Emergency response system based on intelligent communication network

Abstract

The application discloses a distress emergency response system based on an intelligent communication network, and particularly relates to the field of intelligent communication, and comprises a multi-source data acquisition layer, a preprocessing and feature extraction layer, a progressive analysis engine, a decision fusion center and an execution control layer. The application significantly improves data integrity and anti-interference capability, a space-time reference unified module adopts a PTP protocol and a coordinate system conversion technology, realizes accurate alignment of multi-source data, and generates a multi-dimensional feature matrix through time-frequency joint analysis, solves the problem of single dimension of feature extraction in traditional systems, the progressive analysis engine is based on dynamic fusion of transmission robustness index, spatial confidence and environmental risk value, triggers a hierarchical response strategy, combines terrain refraction compensation and time effectiveness attenuation mechanism to optimize decision logic, breaks through the limitation of a fixed threshold response mode, adapts to communication quality fluctuation and environmental mutation, reduces rescue delay risk, and improves resource scheduling accuracy and system intelligent level.

Description

Technical Field

[0001] This invention relates to the field of intelligent communication technology, and more specifically, to an emergency call response system based on an intelligent communication network. Background Technology

[0002] Traditional emergency response systems typically rely on a single communication network or positioning technology to receive and process distress signals. For example, they may use cellular networks to obtain user location information and trigger rescue procedures, or combine GPS positioning with voice communication to complete a basic response. The operational process of existing technologies mainly includes receiving distress signals, parsing location information, and manually dispatching rescue resources. Its core relies on preset fixed thresholds to determine the risk level, lacking real-time adaptability to dynamic environments and communication status.

[0003] However, existing technologies have significant shortcomings in practical applications. First, data acquisition is limited to a single dimension, relying solely on communication signals or location information, making it difficult to cope with signal interference, location drift, and sudden weather changes in complex environments. Second, the lack of unified spatiotemporal benchmarks for multi-source data leads to difficulties in data fusion, inefficient preprocessing, and the inability to generate effective joint feature vectors. Furthermore, existing systems lack dynamic risk assessment capabilities and cannot optimize response strategies through environmental coupling analysis and real-time feedback mechanisms, resulting in frequent delays in rescue efforts or resource misallocation. Summary of the Invention

[0004] To overcome the aforementioned deficiencies of the prior art, this invention provides an emergency call response system based on an intelligent communication network. The system addresses the problems mentioned in the background art, such as reliance on a single data source, inconsistent spatiotemporal references leading to difficulties in multi-source fusion, lack of dynamic environmental risk assessment capabilities, low preprocessing efficiency, susceptibility to signal interference and positioning drift, and high risk of delayed rescue response.

[0005] To achieve the above objectives, the present invention provides the following technical solution: an emergency call response system based on an intelligent communication network, comprising:

[0006] Multi-source data acquisition layer: Real-time acquisition of three major categories of raw data: communication network, spatial positioning and environmental perception, including channel quality indicators of communication parameters, motion state data of positioning parameters and meteorological and terrain information of environmental parameters, forming standardized data packets containing timestamps, device identifiers and acquired values;

[0007] Preprocessing and feature extraction layer: Cleans and filters the raw data, unifies the spatiotemporal reference and performs primary feature calculations, eliminates sensor noise and solves the problem of spatiotemporal inconsistency of multi-source data, generates channel coherence coefficient, motion distortion factor and disaster propagation coefficient by calculation, and outputs a standardized feature vector with unified dimension.

[0008] The progressive analysis engine includes a communication health analysis unit, a spatial confidence analysis unit, and an environmental coupling analysis unit. It first assesses the communication channel quality and determines whether an emergency channel handover is triggered. When communication is normal, it sequentially verifies the positioning accuracy and models the environmental risk. Finally, it outputs three types of analysis results: channel status label, spatial confidence score, and environmental risk level.

[0009] Decision Fusion Center: Integrates communication health, spatial credibility, and environmental risk data through nonlinear fusion algorithms, calculates a comprehensive emergency index, matches preset response strategies, and generates tiered response instructions that include voice confirmation, video verification, and drone rescue.

[0010] The execution control layer consists of three execution units: a multi-mode communication switching controller, a resource scheduler, and a multi-mode alarm generator. These units perform physical operations based on decision instructions, while simultaneously monitoring the execution status and feeding it back to the system data.

[0011] Preferably, the communication parameters include signal-to-noise ratio, MAC layer retransmission rate, and channel impulse response; the positioning parameters include Doppler frequency shift, elevation difference, and number of visible satellites; and the environmental parameters include weather radar reflectivity, base station load rate, and terrain roughness.

[0012] Preferably, the communication data is collected in real time through a distributed sensor network. The terminal device has a built-in radio frequency front-end chipset that measures the instantaneous signal-to-noise ratio at 10ms intervals. An NI PXIe-5663 vector signal analyzer is used to capture IQ signals and simultaneously record the peak-to-average power ratio characteristics of the physical layer channel impulse response. The MAC layer retransmission rate data is obtained by parsing the Retry bit in the 802.11 protocol control frame. Probe devices are deployed on the base station side to collect the PDCP layer packet loss rate and RLC layer buffer status. All communication parameters are timestamped and then encapsulated into communication quality data frames conforming to the IEEE 1901.1 standard.

[0013] Preferably, the positioning data adopts a multi-source fusion acquisition architecture. The GNSS module uses a ublox ZED-F9P high-precision receiver to output raw observations at a frequency of 10Hz. The built-in MPU-9250 nine-axis inertial sensor captures three-dimensional acceleration and angular velocity at a sampling rate of 200Hz. The Doppler frequency shift value is obtained by calculating the carrier phase change rate. The elevation difference data is simultaneously acquired from the absolute altitude value of the barometer MS5611 and the GNSS elevation data. The RTK positioning results after differential correction are compared in real time with the OpenStreetMap elevation database. The number of visible satellites is dynamically updated by parsing the GSV statement of the NMEA-0183 protocol.

