves the robustness and accuracy of motion state classification in complex outdoor electromagnetic environments.

[0174] Example 7: Multi-node collaborative consistency verification and fusion method.

[0175] In this embodiment, before generating the communication information collection record corresponding to the target terminal based on the outdoor status information, the method further includes:

[0176] Obtain the collection timestamp and node location identifier synchronously recorded by the network-side node when collecting the outdoor status information;

[0177] Acquire reference outdoor status information independently collected by at least one other network-side node for the target terminal, wherein the reference outdoor status information carries a corresponding collection timestamp and node location identifier;

[0178] Based on the collection timestamp and the node location identifier corresponding to each network side node, the outdoor status information and the reference outdoor status information are spatiotemporally aligned to obtain an aligned status information set in a unified spatiotemporal coordinate system.

[0179] Calculate the Mahalanobis distance between any two states in the alignment state information set, mark the states whose Mahalanobis distance exceeds a preset Mahalanobis distance threshold as abnormal states, and remove the abnormal states from the alignment state information set.

[0180] For each state information in the aligned state information set after removing abnormal state information, a weighted average is performed using the signal-to-noise ratio weighting coefficient corresponding to each state information to obtain the fused outdoor state information, and the fused outdoor state information is used as the outdoor state information for generating the communication information collection record.

[0181] in:

[0182] Acquisition timestamp: This refers to the time stamp recorded by the internal clock module of a network-side node when it completes an outdoor status information acquisition. The accuracy of the acquisition timestamp directly determines the accuracy of subsequent spatiotemporal alignment. Therefore, it is required that the clocks of each network-side node be synchronized via NTP or PTP protocols, with clock deviation controlled within ±1ms. The acquisition timestamp corresponds one-to-one with each piece of outdoor status information, serving as its time dimension attribute.

[0183] Node location identifier: This refers to the geographic coordinates of a network node after deployment, measured and recorded by its built-in positioning module (such as a GNSS receiver). It is typically expressed as longitude, latitude, and altitude. The node location identifier is used as a coordinate transformation reference when spatially aligning information from multiple nodes.

[0184] Reference outdoor status information refers to outdoor status information obtained by other network-side nodes, different from the current network-side node, independently executing procedures S11-S13 to collect data from the same target terminal. Reference outdoor status information carries the collection timestamp and node location identifier of its collecting node to ensure spatiotemporal alignment with the outdoor status information of the current node. Methods for obtaining reference outdoor status information include, but are not limited to: direct interaction via wired or wireless backhaul links between nodes, or relaying through the core network / application server.

[0185] Signal-to-noise ratio (SNR) weighting coefficient: This refers to the weighting coefficient assigned to each piece of outdoor status information, which is positively correlated with its corresponding SNR estimate. It is used for weighted average fusion, and its calculation method is as follows:

[0186] w_i=SNR_i / ∑SNR_j;

[0187] Where SNR_i is the estimated signal-to-noise ratio (SNR) of the i-th outdoor state information (linear value, not dB value), and ∑SNR_j is the sum of the SNRs of all outdoor state information participating in the fusion. The physical meaning of this weighting method is: the higher the SNR of the acquisition node, the more reliable its outdoor state information, and it should occupy a larger weight in the fusion result.

[0188] The specific implementation of this embodiment is as follows:

[0189] First, the data aggregation unit of the network-side node acquires the collection timestamp and node location identifier synchronously recorded by the node when collecting outdoor status information. Simultaneously, through the inter-node communication link, it sends a status information request to at least one other network-side node and receives the reference outdoor status information returned by that node, along with its collection timestamp and node location identifier. The suggested value range for the number of reference outdoor status information entries, M, is 1 to 5, to balance fusion gain and communication overhead.

[0190] Secondly, the spatiotemporal alignment unit unifies all outdoor status information into the same spatiotemporal coordinate system based on the acquisition timestamps and node location identifiers carried by each outdoor status information. Specifically, the operation is as follows: Time alignment—using the acquisition timestamp of the current node as the reference time, and based on the deviation Δt between the acquisition timestamps of each reference outdoor status information and the reference time, the state value of each reference outdoor status information at the reference time is predicted. The prediction can use a uniform motion model: the predicted position (radial distance) d_pred = d_meas + v_meas·Δt, where d_meas and v_meas are the radial distance and radial velocity in the reference outdoor status information, respectively. Spatial alignment—the radial distance in each reference outdoor status information is converted into an equivalent radial distance with the current node as the reference point, based on the spatial geometric relationship between each node and the target terminal, or directly converted into absolute position coordinates based on a common geographic coordinate system. The specific calculation of spatial alignment depends on the known positions of each node and the coarse azimuth angle of the target terminal.

[0191] Then, the consistency verification unit calculates the Mahalanobis distance between any two outdoor state information pieces in the aligned state information set. The formula for calculating the Mahalanobis distance is as follows:

[0192] D_M(x_i,x_j)=√[(x_i-x_j)ᵀ·S⁻¹·(x_i-x_j)];

[0193] Where x_i and x_j are two outdoor state information vectors (containing dimensions such as radial distance and radial velocity), S is the covariance matrix of the outdoor state information vectors, and S⁻¹ is the inverse matrix of S. The covariance matrix S can be obtained through statistical analysis of historical data or by modeling based on the measurement accuracy parameters of each node. Compared with Euclidean distance, Mahalanobis distance considers the correlation and scale differences between various outdoor state dimensions, and can more reasonably measure the similarity of multidimensional outdoor states. The typical range of the preset Mahalanobis distance threshold is 2.0 to 4.0, and the preferred value is 3.0, corresponding to a confidence interval of approximately 99.7% (if the state follows a normal distribution). State information whose Mahalanobis distance exceeds the threshold is marked as anomaly and removed.