[0014] Preferably, the environmental data establishes a multi-dimensional sensing system. Meteorological data is accessed through the MODBUS-RTU interface of the local weather station WS-3000 to obtain temperature, humidity, and air pressure change rates in real time. Radar reflectivity data is obtained by subscribing to the meteorological department's API interface to obtain 2.5km gridded numerical forecast products. Base station load rate is converted from the ifInOctets counter of the core network management system SNMP agent. Terrain roughness is generated by scanning with RIEGL VZ-400 lidar to generate point cloud data. Elevation changes are monitored by the Bosch BMP388 barometric pressure sensor built into the mobile terminal, and the fractal dimension of the land surface is calculated by combining it with USGS 30-meter DEM data.

[0015] Preferably, the cleaning and filtering process performs anomaly detection on the original data stream based on a sliding window mechanism. In communication data, an improved Z-score algorithm is applied to identify outliers. In location data, abnormal location points with deviations exceeding 4σ are removed after calculating the three-dimensional spatial consistency using Mahalanobis distance. In environmental data, wavelet threshold denoising is used to eliminate impulse interference.

[0016] Preferably, the spatiotemporal reference is uniformly deployed with the PTP precise time protocol to achieve microsecond-level clock synchronization, the spatial reference transformation uses the WGS84 to ECEF coordinate system transformation matrix to unify all positioning data, and the non-uniform sampling data is resampled to a 10Hz reference frequency using cubic spline interpolation.

[0017] Preferably, the primary feature calculation generates multi-dimensional features through joint time-frequency analysis, including communication features, positioning features, and environmental features. Communication features include signal-to-noise ratio gradient, channel energy entropy, and MAC layer burst index. Positioning features include three-dimensional radius of curvature, elevation difference variance, Doppler diffusion factor, and satellite geometric accuracy factor. Environmental features include second derivative of air pressure, radar reflectivity slope, terrain roughness spectrum, base station load change rate, and electromagnetic interference correlation coefficient. Finally, a dimensionally regular feature matrix is ​​output to the progressive analysis engine.

[0018] Preferably, the communication features are constructed by deep analysis of the original signal-to-noise ratio (SNR), channel impulse response, and MAC layer retransmission rate data. First, a sliding window differential operation is performed on the SNR sequence to extract the signal quality change trend and generate an SNR gradient value characterizing communication stability. Then, wavelet packet transform technology is used to decompose the channel impulse response waveform, and the channel energy entropy index is constructed by calculating the Shannon entropy value of the energy distribution of each sub-band. Finally, a Poisson process model is established based on the temporal characteristics of the MAC layer retransmission rate, and the burst transmission anomaly intensity is quantified by the deviation between the actual observed value and the theoretical value to form the MAC layer burst index.

[0019] Preferably, the signal-to-noise ratio gradient value is specifically expressed as: G_{SNR}[n]: signal-to-noise ratio gradient value, SNR[n]: signal-to-noise ratio of the nth sampling point, T_s: sampling interval, N: sliding window length, k: the kth sampling position before the current calculation point; the channel energy entropy is specifically expressed as: E_CEP: Channel energy entropy, WPD_i: Wavelet packet decomposition coefficient of the i-th subband, M: Decomposition layer number; The MAC layer burst index is specifically expressed as: B_MAC: MAC layer burst index, λ_obs: observed MAC retransmission rate, λ_poi: Poisson distribution expectation.

[0020] Preferably, the positioning features originate from the spatiotemporal fusion analysis of raw data on Doppler frequency shift, elevation difference, and the number of visible satellites. First, the instantaneous radius of curvature of the three-dimensional motion trajectory is constructed by combining the second derivative of the Doppler frequency shift change rate and the elevation difference. The elevation difference variance index is generated by evaluating the intensity of elevation data fluctuation using the sliding window variance analysis method. The signal scattering characteristics are extracted by analyzing the autocorrelation function of the Doppler frequency shift sequence to form the Doppler diffusion factor. The accuracy attenuation factor is calculated based on the spatial geometric distribution of visible satellites.

[0021] Preferably, the three-dimensional radius of curvature is specifically expressed as: R_curv: three-dimensional radius of curvature, \vec{v}: velocity vector, \vec{a}: acceleration vector; the elevation difference variance is specifically expressed as: σ_h^2: elevation difference variance, Δh_k=h[k]-h[k-1]: difference between adjacent elevation measurements, K: calculation window size, μ_Δh: arithmetic mean of the elevation difference sequence; the Doppler diffusion factor is specifically expressed as: D_dop: Doppler spread factor, f_d[m]: m-th Doppler frequency shift measurement, τ: correlation delay; the satellite geometric accuracy factor is specifically expressed as: GDOP: Geometric precision factor for satellites, H: Geometric matrix of visible satellites, tr: Matrix trace operation, T: Matrix transpose operation.

[0022] Preferably, the environmental features integrate multimodal data of meteorological radar reflectivity, base station load rate, and terrain roughness; second-order difference processing is performed on the air pressure time series data to capture precursor signals of meteorological changes; linear regression analysis is used to analyze the trend of radar reflectivity changes to construct a disaster development slope index; fast Fourier transform is used to convert terrain roughness into frequency domain energy spectral density features; an exponentially weighted moving average model of base station load rate is established to quantify the dynamic changes of network resources; and mutual information analysis is used to reveal the coupling relationship between electromagnetic environment and terrain features.

[0023] Preferably, the second derivative of the air pressure is specifically expressed as: , ∇^{2}P: second derivative of air pressure, P[t]: air pressure measurement at time t, Δt: differential interval, t: discretized time index; the radar reflectivity slope is specifically expressed as: S_rad: radar reflectivity slope, Z_i: i-th radar reflectivity measurement, N_e: regression window size; the terrain roughness spectrum is specifically represented as: R_rongh: terrain roughness spectrum, X[k]: FFT coefficients of the terrain roughness sequence, K_e: Fourier transform length; the base station load change rate is specifically expressed as: ΔL: base station load change rate, L[t]: current base station load rate, α: forgetting factor; the electromagnetic interference correlation coefficient is specifically expressed as: C_emi: Electromagnetic interference correlation coefficient, L: Base station load rate sequence, R: Terrain roughness sequence, cov: Covariance operation, σ: Standard deviation.

[0024] Preferably, the channel coherence coefficient is generated by dynamically coupling the real-time acquired signal-to-noise ratio sequence with the channel impulse response characteristics. 250 signal-to-noise ratio sampling points are continuously acquired with a time window of 5 seconds. An evaluation model is constructed by combining the deviation of the current channel impulse response peak-to-average ratio with its 300-second sliding historical average. The noise basis correction in the denominator is dynamically adjusted according to the receiver sensitivity parameters. The numerator reflects the instantaneous signal quality. The final output value is normalized to characterize the channel stability.

[0025] Preferably, the channel coherence coefficient is specifically expressed as follows: b_1: Channel coherence coefficient, SNR_k: Signal-to-noise ratio at the kth sampling point, CIR_PAPR: Peak-to-average power ratio of channel impulse response, μ_CIR: Historical average CIR, γ_0: Noise basis correction, N_b: Number of sampling windows.

[0026] Preferably, the motion distortion factor is constructed by fusing Doppler frequency shift measurements and elevation change gradient data to form a two-dimensional motion feature vector. First, the elevation data sequence is subjected to time difference operation to obtain the vertical change rate, which is then combined with the horizontal Doppler frequency shift to form a composite motion vector. The degree of motion anomaly is then quantified by the ratio of the magnitude of this vector to the preset maximum motion speed.

[0027] Preferably, the motion distortion factor is specifically expressed as: b_3: motion distortion factor, f_d: absolute value of Doppler frequency shift, ∇h: elevation change gradient, ν_max: maximum motion velocity.

[0028] Preferably, the disaster propagation coefficient is constructed by integrating the time integral of meteorological radar reflectivity with the base station load status to build an environmental risk model. A 30-minute sliding window is used to numerically integrate the radar reflectivity data to assess the intensity of persistent meteorological threats. At the same time, an exponential correction term for the base station load rate is introduced to enhance the network congestion effect. The network topology factor in the exponent is calculated based on the Voronoi diagram density of the base station distribution. The base station density is converted into an influence coefficient of 0.2-0.8 through the sigmoid function. The final output value is a product that integrates both meteorological and network risk factors.

[0029] Preferably, the disaster propagation coefficient is specifically expressed as: Γ: disaster propagation coefficient, Z(τ): weather radar reflectivity, L: base station load rate, T: integration time window, δ: network topology factor.

[0030] Preferably, the progressive analysis engine adopts a three-level serial processing architecture. First, the communication health analysis unit receives the channel coherence coefficient and MAC layer retransmission rate data provided by the preprocessing layer. It calculates the transmission robustness index by fusing the exponential decay model and the sigmoid function. When the index is lower than the 0.4 threshold, it immediately triggers the multi-mode channel switching protocol and sends control commands to the execution layer. When the communication status is normal, the spatial reliability assessment unit is activated.

[0031] Preferably, the transmission robustness index is specifically expressed as: b_2: Transmission robustness index, λ: MAC layer retransmission rate attenuation coefficient, θ: Channel coherence threshold; When b_2 < 0.4: Activate channel switching protocol and send frequency hopping command to execution layer; When 0.4 ≤ b_2 < 0.6: Start bandwidth adaptive adjustment; When b_2 ≥ 0.6: Release spatial analysis enable signal.

[0032] Preferably, λ is obtained by constructing a Poisson regression model using historical operational data, and θ is specified in Section 6.3.2 of 3GPP TS 23.203 V16.2.0 standard.

[0033] Preferably, the spatial confidence assessment unit receives Doppler frequency shift and elevation difference data, constructs kinematic constraint equations including angular velocity vector and elevation gradient, calculates multi-source positioning dispersion and uses hyperbolic tangent function normalization to generate spatial confidence score, and starts environmental coupling analysis unit when the score exceeds 0.5 confidence threshold.

[0034] Preferably, the spatial reliability assessment unit constructs kinematic constraint equations when 0.4 ≤ b_2: , : 3D acceleration vector, ω: angular velocity vector, β: terrain coupling coefficient; Calculate multi-source positioning discreteness: σ_pos: Multi-source localization dispersion, N_σ: Number of effective localization sources, x_i, y_i: Planar coordinates provided by the i-th localization source, μ_x, μ_y: Arithmetic mean of coordinates of all effective localization sources; Output spatial confidence: C_s: spatial confidence, α: weighting coefficient of positioning dispersion, β_c: motion mutation penalty factor.

[0035] Preferably, the environmental coupling degree analysis unit integrates the time-series integral of meteorological radar reflectivity and the differential data of base station load change rate, and performs refraction compensation on the disaster propagation model by introducing terrain roughness spectrum, and finally outputs a comprehensive risk value weighted by communication quality. In this process, a dynamic feedback mechanism is established, and when the spatial confidence level is lower than 0.3 for three consecutive times, it is automatically upgraded to the manual handling process.

[0036] Preferably, the environmental coupling analysis unit initiates environmental analysis when C_s>0.5: Disaster propagation model: γ: Coupling coefficient; Terrain refraction compensation: Generate a comprehensive risk value: R: Comprehensive risk value, ε: Zero-prevention constant; When R>2.0, the MAC layer priority is automatically increased, and when C_s<0.3 for 3 consecutive times, the manual intervention protocol is triggered.

[0037] Preferably, the comprehensive emergency index is specifically expressed as follows: E: Comprehensive emergency index, γ_e: Timeliness decay coefficient.

[0038] Preferably, the comprehensive emergency index adopts a nonlinear fusion algorithm. First, the transmission robustness index output by the communication health analysis unit and the three-dimensional location reliability calculated by the spatial reliability assessment unit are normalized and multiplied to construct a basic guarantee factor. Then, the logarithmic attenuation operation is performed with the disaster propagation coefficient obtained by the environmental coupling degree analysis unit to amplify the sensitivity of high-risk scenarios. At the same time, a timeliness attenuation term is introduced to dynamically reduce the weight of historical data. γ_e is jointly determined by the system running timestamp difference Δt and the current network load status. Finally, the output value is compressed to the [0, 1] interval through the Sigmoid function to form a standardized emergency index.

[0039] Preferably, the decision fusion center initiates a first-level response when E∈[0, 0.3), initiates two-way voice verification through the VoLTE channel and activates base station-level fingerprint positioning; when E∈[0.3, 0.6), a second-level response is triggered, scheduling smart cameras within a 300-meter radius to perform multi-angle video verification, and simultaneously enabling carrier phase differential technology to improve positioning accuracy to sub-meter level; when E≥0.6, a third-level response is executed, automatically dispatching tethered drones to establish temporary communication relays, and linking with the 120 / 110 system to generate a joint rescue work order. A hysteresis band of 0.05 is introduced at the boundaries of each threshold interval to prevent state oscillation, and the threshold benchmark is dynamically calibrated every 8 hours based on the output of the environmental coupling calculator.