[0194] Then, the weighted fusion unit performs weighted average fusion on the remaining state information after removing anomalies. The weighting coefficients are calculated as described above, and can be the ratio of the linear values ​​of the signal-to-noise ratio. The fusion formula is as follows:

[0195] d_fusion=∑w_i·d_i, v_fusion=∑w_i·v_i;

[0196] Where d_i and v_i are the radial distance and radial velocity of the i-th outdoor state information, respectively, and w_i is the normalized signal-to-noise ratio weighting coefficient. For the discrete variable of motion state type, the fusion result can be determined by majority voting or weighted voting.

[0197] Finally, the fused outdoor status information is used as the final status information for generating communication information collection records, replacing the original single-node collection results.

[0198] It should be noted that the above consistency verification and fusion methods are not limited to Mahalanobis distance plus signal-to-noise ratio weighting. In one example, a sequential fusion method based on Kalman filtering can be used, where the outdoor state information of each network-side node is input as a measurement value into the Kalman filter, and the filter automatically completes time alignment, anomaly detection, and optimal weighted fusion. In another example, a batch processing fusion method based on graph optimization can be used, constructing a factor graph with the state at each time step as nodes and the state information as constraint edges, and solving for the optimal state trajectory through nonlinear least squares optimization, which is suitable for post-event precision analysis scenarios. Any technical means that can perform consistency verification and fusion of outdoor state information of multiple network-side nodes is within the protection scope of this invention.

[0199] For example, following the previous example, after the emergency communication vehicle (network-side node A, hereinafter referred to as node A) collects the outdoor status information of member A's target terminal, it requests reference outdoor status information from the UAV relay platform (network-side node B, hereinafter referred to as node B) deployed on a nearby ridge via a wireless link. Node B returns the reference outdoor status information of member A collected approximately 10ms earlier: radial distance approximately 310m, radial velocity approximately -1.2m / s, motion state type "walking", collection timestamp T3-10ms, and node position N30.12400° E120.67950°. The spatiotemporal alignment unit aligns the reference collection time of node B to time T3, predicting the radial distance = 310 + (-1.2) × 0.01 ≈ 309.99m. The consistency verification unit calculates the Mahalanobis distance between the two node status information to be approximately 1.5, which is less than the threshold of 3.0, and no anomaly is detected. The weighted fusion unit calculates weights based on the signal-to-noise ratios of the two nodes (approximately 15 dB for node A and 12 dB for node B, with linear values ​​of approximately 31.6 and 15.8 respectively). Node A has a weight of approximately 0.67, and node B has a weight of approximately 0.33. After fusion, the radial distance is approximately 0.67 × 300 + 0.33 × 310 ≈ 303.3 m, and the radial velocity is approximately 0.67 × (-1.3) + 0.33 × (-1.2) ≈ -1.27 m / s. The fusion result replaces the original acquisition result of node A and is used to generate the communication information acquisition record.

[0200] In this embodiment, by introducing multi-node collaborative acquisition, spatiotemporal alignment, Mahalanobis distance consistency verification, and signal-to-noise ratio weighted fusion mechanism, the independent acquisition of a single node is transformed into collaborative acquisition with multi-node cross-verification. This solves the problem of insufficient reliability of outdoor status information caused by single-node measurement noise or local interference, and significantly improves the accuracy and reliability of outdoor status information.

[0201] Example 8: Method for detecting and suppressing adversarial interference.

[0202] In this embodiment, after receiving the echo signal formed by the downlink sensing signal reflected by at least one target terminal, the method further includes:

[0203] Perform autocorrelation calculation on the echo signal to obtain the autocorrelation function curve;

[0204] The peak values ​​in the autocorrelation function curve whose amplitudes exceed a preset noise floor threshold are extracted as abnormal correlation peaks;

[0205] When the abnormal correlation peak is detected, the time delay-Doppler feature corresponding to the abnormal correlation peak is obtained, and the time delay-Doppler feature is matched with the time delay-Doppler feature of known interference modes stored in the preset fake signal feature library. The preset fake signal feature library is constructed by pre-collected and labeled adversarial interference samples.

[0206] If a match is successful, the type of interference corresponding to the known interference pattern is determined, and the target suppression strategy corresponding to the interference type is called from the preset set of interference suppression strategies.

[0207] Using the target suppression strategy, the signal component corresponding to the abnormal correlation peak in the echo signal is subjected to notch filtering in the time-delay-Doppler domain to obtain a purified echo signal, and the purified echo signal is used as the echo signal for separating the target reflection signal component.

[0208] in:

[0209] Autocorrelation function curve: This refers to the curve obtained by performing autocorrelation calculation on the echo signal, with time delay as the horizontal axis and autocorrelation amplitude as the vertical axis. The mathematical definition of autocorrelation calculation can be:

[0210] R_x(τ)=E[x(t)·x*(t-τ)];

[0211] Where t represents the time variable, x(t) is the echo signal, τ is the time delay, x*(t-τ) represents the complex conjugate of x(t-τ), and E[·] represents the statistical average of the random variables within the brackets. For periodic spoofing interference signals (such as periodic pulses emitted by deployed deceptive jammers), their autocorrelation function curve will exhibit abnormal peaks at the repetition period of the interference pulse and its integer multiples. The general calculation method for the autocorrelation function curve is a well-known technique in this field and will not be elaborated here.

[0212] The preset noise floor threshold is an amplitude threshold used to distinguish between normal noise fluctuations and abnormal correlation peaks. It is determined by collecting environmental echo signal samples under interference-free conditions, calculating the mean and standard deviation of their autocorrelation functions, and setting the threshold to the mean plus K times the standard deviation. A typical value for K is 5 to 10, with an optimal value of 6, to ensure that the false alarm probability is below 10⁻ under normal noise conditions. 6 .

[0213] The pre-defined pseudo signal feature library refers to a database that is pre-constructed by collecting and labeling adversarial interference samples and stores the feature parameters of various known interference modes in the time delay-Doppler domain. The construction process will be detailed in the specific implementation of this embodiment.

[0214] Delay-Doppler domain notch filtering: refers to the signal processing operation that filters out the interference component by setting the amplitude to zero or multiplying it by a very small suppression coefficient in the region (i.e., the delay-Doppler unit) where the signal component corresponding to the abnormal correlation peak is located on the delay-Doppler two-dimensional plane.

[0215] The specific implementation of this embodiment is as follows:

[0216] First, before the received echo signal enters the separation process of S12, the interference detection unit of the network-side node performs real-time autocorrelation calculation on the echo signal. The autocorrelation calculation can be implemented by Fast Fourier Transform (FFT): the signal x(t) is transformed to the frequency domain by FFT to obtain X(f), the power spectrum |X(f)|² is calculated, and then the autocorrelation function R_x(τ) is obtained by IFFT.

[0217] Secondly, the interference detection unit extracts peak values ​​on the autocorrelation function curve whose amplitudes exceed a preset noise floor threshold as abnormal correlation peaks. The extraction process includes: searching for peaks in the non-zero delay region of the autocorrelation function curve; the lowest peak height for peak determination is the noise floor threshold, and the lowest peak spacing is the delay corresponding to the distance resolution of the sensed signal. For each detected abnormal correlation peak, its delay position τ_int and peak amplitude A_int are recorded.

[0218] Then, upon detecting an abnormal correlation peak, the interference identification unit extracts a signal segment from the echo signal corresponding to the time delay interval, centered on the time delay position τ_int of the abnormal correlation peak. It performs a time-delay-Doppler two-dimensional Fourier transform on this segment (same as in Example 5), extracts its two-dimensional time-delay-Doppler feature map, and matches this feature map with the time-delay-Doppler features of known interference modes stored in a preset spoofing signal feature library. Matching can be performed using two-dimensional cross-correlation or feature point matching methods. A successful match is determined when the matching degree exceeds a preset threshold (e.g., 0.7). The preset spoofing signal feature library can be constructed as follows: before system deployment or during operation, signal samples emitted by various known adversarial interference devices (such as deceptive repeaters, suppression noise jammers, etc.) are collected. The interference type is manually or semi-automatically labeled, and the time-delay-Doppler features of each sample are extracted and stored in the preset spoofing signal feature library.

[0219] Next, if a match is successful, the interference suppression unit calls the target suppression strategy corresponding to the matched interference type from a preset set of interference suppression strategies. The preset set of interference suppression strategies includes at least the following strategies: a time-delay-Doppler domain notch filtering strategy for spoofing forwarding interference, an adaptive filtering strategy for suppressing noise interference, and a frequency domain notch filtering strategy for frequency sweeping interference.

[0220] Finally, the interference suppression unit employs a targeted suppression strategy to process the echo signal. Taking time-delay-Doppler domain notch filtering as an example: the echo signal is transformed to the time-delay-Doppler domain, and a notch filter (i.e., amplitude multiplied by a suppression coefficient approaching 0) is applied to the time-delay-Doppler unit corresponding to the anomalous correlation peak. Then, it is inversely transformed back to the time domain to obtain the purified echo signal. The purified echo signal replaces the original echo signal and enters the subsequent separation process in S12.

[0221] It should be noted that the aforementioned interference detection and suppression methods are not limited to autocorrelation anomaly detection plus time-delay-Doppler domain notch filtering. In one example, a spatial interference suppression method based on adaptive beamforming can be used, which suppresses the interference signal by forming nulls in the interference direction, suitable for scenarios where the direction of the interference source and the direction of the target terminal can be distinguished. In another example, an interference cancellation method based on blind source separation can be used, which separates and removes the interference signal as an independent source from the mixed signal, without needing to know the interference characteristics, suitable for scenarios where the type of interference is unknown. Any technical means that can detect and suppress the anti-interference component in the echo signal is within the protection scope of this invention.

[0222] For example, following the previous example, in a field exploration scenario, a deceptive jammer deployed by a non-cooperative party may exist in a certain area. This jammer periodically emits spoofed signals similar to downlink sensing signals, intending to disrupt normal monitoring of member terminals. After performing autocorrelation calculations on the echo signal, the network-side node detects an abnormal correlation peak at a time delay of approximately 15 μs (corresponding to a distance of approximately 2250 m between the jammer and the node). The peak amplitude exceeds the noise floor threshold by approximately 12 dB. The time-delay-Doppler feature corresponding to this abnormal correlation peak is extracted, and its matching degree with the "periodic pulse deceptive interference" pattern in the spoofed signal feature library is approximately 0.92, indicating a successful match. The interference suppression unit invokes the "time-delay-Doppler domain notch filtering" strategy, applying approximately -60 dB of suppression notch filtering to the time-delay-Doppler unit where the abnormal peak is located, eliminating the interference component. In the purified echo signal, the true signal component of member terminal A at a time delay of approximately 2.0 μs is clearly presented.

[0223] In this embodiment, by introducing a complete adversarial defense chain of autocorrelation anomaly detection, feature library matching and identification, and time-delay-Doppler domain notch suppression, the passive sensing system in traditional technology that is "undefended" is upgraded to a robust sensing system with active defense capabilities. This solves the problem of false alarms and missed detections in target detection caused by malicious spoofing signals, and significantly improves the survivability and reliability of the system in non-cooperative electromagnetic environments.