[0040] Preferably, the execution control layer adopts a hardware-in-the-loop architecture to implement the physical execution of decision commands. The multi-mode communication handover controller dynamically scans the channel quality of satellite / 5G / Mesh networks through a software-defined radio interface. Based on the target signal-to-noise ratio threshold value issued by the decision center, it selects the optimal communication link using an ε-greedy algorithm. During the handover process, the base station power amplifier gain is simultaneously adjusted to suppress transient interference.

[0041] Preferably, the resource scheduler has a built-in FPGA dynamic partition controller, which establishes a three-level pipeline structure based on task priority: the three-level response tasks are allocated 64-bit floating-point arithmetic units and dedicated DMA channels, the two-level response tasks use a shared computing array, the first-level response tasks are limited to execution in preemptible time slots, and memory access is optimized through AXI crossbar switches.

[0042] Preferably, the multimodal alarm generator integrates a PWM dimming circuit and a soundprint synthesis chip, and uses fractal time-frequency modulation technology to generate a spatially directional sound and light alarm signal. The light alarm is configured with a pulse mode in accordance with the CIE 62471 safety standard, and the sound alarm is constructed with a swept-frequency sine wave fundamental frequency according to the EN 54-3 standard. At the same time, a tactile feedback conforming to the impact spectrum characteristics of MIL-STD-810G is generated through a vibration motor drive circuit.

[0043] Preferably, the execution control layer maintains deterministic communication with the decision center via time-triggered Ethernet with a 10ms cycle. The execution status data is written into the dual-port RAM after CRC32 verification, so that the self-testing module of the preprocessing and feature extraction layer can perform reverse trace verification.

[0044] The technical effects and advantages of this invention are as follows:

[0045] 1. This invention significantly improves the data integrity and analysis accuracy of emergency response by constructing a multi-source data acquisition layer and preprocessing mechanism. The system integrates three major categories of data: communication network, spatial positioning, and environmental perception. It adopts a distributed sensor network and multi-source fusion architecture to capture multi-dimensional parameters such as signal-to-noise ratio, Doppler frequency shift, and radar reflectivity in real time. Data cleaning is achieved through techniques such as sliding window anomaly detection and Mahalanobis distance removal, solving the problems of single data source and severe noise interference in traditional systems, and providing a highly reliable standardized data stream for subsequent analysis.

[0046] 2. To address the shortcomings of inconsistent spatiotemporal references in multi-source data, this invention designs a spatiotemporal reference unification module, deploying the PTP precise time protocol and the WGS84 to ECEF coordinate system transformation matrix. Combined with cubic spline interpolation, it achieves resampling of non-uniform data, ensuring spatiotemporal alignment of communication, positioning, and environmental data. Furthermore, through joint time-frequency analysis and feature matrix generation, the system can dynamically extract key features such as channel energy entropy, three-dimensional radius of curvature, and disaster propagation coefficient, overcoming the limitation of traditional systems' single-dimensional feature extraction and significantly enhancing data analysis capabilities in complex scenarios.

[0047] 3. This invention achieves intelligent dynamic risk assessment and graded response through a progressive analysis engine and decision fusion center. The system dynamically triggers multi-level response strategies such as voice confirmation, video verification, and drone rescue based on the nonlinear fusion of transmission robustness index, spatial confidence score, and comprehensive risk value. It also introduces timeliness attenuation and terrain refraction compensation mechanisms to optimize decision logic. Compared with the traditional fixed threshold response mode, this invention can adapt to communication quality, environmental changes, and network load changes, effectively reducing the risk of rescue delay and improving the accuracy and timeliness of emergency resource scheduling. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0049] Figure 2 This is a schematic diagram of the multi-source data acquisition layer structure of the present invention.

[0050] Figure 3 This is a schematic diagram of the preprocessing and feature extraction layer structure of the present invention.

[0051] Figure 4 This is a schematic diagram of the progressive analysis engine structure of the present invention.

[0052] Figure 5 This is a schematic diagram of the decision fusion center structure of the present invention.

[0053] Figure 6 This is a schematic diagram of the execution control layer structure of the present invention. Detailed Implementation

[0054] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0055] refer to Figures 1-6 The emergency call response system based on intelligent communication networks shown includes:

[0056] Multi-source data acquisition layer: Real-time acquisition of three major categories of raw data: communication network, spatial positioning, and environmental perception. This includes channel quality indicators for communication parameters, motion state data for positioning parameters, and meteorological and terrain information for environmental parameters, forming standardized data packets containing timestamps, device identifiers, and acquired values.

[0057] The multi-source data acquisition layer serves as the system's data input end, directly connecting with the preprocessing and feature extraction layers. It provides the preprocessing and feature extraction layers with unprocessed raw data streams, serving as the foundational data source for all subsequent analysis processes.

[0058] The communication parameters include signal-to-noise ratio, MAC layer retransmission rate, and channel impulse response; the positioning parameters include Doppler frequency shift, elevation difference, and number of visible satellites; and the environmental parameters include weather radar reflectivity, base station load rate, and terrain roughness.

[0059] The communication data is collected in real time through a distributed sensor network. The terminal device's built-in RF front-end chipset measures the instantaneous signal-to-noise ratio at 10ms intervals. An NI PXIe-5663 vector signal analyzer is used to capture IQ signals and simultaneously record the peak-to-average power ratio characteristics of the physical layer channel impulse response. The MAC layer retransmission rate data is obtained by parsing the Retry bit in the 802.11 protocol control frame. Probe devices are deployed on the base station side to collect the PDCP layer packet loss rate and RLC layer buffer status. All communication parameters are timestamped and encapsulated into communication quality data frames conforming to the IEEE 1901.1 standard.

[0060] The positioning data adopts a multi-source fusion acquisition architecture. The GNSS module uses a ublox ZED-F9P high-precision receiver to output raw observations at a frequency of 10Hz. The built-in MPU-9250 nine-axis inertial sensor captures three-dimensional acceleration and angular velocity at a sampling rate of 200Hz. The Doppler frequency shift value is obtained by calculating the carrier phase change rate. The elevation difference data is simultaneously acquired from the absolute altitude value of the barometer MS5611 and the GNSS elevation data. The RTK positioning results after differential correction are compared in real time with the OpenStreetMap elevation database. The number of visible satellites is dynamically updated by parsing the GSV statement of the NMEA-0183 protocol.