[0224] Example 9: Closed-loop adaptive update method based on terminal reported reference.

[0225] In this embodiment, the communication information collection record includes a corresponding collection timestamp; after generating the communication information collection record corresponding to the target terminal, it further includes:

[0226] The target terminal actively reports outdoor status information via cellular network or satellite link within a preset reference time window. The reported outdoor status information carries a corresponding reporting timestamp and is sent by the target terminal after the target terminal restores its network connection.

[0227] Based on the collection timestamp and the reporting timestamp, a correspondence is established between the outdoor status information with time deviations within a preset matching threshold range and the reported outdoor status information.

[0228] The outdoor status information with the established correspondence is compared with the reported outdoor status information, and the deviation value of the outdoor status information relative to the reported outdoor status information is calculated.

[0229] Based on the deviation value, calculate the gradient correction amount of the feature extraction parameters used for the physical layer feature extraction and / or the mapping model parameters used for the mapping;

[0230] Based on the gradient correction amount, a preset learning rate is used to update the feature extraction parameters and / or the mapping model parameters using stochastic gradient descent, so that the deviation between the updated outdoor state information and the corresponding reported outdoor state information approaches the preset deviation convergence target.

[0231] in:

[0232] Reporting timestamp: This refers to the time stamp recorded by the target terminal's internal clock module and sent along with the outdoor status information when the target terminal actively reports its outdoor status information. Since the target terminal reports after the network connection is restored, there may be a certain delay between the reporting timestamp and the actual time of the status event. However, its relative relationship with the timestamp collected by the network-side nodes can be used for time matching.

[0233] Reported outdoor status information: This refers to the historical status information that a target terminal actively sends to the network side via cellular networks (such as 4G LTE, 5G NR) or satellite links (such as BeiDou short message service, Iridium SBD) after reconnecting to the communication network. This information is measured by the target terminal's own positioning module (such as a GNSS receiver). Compared to outdoor status information collected by network-side nodes, reported outdoor status information typically has higher absolute accuracy (e.g., GNSS positioning accuracy can reach meter-level), but its acquisition is delayed and energy-intensive. Therefore, reported outdoor status information can serve as a "real-world" reference for offline training and parameter updates.

[0234] Deviation value: This refers to the measure of the difference between the outdoor status information collected by network-side nodes and the outdoor status information reported by the target terminal. For continuous variables (such as radial distance and radial velocity), the deviation value can be expressed as mean squared error (MSE) or its component form; for discrete variables (such as motion state type), the deviation value can be expressed as 0-1 loss or cross-entropy loss. The total deviation value is the weighted sum of the deviation values ​​of each dimension, and the weighting coefficients can be preset.

[0235] The specific implementation of this embodiment is as follows:

[0236] First, after the target terminal restores its network connection, the network-side node receives the outdoor status information actively reported by the target terminal within a preset reference time window via the backhaul network. The typical range of the preset reference time window is 1 to 30 minutes, with a preferred value of 5 minutes. The window length must cover a sufficient number of sample pairs to ensure the statistical stability of parameter updates. The reported outdoor status information may include the terminal's GNSS positioning data, the terminal's motion sensor data, and the corresponding reporting timestamp.

[0237] Secondly, the time matching unit establishes a time correspondence between the collection timestamp in the communication information collection record and the reporting timestamp carried in the reported outdoor status information. The matching rule can be as follows: for each reported outdoor status information, the unit searches for records in the communication information collection record where the absolute value of the difference between the collection timestamp and the reporting timestamp is less than a preset matching threshold (typically 1s to 5s, preferably 2s). If a match is found, a pairing is established; unpaired records are discarded.

[0238] Then, the deviation calculation unit calculates the deviation value of the outdoor status information relative to the reported outdoor status information for each paired record. Wherein:

[0239] The formula for calculating radial distance deviation is: e_d=(d_meas-d_ref)², where d_meas is the radial distance in the outdoor status information and d_ref is the radial distance in the reported outdoor status information;

[0240] The formula for calculating radial velocity deviation is: e_v=(v_meas-v_ref)², where v_meas is the radial velocity in the outdoor status information and v_ref is the radial velocity in the reported outdoor status information.

[0241] The formula for calculating the motion state type deviation can be: e_type = 0 (if the classification is consistent) or 1 (if the classification is inconsistent);

[0242] The total deviation E = ∑(α·e_d + β·e_v + γ·e_type), where α, β, and γ are preset deviation weights for each dimension, with a typical value of 1 / 3.

[0243] Next, the gradient calculation unit calculates the gradients of the feature extraction parameters θ_feat and the mapping model parameters θ_map based on the total bias E. Gradient calculation can be performed using automatic differentiation tools (such as the computation graph engine of a deep learning framework) or by manually deriving the gradient formula. For non-differentiable discrete operations (such as threshold comparison in CFAR detection), the gradient can be approximated using a pass-through estimator (STE) or a policy gradient method from reinforcement learning.

[0244] Then, the parameter update unit updates the parameters using the stochastic gradient descent (SGD) algorithm or its variants (such as Adam, RMSprop). The update formula is: θ←θ-η·∇_θE, where η is the preset learning rate, typically ranging from 10⁻⁻⁶. 4 The optimal value is 10⁻³, which can be determined through hyperparameter search based on convergence speed and stability; ∇_θE represents the gradient. The updated parameters replace the original parameters and are applied to subsequent physical layer feature extraction and state mapping processes.

[0245] Finally, after several consecutive update cycles, the parameter update unit evaluates whether the total deviation E approaches the preset deviation convergence target. A typical value for the deviation convergence target is 10% to 50% of the initial deviation value, with a preferred value of 20%. If convergence is achieved, the learning rate can be reduced or updates can be paused to lock the parameters.

[0246] It should be noted that the parameter update methods described above are not limited to stochastic gradient descent. In one example, mini-batch gradient descent (MBGD) can be used, accumulating the gradients of multiple paired samples before updating the parameters, suitable for scenarios where reported data arrives infrequently. In another example, an online passive-aggressive learning approach can be used, triggering parameter updates only when the deviation exceeds a preset tolerance threshold, avoiding frequent updates due to minor fluctuations. Any technique that can adaptively optimize the collected model parameters using reference state information reported by the terminal falls within the scope of protection of this invention.

[0247] For example, following the previous example, after member A's target terminal leaves an area without network coverage and restores its cellular network connection, it automatically reports its GNSS positioning history data for the past 5 minutes, including timestamps and latitude / longitude coordinates, to the network side. The time matching unit successfully matched approximately 30 pairs of records. The deviation calculation unit calculated the initial average radial distance deviation to be approximately 8.2m. The gradient calculation unit extracted the gradients of parameters (such as CFAR threshold factor and FFT points) and mapping model parameters (such as Doppler-velocity conversion coefficients) based on the deviation calculation features. The parameter update unit updated the parameters using SGD with a learning rate of approximately 0.001. After iterative updates using approximately 50 pairs of matched records, the average radial distance deviation converged to approximately 1.5m, approaching the convergence target of approximately 1.6m (approximately 20% of the initial deviation). The update stopped, and the updated parameters were locked for subsequent real-time status acquisition of member A.

[0248] In this embodiment, by using the high-precision GNSS status information actively reported by the terminal after reconnection as a reference truth, the closed-loop adaptive update of feature extraction parameters and mapping model parameters is driven, and a self-learning mechanism of "passive acquisition on the network side - active verification on the terminal side - closed-loop optimization of model parameters" is constructed. This solves the problem of performance degradation of fixed parameters of the acquisition model caused by environmental changes or individual differences of terminals, and enables the system to have the evolutionary ability to continuously self-optimize from the running data and gradually approach the optimal acquisition accuracy.

[0249] Example 10: An integrated example of combining multiple technical features.

[0250] The outdoor communication information collection methods described in the above embodiments can be recombined as needed to obtain combined implementation schemes, but all are within the protection scope claimed by this invention.

[0251] For example, combining the basic scheme of Embodiment 1 with the multi-node fusion scheme of Embodiment 7 and the closed-loop adaptive update scheme of Embodiment 9 can constitute an enhanced outdoor communication information acquisition system that possesses both the advantages of multi-node collaborative accuracy and continuous self-optimization capabilities. As another example, combining the basic scheme of Embodiment 1 with the adversarial interference defense scheme of Embodiment 8 can constitute a robust outdoor communication information acquisition system suitable for complex outdoor environments with malicious electromagnetic interference. Furthermore, combining the schemes of Embodiments 2, 3, and 4 can constitute a refined passive sensing front-end with high-precision identification signal separation capabilities and continuous terminal motion trajectory tracking capabilities. The above combinations are merely illustrative examples and do not constitute a limitation on the scope of protection of this invention.

[0252] Example 11: An outdoor communication information collection system.

[0253] In this embodiment, an outdoor communication information collection system is provided, applied to a network-side node. This outdoor communication information collection system corresponds one-to-one with the outdoor communication information collection method described in the previous embodiments. Please refer to [link / reference]. Figure 3 , Figure 3 This is a schematic block diagram of an outdoor communication information collection system provided in an embodiment of the present invention. Figure 3 As shown, the outdoor communication information acquisition system 30 includes a first broadcast module 31, a first separation module 32, a first extraction module 33, and a first generation module 34. The detailed descriptions of each functional module are as follows:

[0254] The first broadcast module 31 is used to broadcast downlink sensing signals, which are wireless signals with sensing functions that are not only used for data communication;

[0255] The first separation module 32 is used to receive the echo signal formed by the downlink sensing signal reflected by at least one target terminal, and to separate the target reflection signal component corresponding to the target terminal from the echo signal based on the inherent passive scattering characteristics of the target terminal.

[0256] The first extraction module 33 is used to extract physical layer features from the target reflected signal components and map the extracted features to the outdoor status information of the target terminal.

[0257] The first generation module 34 is used to generate a communication information collection record corresponding to the target terminal based on the outdoor status information, so as to complete the information collection.

[0258] Furthermore, the first separation module 32 includes:

[0259] The first acquisition submodule is used to acquire a preset passive scattering feature template corresponding to the target terminal. The preset passive scattering feature template includes a sequence of complex reflection coefficients of the radio frequency front end of the target terminal in a preset frequency band. Each complex reflection coefficient in the sequence includes an amplitude response value and a phase offset value.

[0260] The first operation submodule is used to perform conjugate multiplication of the complex received values ​​of the echo signal at each frequency point in the preset frequency band with the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and to perform inverse Fourier transform based on the conjugate multiplication results at each frequency point to obtain the frequency domain cross-correlation spectrum between the echo signal and the preset passive scattering feature template.

[0261] The first determination submodule is used to determine the peak position and peak amplitude of the frequency domain cross-correlation spectrum, and to determine the signal components within the time delay interval corresponding to the peak position where the peak amplitude exceeds a preset matching threshold as target reflection signal components.

[0262] Furthermore, the first acquisition submodule includes:

[0263] The first broadcast submodule is used to broadcast a preset frequency sweep excitation signal to the space where the target terminal is located when the target terminal is in a non-working state. The frequency range of the preset frequency sweep excitation signal covers the preset frequency band.