[0061] The environmental data establishes a multi-dimensional sensing system. Meteorological data is accessed through the MODBUS-RTU interface of the local weather station WS-3000 to obtain temperature, humidity, and air pressure change rates in real time. Radar reflectivity data is obtained by subscribing to the meteorological department's API interface to obtain 2.5km gridded numerical forecast products. Base station load rate is converted from the ifInOctets counter of the core network management system SNMP agent. Terrain roughness is generated by scanning with RIEGL VZ-400 lidar to generate point cloud data. Elevation changes are monitored by the Bosch BMP388 barometric pressure sensor built into the mobile terminal, and the fractal dimension of the land surface is calculated by combining it with USGS 30-meter DEM data.

[0062] Preprocessing and feature extraction layer: Cleans and filters the raw data, unifies the spatiotemporal reference, and performs primary feature calculations. It eliminates sensor noise and solves the problem of spatiotemporal inconsistency of multi-source data. It generates channel coherence coefficient, motion distortion factor, and disaster propagation coefficient by calculation, and outputs a standardized feature vector with unified dimensions.

[0063] The cleaning and filtering process uses a sliding window mechanism to detect anomalies in the raw data stream. In communication data, an improved Z-score algorithm is applied to identify outliers. In location data, three-dimensional spatial consistency is calculated using Mahalanobis distance, and abnormal location points with deviations exceeding 4σ are removed. In environmental data, wavelet thresholding is used to eliminate impulse interference.

[0064] The unified deployment of the spatiotemporal reference PTP precise time protocol achieves microsecond-level clock synchronization. The spatial reference transformation uses the WGS84 to ECEF coordinate system transformation matrix to unify all positioning data. For non-uniform sampling data, cubic spline interpolation is used to resample to a 10Hz reference frequency.

[0065] The initial feature calculation generates multidimensional features through joint time-frequency analysis, including communication features, positioning features, and environmental features. Communication features include signal-to-noise ratio gradient, channel energy entropy, and MAC layer burst index. Positioning features include three-dimensional radius of curvature, elevation difference variance, Doppler diffusion factor, and satellite geometric accuracy factor. Environmental features include second derivative of air pressure, radar reflectivity slope, terrain roughness spectrum, base station load change rate, and electromagnetic interference correlation coefficient. Finally, a dimensionally regular feature matrix is ​​output to the progressive analysis engine.

[0066] The communication features are constructed through deep analysis of raw signal-to-noise ratio (SNR), channel impulse response (CTR), and MAC layer retransmission rate (MRR) data. First, a sliding window differential operation is performed on the SNR sequence to extract the signal quality change trend, generating an SNR gradient value that characterizes communication stability. Then, wavelet packet transform technology is used to decompose the channel impulse response waveform, and the channel energy entropy index is constructed by calculating the Shannon entropy value of the energy distribution of each sub-band. Finally, a Poisson process model is established based on the temporal characteristics of the MAC layer MRR, and the burst transmission anomaly intensity is quantified by the deviation between the actual observed value and the theoretical value, forming the MAC layer burst index.

[0067] The signal-to-noise ratio gradient value is specifically expressed as follows: G_{SNR}[n]: signal-to-noise ratio gradient value, SNR[n]: signal-to-noise ratio of the nth sampling point, T_s: sampling interval, N: sliding window length, k: the kth sampling position before the current calculation point; the channel energy entropy is specifically expressed as: E_CEP: Channel energy entropy, WPD_i: Wavelet packet decomposition coefficient of the i-th subband, M: Decomposition layer number; The MAC layer burst index is specifically expressed as: B_MAC: MAC layer burst index, λ_obs: observed MAC retransmission rate, λ_poi: Poisson distribution expectation.

[0068] The positioning features are derived from the spatiotemporal fusion analysis of raw data on Doppler frequency shift, elevation difference, and the number of visible satellites. First, the instantaneous radius of curvature of the three-dimensional motion trajectory is constructed by combining the second derivative of the Doppler frequency shift change rate and the elevation difference. The elevation difference variance index is generated by evaluating the intensity of elevation data fluctuation using the sliding window variance analysis method. The signal scattering characteristics are extracted by analyzing the autocorrelation function of the Doppler frequency shift sequence to form the Doppler diffusion factor. The accuracy attenuation factor is calculated based on the spatial geometric distribution of visible satellites.

[0069] The three-dimensional radius of curvature is specifically expressed as: R_curv: three-dimensional radius of curvature, \vec{v}: velocity vector, \vec{a}: acceleration vector; the elevation difference variance is specifically expressed as: σ_h^2: elevation difference variance, Δh_k=h[k]-h[k-1]: difference between adjacent elevation measurements, K: calculation window size, μ_Δh: arithmetic mean of the elevation difference sequence; the Doppler diffusion factor is specifically expressed as: D_dop: Doppler spread factor, f_d[m]: m-th Doppler frequency shift measurement, τ: correlation delay; the satellite geometric accuracy factor is specifically expressed as: GDOP: Geometric precision factor for satellites, H: Geometric matrix of visible satellites, tr: Matrix trace operation, T: Matrix transpose operation.

[0070] The environmental features integrate multimodal data of meteorological radar reflectivity, base station load rate, and terrain roughness. Second-order difference processing is performed on the time series data of air pressure to capture precursor signals of meteorological changes. A disaster development slope index is constructed by analyzing the trend of radar reflectivity changes through linear regression. The terrain roughness is converted into frequency domain energy spectral density features using fast Fourier transform. An exponentially weighted moving average model of base station load rate is established to quantify the dynamic changes of network resources. The coupling relationship between electromagnetic environment and terrain features is revealed by mutual information analysis.