[0264] The first separation submodule is used to receive the reflection response signal generated by the target terminal under the illumination of the preset frequency sweep excitation signal, and to separate the scattering component related to the radio frequency front-end impedance characteristics of the target terminal from the reflection response signal;

[0265] The first construction submodule is used to perform complex demodulation on the scattering components, extract the amplitude response value and phase offset value at each frequency sweep point, construct the complex reflection coefficient sequence, and associate and store the complex reflection coefficient sequence with the terminal identifier of the target terminal to form the preset passive scattering feature template.

[0266] Furthermore, the first computational submodule includes:

[0267] The first division submodule is used to divide the echo signal into continuous time-domain signal frames according to a preset time frame length, and the adjacent time-domain signal frames have a preset ratio of overlap interval.

[0268] The first transformation submodule is used to perform a Fourier transform on each of the time-domain signal frames to obtain a frequency-domain receiving vector corresponding to each time-domain signal frame.

[0269] The first submodule is used to perform element-wise conjugate multiplication on each frequency domain receiving vector and the frequency domain template vector formed by the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and to perform inverse Fourier transform on the result of the element-wise conjugate multiplication to obtain the time delay domain cross-correlation spectrum corresponding to each time domain signal frame.

[0270] The first splicing submodule is used to splice the time-delay domain cross-correlation spectra corresponding to each time-domain signal frame in chronological order to form a time-delay-time two-dimensional cross-correlation distribution map, which is a frame-refined representation of the frequency domain cross-correlation spectrum.

[0271] Furthermore, the first extraction module 33 includes:

[0272] The second transformation submodule is used to perform a time-delay-Doppler two-dimensional Fourier transform on the target reflected signal component to obtain the time-delay-Doppler spectrum of the target reflected signal component;

[0273] The first extraction submodule is used to extract the main peak position, main peak width and Doppler expansion parameters corresponding to the target terminal from the time-delay Doppler spectrum;

[0274] The first determining submodule is used to determine the radial distance of the target terminal relative to the network-side node based on the main peak position, determine the radial velocity of the target terminal based on the time change rate of the main peak position, and determine the motion state type of the target terminal based on the Doppler expansion parameters.

[0275] The second determining submodule is used to use the radial distance, the radial velocity, and the motion state type as the outdoor state information of the target terminal.

[0276] Furthermore, the first determining submodule includes:

[0277] The second acquisition submodule is used to acquire the signal-to-noise ratio estimate of the target reflected signal component within a preset time period;

[0278] The first selection submodule is used to dynamically select a subset of target threshold intervals that match the signal-to-noise ratio estimate from a preset set of Doppler expansion parameter threshold intervals. The preset set of Doppler expansion parameter threshold intervals includes at least one preset static state threshold interval, at least one preset walking state threshold interval, and at least one preset fast movement state threshold interval. Each threshold interval in the preset set of Doppler expansion parameter threshold intervals has a preset one-to-one correspondence with a preset motion state label.

[0279] The first comparison submodule is used to compare the Doppler expansion parameter with the target threshold interval subset to determine the target threshold interval in which the Doppler expansion parameter falls.

[0280] The third determining submodule is used to determine the preset motion state label corresponding to the target threshold range according to the preset one-to-one correspondence, and to use the determined preset motion state label as the motion state type of the target terminal.

[0281] Furthermore, the outdoor communication information collection system 30 also includes:

[0282] The first acquisition module is used to acquire the collection timestamp and node location identifier synchronously recorded by the network-side node when collecting the outdoor status information;

[0283] The second acquisition module is used to acquire reference outdoor status information independently collected by at least one other network-side node on the target terminal, the reference outdoor status information carrying a corresponding collection timestamp and node location identifier.

[0284] The first alignment module is used to perform spatiotemporal alignment of the outdoor state information and the reference outdoor state information based on the collection timestamp and the node location identifier corresponding to each network side node, so as to obtain an alignment state information set under a unified spatiotemporal coordinate system.

[0285] The first elimination module is used to calculate the Mahalanobis distance between any two states in the alignment state information set, mark the states whose Mahalanobis distance exceeds a preset Mahalanobis distance threshold as abnormal states, and eliminate the abnormal states from the alignment state information set.

[0286] The first fusion module is used to perform a weighted average of each state information in the aligned state information set after removing abnormal state information, using the signal-to-noise ratio weighting coefficient corresponding to each state information, to obtain the fused outdoor state information, and to use the fused outdoor state information as the outdoor state information for generating the communication information collection record.

[0287] Furthermore, the outdoor communication information collection system 30 also includes:

[0288] The first calculation module is used to perform autocorrelation calculation on the echo signal to obtain the autocorrelation function curve;

[0289] The second extraction module is used to extract the peak values ​​in the autocorrelation function curve whose amplitude exceeds a preset noise floor threshold, as abnormal correlation peaks;

[0290] The first matching module is used to obtain the time-delay-Doppler feature corresponding to the abnormal correlation peak when the abnormal correlation peak is detected, and to match the time-delay-Doppler feature with the time-delay-Doppler feature of known interference modes stored in a preset fake signal feature library. The preset fake signal feature library is constructed by pre-collecting and labeling adversarial interference samples.

[0291] The first calling module is used to determine the interference type corresponding to the known interference pattern that was matched when the match is successful, and to call the target suppression strategy corresponding to the interference type from the preset interference suppression strategy set.

[0292] The first processing module is used to employ the target suppression strategy to perform notch filtering on the signal component corresponding to the abnormal correlation peak in the echo signal in the time-delay-Doppler domain to obtain a purified echo signal, and to use the purified echo signal as the echo signal for separating the target reflection signal component.