[0071] The second derivative of the air pressure is specifically expressed as: , ∇^{2}P: second derivative of air pressure, P[t]: air pressure measurement at time t, Δt: differential interval, t: discretized time index; the radar reflectivity slope is specifically expressed as: S_rad: radar reflectivity slope, Z_i: i-th radar reflectivity measurement, N_e: regression window size; the terrain roughness spectrum is specifically represented as: R_rongh: terrain roughness spectrum, X[k]: FFT coefficients of the terrain roughness sequence, K_e: Fourier transform length; the base station load change rate is specifically expressed as: ΔL: base station load change rate, L[t]: current base station load rate, α: forgetting factor; the electromagnetic interference correlation coefficient is specifically expressed as: C_emi: Electromagnetic interference correlation coefficient, L: Base station load rate sequence, R: Terrain roughness sequence, cov: Covariance operation, σ: Standard deviation.

[0072] The channel coherence coefficient is generated by dynamically coupling the real-time acquired signal-to-noise ratio sequence with the channel impulse response characteristics. 250 signal-to-noise ratio sampling points are continuously acquired with a time window of 5 seconds. An evaluation model is constructed by combining the deviation of the current channel impulse response peak-to-average ratio with its 300-second sliding historical average. The noise basis correction in the denominator is dynamically adjusted according to the receiver sensitivity parameters. The numerator reflects the instantaneous signal quality. The final output value is normalized to characterize the channel stability.

[0073] The channel coherence coefficient is specifically expressed as follows: b_1: Channel coherence coefficient, SNR_k: Signal-to-noise ratio at the kth sampling point, CIR_PAPR: Peak-to-average power ratio of channel impulse response, μ_CIR: Historical average CIR, γ_0: Noise basis correction, N_b: Number of sampling windows.

[0074] The motion distortion factor constructs a two-dimensional motion feature vector by fusing Doppler frequency shift measurements with elevation change gradient data. First, the elevation data sequence is subjected to time difference operation to obtain the vertical change rate, which is then combined with the horizontal Doppler frequency shift to form a composite motion vector. The degree of motion anomaly is then quantified by the ratio of the magnitude of this vector to the preset maximum motion speed.

[0075] The motion distortion factor is specifically expressed as follows: b_3: motion distortion factor, f_d: absolute value of Doppler frequency shift, ∇h: elevation change gradient, ν_max: maximum motion velocity.

[0076] The disaster propagation coefficient is constructed by integrating the time integral of meteorological radar reflectivity with the base station load status to build an environmental risk model. A 30-minute sliding window is used to numerically integrate the radar reflectivity data to assess the intensity of persistent meteorological threats. At the same time, an exponential correction term for the base station load rate is introduced to enhance the network congestion effect. The network topology factor in the exponent is calculated based on the Voronoi diagram density of the base station distribution. The base station density is converted into an influence coefficient of 0.2-0.8 through the sigmoid function. The final output value is a product that integrates both meteorological and network risk factors.

[0077] The disaster propagation coefficient is specifically expressed as follows: Γ: disaster propagation coefficient, Z(τ): weather radar reflectivity, L: base station load rate, T: integration time window, δ: network topology factor.

[0078] The progressive analysis engine includes a communication health analysis unit, a spatial confidence analysis unit, and an environmental coupling analysis unit. It first assesses the communication channel quality and determines whether an emergency channel handover is triggered. When communication is normal, it sequentially verifies the positioning accuracy and models the environmental risk. Finally, it outputs three types of analysis results: channel status label, spatial confidence score, and environmental risk level.

[0079] The progressive analysis engine receives feature data from the preprocessing and feature extraction layers, dynamically controls the analysis process branches based on the communication health status, and its output results are directly input into the decision fusion center, constituting the core decision basis generation link of the system's intelligent analysis.

[0080] The progressive analysis engine adopts a three-level serial processing architecture. First, the communication health analysis unit receives the channel coherence coefficient and MAC layer retransmission rate data provided by the preprocessing layer. It calculates the transmission robustness index by fusing the exponential decay model and the sigmoid function. When the index is lower than the 0.4 threshold, it immediately triggers the multi-mode channel switching protocol and sends control commands to the execution layer. When the communication status is normal, the spatial reliability assessment unit is activated.

[0081] The transmission robustness index is specifically expressed as: b_2: Transmission robustness index, λ: MAC layer retransmission rate attenuation coefficient, θ: Channel coherence threshold; When b_2 < 0.4: Activate channel switching protocol and send frequency hopping command to execution layer; When 0.4 ≤ b_2 < 0.6: Start bandwidth adaptive adjustment; When b_2 ≥ 0.6: Release spatial analysis enable signal.

[0082] The λ is obtained by constructing a Poisson regression model using historical operational data, and θ is specified in Section 6.3.2 of 3GPP TS 23.203 V16.2.0.

[0083] The spatial confidence assessment unit receives Doppler frequency shift and elevation difference data, constructs kinematic constraint equations including angular velocity vector and elevation gradient, calculates multi-source positioning dispersion and uses hyperbolic tangent function normalization to generate spatial confidence score. When the score exceeds 0.5 confidence threshold, the environmental coupling analysis unit is activated.

[0084] The spatial reliability assessment unit constructs kinematic constraint equations when 0.4 ≤ b_2: , : 3D acceleration vector, ω: angular velocity vector, β: terrain coupling coefficient; Calculate multi-source positioning discreteness: σ_pos: Multi-source localization dispersion, N_σ: Number of effective localization sources, x_i, y_i: Planar coordinates provided by the i-th localization source, μ_x, μ_y: Arithmetic mean of coordinates of all effective localization sources; Output spatial confidence: C_s: spatial confidence, α: weighting coefficient of positioning dispersion, β_c: motion mutation penalty factor.

[0085] The α value was obtained through optimization in a typical urban canyon scenario using Monte Carlo simulation, while β_c was derived from regression analysis based on vehicle-mounted experimental data.

[0086] The environmental coupling degree analysis unit integrates the time-series integral of meteorological radar reflectivity and the differential data of base station load change rate. By introducing the terrain roughness spectrum, it performs refraction compensation on the disaster propagation model and finally outputs a comprehensive risk value weighted by communication quality. In this process, a dynamic feedback mechanism is established, which automatically upgrades to the manual handling process when the spatial confidence level is lower than 0.3 for three consecutive times.

[0087] The environmental coupling analysis unit initiates environmental analysis when C_s>0.5: Disaster propagation model: γ: Coupling coefficient; Terrain refraction compensation: Generate a comprehensive risk value: R: Comprehensive risk value, ε: Zero-prevention constant; When R>2.0, the MAC layer priority is automatically increased, and when C_s<0.3 for 3 consecutive times, the manual intervention protocol is triggered.

[0088] Decision Fusion Center: Integrates communication health, spatial credibility, and environmental risk data through nonlinear fusion algorithms, calculates a comprehensive emergency index, matches it with preset response strategies, and generates tiered response instructions that include voice confirmation, video verification, and drone rescue.

[0089] The comprehensive emergency response index is specifically expressed as follows: E: Comprehensive emergency index, γ_e: Timeliness decay coefficient.

[0090] The comprehensive emergency index adopts a nonlinear fusion algorithm. First, the transmission robustness index output by the communication health analysis unit and the three-dimensional location reliability calculated by the spatial reliability assessment unit are normalized and multiplied to construct a basic guarantee factor. Then, the logarithmic attenuation operation is performed with the disaster propagation coefficient obtained by the environmental coupling degree analysis unit to amplify the sensitivity of high-risk scenarios. At the same time, a timeliness attenuation term is introduced to dynamically reduce the weight of historical data. γ_e is jointly determined by the system running timestamp difference Δt and the current network load status. Finally, the output value is compressed to the [0, 1] interval through the Sigmoid function to form a standardized emergency index.

[0091] The decision fusion center initiates a level one response when E∈[0, 0.3), initiates two-way voice verification through the VoLTE channel and activates base station-level fingerprint positioning; when E∈[0.3, 0.6), a level two response is triggered, scheduling smart cameras within a 300-meter radius to perform multi-angle video verification, and simultaneously enabling carrier phase differential technology to improve positioning accuracy to sub-meter level; when E≥0.6, a level three response is executed, automatically dispatching tethered drones to establish temporary communication relays, and linking with the 120 / 110 system to generate a joint rescue work order. A hysteresis band of 0.05 is introduced at the boundaries of each threshold interval to prevent state oscillation, and the threshold benchmark is dynamically calibrated every 8 hours based on the output of the environmental coupling calculator.

[0092] The execution control layer consists of three execution units: a multi-mode communication switching controller, a resource scheduler, and a multi-mode alarm generator. These units perform physical operations based on decision instructions, while simultaneously monitoring the execution status and feeding it back to the system database.

[0093] The execution control layer adopts a hardware-in-the-loop architecture to physically execute decision commands. The multi-mode communication handover controller dynamically scans the channel quality of satellite / 5G / Mesh networks through a software-defined radio interface. Based on the target signal-to-noise ratio threshold value issued by the decision center, it uses the ε-greedy algorithm to select the optimal communication link. During the handover process, the base station power amplifier gain is simultaneously adjusted to suppress transient interference.

[0094] The resource scheduler has a built-in FPGA dynamic partition controller and establishes a three-level pipeline structure based on task priority: the three-level response tasks are allocated a 64-bit floating-point arithmetic unit and a dedicated DMA channel, the two-level response tasks use a shared computing array, the first-level response tasks are limited to execution in preemptible time slots, and memory access is optimized through AXI crossbar switches.

[0095] The multimodal alarm generator integrates a PWM dimming circuit and a soundprint synthesis chip. It uses fractal time-frequency modulation technology to generate spatially directional sound and light alarm signals. The light alarm is configured with a pulse mode in accordance with the CIE 62471 safety standard, and the sound alarm is constructed with a swept-frequency sine wave fundamental frequency according to the EN 54-3 standard. At the same time, it generates tactile feedback that conforms to the impact spectrum characteristics of MIL-STD-810G through a vibration motor drive circuit.

[0096] The execution control layer maintains deterministic communication with the decision center via time-triggered Ethernet with a 10ms cycle. The execution status data is written into the dual-port RAM after CRC32 verification, so that the self-testing module of the preprocessing and feature extraction layer can perform reverse trace verification.

[0097] In this embodiment, the determination of the constant threshold and formula parameters is based on theoretical standards, measured data, and dynamic calibration, and will not be specifically explained in this embodiment.

[0098] This invention acquires three main categories of raw data in real time: communication network, spatial positioning, and environmental perception data through a multi-source data acquisition layer. Communication parameters, including signal-to-noise ratio, MAC layer retransmission rate, and channel impulse response, are captured at 10ms intervals using a distributed sensor network and a dedicated analyzer. Positioning parameters are achieved through multi-source fusion using a GNSS module, inertial sensors, and Doppler frequency shift calculation, combined with RTK positioning and elevation database comparison to generate high-precision motion state data. Environmental parameters are integrated with meteorological station, radar reflectivity API, and lidar scanning data to construct a comprehensive sensing system encompassing meteorological, topographical, and base station load data. The acquired raw data undergoes preprocessing and feature extraction layers for cleaning and filtering, spatiotemporal benchmark unification, and primary feature calculations, outputting standardized feature vectors to the progressive analysis engine. The progressive analysis engine executes three levels of cascaded analysis sequentially: The communication health analysis unit calculates the transmission robustness index based on the channel coherence coefficient and MAC layer retransmission rate. If it is below 0.4, a multi-mode channel handover protocol is triggered. When communication is normal, the spatial credibility analysis unit is activated, which calculates the spatial confidence score through kinematic constraint equations and multi-source positioning discreteness. If it exceeds 0.5, the environmental coupling analysis unit is activated, integrating meteorological radar reflectivity integral and base station load rate differential data, and introducing terrain refraction compensation to generate a comprehensive risk value. The decision fusion center converts communication health, spatial credibility, and environmental risk data into a comprehensive emergency index through a nonlinear fusion algorithm, and triggers a graded response according to the index range: Level 1 response initiates voice verification and fingerprint positioning; Level 2 response dispatches intelligent camera video verification and improves positioning accuracy; Level 3 response dispatches tethered drones to establish relay and link with the rescue system. The execution control layer executes commands through a multi-mode communication switching controller, resource scheduler, and multi-mode alarm generator, and feeds back the status to the system database in real time via time-triggered Ethernet, forming a closed-loop control. All process data is verified by CRC32 check and reverse traceback to ensure system reliability and real-time performance, ultimately achieving intelligent emergency call response.