[0293] Furthermore, the communication information collection record includes a corresponding collection timestamp; the outdoor communication information collection system 30 also includes:

[0294] The third acquisition module is used to acquire the outdoor status information actively reported by the target terminal through a cellular network or satellite link within a preset reference time window. The outdoor status information carries a corresponding reporting timestamp and is sent by the target terminal after the target terminal restores its network connection.

[0295] The first establishment module is used to establish a correspondence between the outdoor status information with time deviations within a preset matching threshold range and the reported outdoor status information based on the collection timestamp and the reporting timestamp.

[0296] The first comparison module is used to compare the outdoor status information with the reported outdoor status information with the established correspondence, and to calculate the deviation value of the outdoor status information relative to the reported outdoor status information.

[0297] The first calculation module is used to calculate the gradient correction amount of the feature extraction parameters used for the physical layer feature extraction and / or the mapping model parameters used for the mapping, based on the deviation value.

[0298] The first update module is used to update the feature extraction parameters and / or the mapping model parameters by stochastic gradient descent based on the gradient correction amount and using a preset learning rate, so that the deviation between the updated outdoor state information and the corresponding reported outdoor state information approaches the preset deviation convergence target.

[0299] Specific limitations regarding outdoor communication information collection systems can be found in the above section on limitations of outdoor communication information collection methods, and will not be repeated here. Each module in the aforementioned outdoor communication information collection system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0300] Those skilled in the art will understand that the methods provided in the embodiments of the present invention can be implemented, in whole or in part, by software, hardware, firmware, or any combination thereof. The methods can also be implemented as a computer program product stored in one or more computer-readable storage media, including but not limited to: disks, optical disks, read-only memory (ROM), random access memory (RAM), flash memory, etc. When the computer program product is executed by one or more data processing devices (such as computers), the devices perform the steps as described in any of the foregoing method embodiments.

[0301] The software tools, components, or models not belonging to our company that appear in the embodiments of this invention are merely illustrative examples and do not represent actual use.

[0302] The data collection methods described in this invention comply with relevant laws and regulations and information security standards.

[0303] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An outdoor communication information collection method, applied to a network-side node, characterized in that, include: Broadcast downlink sensing signals, which are wireless signals with sensing capabilities that are not solely used for data communication; Receive the echo signal formed by the downlink sensing signal reflected by at least one target terminal, and based on the passive scattering characteristics inherent in the target terminal, separate the target reflection signal component corresponding to the target terminal from the echo signal; Physical layer features are extracted from the target reflected signal components, and the extracted features are mapped to the outdoor status information of the target terminal. Based on the outdoor status information, a communication information collection record corresponding to the target terminal is generated to complete the information collection.

2. The outdoor communication information collection method as described in claim 1, characterized in that, Based on the inherent passive scattering characteristics of the target terminal, the target reflected signal component corresponding to the target terminal is separated from the echo signal, including: Obtain a preset passive scattering feature template corresponding to the target terminal. The preset passive scattering feature template contains a sequence of complex reflection coefficients of the radio frequency front end of the target terminal in a preset frequency band. Each complex reflection coefficient in the sequence of complex reflection coefficients contains an amplitude response value and a phase offset value. The complex received values ​​of the echo signal at each frequency point within the preset frequency band are multiplied by the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and an inverse Fourier transform is performed based on the result of the conjugate multiplication at each frequency point to obtain the frequency domain cross-correlation spectrum between the echo signal and the preset passive scattering feature template. The peak position and peak amplitude of the frequency domain cross-correlation spectrum are determined, and the signal components within the time delay interval corresponding to the peak position where the peak amplitude exceeds a preset matching threshold are identified as target reflected signal components.

3. The outdoor communication information collection method as described in claim 2, characterized in that, Obtaining the preset passive scattering feature template corresponding to the target terminal includes: When the target terminal is in a non-working state, a preset frequency sweep excitation signal is broadcast to the space where the target terminal is located. The frequency range of the preset frequency sweep excitation signal covers the preset frequency band. The system receives the reflection response signal generated by the target terminal under the illumination of the preset frequency sweep excitation signal, and separates the scattering component related to the radio frequency front-end impedance characteristics of the target terminal from the reflection response signal. The scattering components are demodulated in complex form to extract the amplitude response value and phase offset value at each frequency sweep point, construct the complex reflection coefficient sequence, and associate the complex reflection coefficient sequence with the terminal identifier of the target terminal to form the preset passive scattering feature template.

4. The outdoor communication information collection method as described in claim 2, characterized in that, The complex received values ​​of the echo signal at each frequency point within the preset frequency band are multiplied by the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and then a conjugate multiplication is performed on each frequency point. Based on the conjugate multiplication results at each frequency point, an inverse Fourier transform is performed to obtain the frequency domain cross-correlation spectrum between the echo signal and the preset passive scattering feature template, including: The echo signal is divided into continuous time-domain signal frames according to a preset time frame length, and adjacent time-domain signal frames have a preset overlap interval. Perform a Fourier transform on each of the time-domain signal frames to obtain the frequency-domain receiving vector corresponding to each time-domain signal frame; Each frequency domain receiving vector is multiplied element-wise by the frequency domain template vector formed by the complex reflection coefficients at the corresponding frequency points in the preset passive scattering feature template, and the result of the element-wise conjugate multiplication is subjected to inverse Fourier transform to obtain the time delay domain cross-correlation spectrum corresponding to each time domain signal frame. The time-delay domain cross-correlation spectra corresponding to each time-domain signal frame are spliced ​​together in chronological order to form a time-delay-time two-dimensional cross-correlation distribution map, which is a frame-by-frame refined representation of the frequency domain cross-correlation spectrum.