[0099] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0100] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An emergency call response system based on an intelligent communication network, characterized in that, include: Multi-source data acquisition layer: Real-time acquisition of three major categories of raw data: communication network, spatial positioning and environmental perception, including channel quality indicators of communication parameters, motion state data of positioning parameters and meteorological and terrain information of environmental parameters, forming standardized data packets containing timestamps, device identifiers and acquired values; Preprocessing and feature extraction layer: Cleans and filters the raw data, unifies the spatiotemporal reference and performs primary feature calculations, eliminates sensor noise and solves the problem of spatiotemporal inconsistency of multi-source data, generates channel coherence coefficient, motion distortion factor and disaster propagation coefficient by calculation, and outputs a standardized feature vector with unified dimension. The initial feature calculation generates multi-dimensional features through joint time-frequency analysis, including communication features, positioning features, and environmental features. Communication features include signal-to-noise ratio gradient, channel energy entropy, and MAC layer burst index. Positioning features include three-dimensional radius of curvature, elevation difference variance, Doppler diffusion factor, and satellite geometric accuracy factor. Environmental features include second derivative of air pressure, radar reflectivity slope, terrain roughness spectrum, base station load change rate, and electromagnetic interference correlation coefficient. Finally, a dimensionally regular feature matrix is ​​output to the progressive analysis engine. The progressive analysis engine includes a communication health analysis unit, a spatial confidence analysis unit, and an environmental coupling analysis unit. It first assesses the communication channel quality and determines whether an emergency channel handover is triggered. When communication is normal, it sequentially verifies the positioning accuracy and models the environmental risk. Finally, it outputs three types of analysis results: channel status label, spatial confidence score, and environmental risk level. The progressive analysis engine adopts a three-level serial processing architecture. First, the communication health analysis unit receives channel coherence coefficient and MAC layer retransmission rate data provided by the preprocessing and feature extraction layers. It calculates the transmission robustness index by fusing the exponential decay model and the sigmoid function. When the index is lower than the 0.4 threshold, the multi-mode channel switching protocol is immediately triggered and a control command is sent to the execution layer. When the communication status is normal, the spatial reliability analysis unit is activated. The spatial confidence analysis unit receives Doppler frequency shift and elevation difference data, constructs kinematic constraint equations including angular velocity vector and elevation gradient, calculates multi-source positioning dispersion and uses hyperbolic tangent function normalization to generate spatial confidence score. When the score exceeds 0.5 confidence threshold, the environmental coupling analysis unit is activated. The environmental coupling degree analysis unit integrates the time-series integral of meteorological radar reflectivity and the differential data of base station load change rate. By introducing the terrain roughness spectrum, it performs refraction compensation on the disaster propagation model and finally outputs a comprehensive risk value weighted by communication quality. In this process, a dynamic feedback mechanism is established, which automatically upgrades to the manual handling process when the spatial confidence level is lower than 0.3 for three consecutive times. Decision Fusion Center: Integrates communication health, spatial confidence scores and environmental risk data through nonlinear fusion algorithms, calculates a comprehensive emergency index and matches it with preset response strategies, and generates tiered response instructions that include voice confirmation, video verification and drone rescue. The comprehensive emergency index adopts a nonlinear fusion algorithm. First, the transmission robustness index output by the communication health analysis unit and the spatial confidence score calculated by the spatial credibility analysis unit are normalized and multiplied to construct a basic guarantee factor. Then, the logarithmic attenuation operation is performed with the disaster propagation coefficient obtained by the environmental coupling degree analysis unit to amplify the sensitivity of high-risk scenarios. At the same time, a timeliness attenuation term is introduced to dynamically reduce the weight of historical data. The timeliness attenuation coefficient γ_e is jointly determined by the system running timestamp difference Δt and the current network load status. Finally, the output value is compressed to the [0, 1] interval through the Sigmoid function to form a standardized emergency index. The decision fusion center initiates a Level 1 response when the comprehensive emergency index E∈[0, 0.3), initiates two-way voice verification through the VoLTE channel and activates base station-level fingerprint positioning; when E∈[0.3, 0.6), a Level 2 response is triggered, scheduling smart cameras within a 300-meter radius to perform multi-angle video verification, and simultaneously enabling carrier phase differential technology to improve positioning accuracy to sub-meter level; when E≥0.6, a Level 3 response is executed, automatically dispatching tethered drones to establish temporary communication relays, and linking with the 120 / 110 system to generate a joint rescue work order. A hysteresis band of 0.05 is introduced at the boundaries of each threshold interval to prevent state oscillation, and the threshold benchmark is dynamically calibrated every 8 hours based on the output of the environmental coupling calculator. The execution control layer consists of three execution units: a multi-mode communication switching controller, a resource scheduler, and a multi-mode alarm generator. These units perform physical operations based on decision instructions, while simultaneously monitoring the execution status and feeding it back to the system database. The execution control layer adopts a hardware-in-the-loop architecture to implement the physical execution of decision commands. The multi-mode communication handover controller dynamically scans the channel quality of satellite / 5G / Mesh networks through a software-defined radio interface. Based on the target signal-to-noise ratio threshold value issued by the decision fusion center, it selects the optimal communication link using the ε-greedy algorithm. During the handover process, the base station power amplifier gain is simultaneously triggered to suppress transient interference. The resource scheduler has a built-in FPGA dynamic partition controller and establishes a three-level pipeline structure based on task priority: the three-level response tasks are allocated a 64-bit floating-point arithmetic unit and a dedicated DMA channel, the two-level response tasks use a shared computing array, the first-level response tasks are limited to execution in preemptible time slots, and memory access is optimized through AXI crossbar switches. The multimodal alarm generator integrates a PWM dimming circuit and a soundprint synthesis chip. It uses fractal time-frequency modulation technology to generate spatially directional sound and light alarm signals. The light alarm is configured with a pulse mode in accordance with the CIE 62471 safety standard, and the sound alarm is constructed with a swept-frequency sine wave fundamental frequency according to the EN 54-3 standard. At the same time, it generates tactile feedback that conforms to the impact spectrum characteristics of MIL-STD-810G through a vibration motor drive circuit.

2. The emergency call response system based on an intelligent communication network according to claim 1, characterized in that: The communication parameters include signal-to-noise ratio, MAC layer retransmission rate, and channel impulse response; the positioning parameters include Doppler frequency shift, elevation difference, and number of visible satellites; and the environmental parameters include weather radar reflectivity, base station load rate, and terrain roughness.

Images