5. The outdoor communication information collection method as described in claim 1, characterized in that, Physical layer feature extraction is performed on the target reflected signal components, and the extracted features are mapped to the outdoor state information of the target terminal, including: Perform a time-delay-Doppler two-dimensional Fourier transform on the target reflected signal component to obtain the time-delay-Doppler spectrum of the target reflected signal component; Extract the main peak position, main peak width, and Doppler expansion parameters corresponding to the target terminal from the time-delay-Doppler spectrum; The radial distance of the target terminal relative to the network-side node is determined based on the position of the main peak, the radial velocity of the target terminal is determined based on the time change rate of the main peak position, and the motion state type of the target terminal is determined based on the Doppler expansion parameters. The radial distance, the radial velocity, and the motion state type are used as the outdoor state information of the target terminal.

6. The outdoor communication information collection method as described in claim 5, characterized in that, Determining the motion state type of the target terminal based on the Doppler extension parameters includes: Obtain the signal-to-noise ratio estimate of the target reflected signal component within a preset time period; Based on the signal-to-noise ratio (SNR) estimate, a subset of target threshold intervals that matches the SNR estimate is dynamically selected from a preset set of Doppler expansion parameter threshold intervals. The preset set of Doppler expansion parameter threshold intervals includes at least one preset static state threshold interval, at least one preset walking state threshold interval, and at least one preset fast movement state threshold interval. Each threshold interval in the preset set of Doppler expansion parameter threshold intervals has a preset one-to-one correspondence with a preset motion state label. The Doppler expansion parameter is compared with the target threshold interval subset to determine the target threshold interval in which the Doppler expansion parameter falls. Based on the preset one-to-one correspondence, a preset motion state label corresponding to the target threshold range is determined, and the determined preset motion state label is used as the motion state type of the target terminal.

7. The outdoor communication information collection method as described in claim 1, characterized in that, Before generating the communication information collection record corresponding to the target terminal based on the outdoor status information, the method further includes: Obtain the collection timestamp and node location identifier synchronously recorded by the network-side node when collecting the outdoor status information; Acquire reference outdoor status information independently collected by at least one other network-side node for the target terminal, wherein the reference outdoor status information carries a corresponding collection timestamp and node location identifier; Based on the collection timestamp and the node location identifier corresponding to each network side node, the outdoor status information and the reference outdoor status information are spatiotemporally aligned to obtain an aligned status information set in a unified spatiotemporal coordinate system. Calculate the Mahalanobis distance between any two states in the alignment state information set, mark the states whose Mahalanobis distance exceeds a preset Mahalanobis distance threshold as abnormal states, and remove the abnormal states from the alignment state information set. For each state information in the aligned state information set after removing abnormal state information, a weighted average is performed using the signal-to-noise ratio weighting coefficient corresponding to each state information to obtain the fused outdoor state information, and the fused outdoor state information is used as the outdoor state information for generating the communication information collection record.

8. The outdoor communication information collection method as described in claim 1, characterized in that, After receiving the echo signal formed by the downlink sensing signal reflected by at least one target terminal, the method further includes: Perform autocorrelation calculation on the echo signal to obtain the autocorrelation function curve; The peak values ​​in the autocorrelation function curve whose amplitudes exceed a preset noise floor threshold are extracted as abnormal correlation peaks; When the abnormal correlation peak is detected, the time delay-Doppler feature corresponding to the abnormal correlation peak is obtained, and the time delay-Doppler feature is matched with the time delay-Doppler feature of known interference modes stored in the preset fake signal feature library. The preset fake signal feature library is constructed by pre-collected and labeled adversarial interference samples. If a match is successful, the type of interference corresponding to the known interference pattern is determined, and the target suppression strategy corresponding to the interference type is called from the preset set of interference suppression strategies. Using the target suppression strategy, the signal component corresponding to the abnormal correlation peak in the echo signal is subjected to notch filtering in the time-delay-Doppler domain to obtain a purified echo signal, and the purified echo signal is used as the echo signal for separating the target reflection signal component.

9. The outdoor communication information collection method as described in claim 1, characterized in that, The communication information collection record includes a corresponding collection timestamp; After generating the communication information collection record corresponding to the target terminal, the method further includes: The target terminal actively reports outdoor status information via cellular network or satellite link within a preset reference time window. The reported outdoor status information carries a corresponding reporting timestamp and is sent by the target terminal after the target terminal restores its network connection. Based on the collection timestamp and the reporting timestamp, a correspondence is established between the outdoor status information with time deviations within a preset matching threshold range and the reported outdoor status information; The outdoor status information with the established correspondence is compared with the reported outdoor status information, and the deviation value of the outdoor status information relative to the reported outdoor status information is calculated. Based on the deviation value, calculate the gradient correction amount of the feature extraction parameters used for the physical layer feature extraction and / or the mapping model parameters used for the mapping; Based on the gradient correction amount, a preset learning rate is used to update the feature extraction parameters and / or the mapping model parameters using stochastic gradient descent, so that the deviation between the updated outdoor state information and the corresponding reported outdoor state information approaches the preset deviation convergence target.

10. An outdoor communication information acquisition system, applied to a network-side node, characterized in that, include: The first broadcast module is used to broadcast downlink sensing signals, which are wireless signals with sensing functions that are not only used for data communication; The first separation module is used to receive the echo signal formed by the downlink sensing signal reflected by at least one target terminal, and to separate the target reflection signal component corresponding to the target terminal from the echo signal based on the inherent passive scattering characteristics of the target terminal. The first extraction module is used to extract physical layer features from the target reflected signal components and map the extracted features to the outdoor status information of the target terminal. The first generation module is used to generate a communication information collection record corresponding to the target terminal based on the outdoor status information, so as to complete the information collection.

Images