A scenario-channel parameter mapping system and method based on multi-source fusion and dynamic learning

Abstract

This invention discloses a scenario-channel parameter mapping system and method based on multi-source fusion and dynamic learning, relating to the fields of wireless communication simulation and digital twin technology. The system includes: a multi-source information sensing and standardization layer: used to receive and standardize heterogeneous scenario input data, outputting a standardized quantized feature vector set; an intelligent mapping and coupling effect modeling layer: serving as the core computing unit of the engine, responsible for mapping the standardized feature vectors into preliminary channel parameters; a parameter synthesis and interface adaptation layer: used to synthesize the basic channel parameters output by the intelligent mapping layer into the complete parameter set required by the channel simulator, and to complete format adaptation and interface compatibility; and a verification and dynamic learning optimization layer: monitoring simulation results in real time, comparing them with the scenario's expected target, forming parameter optimization feedback, and realizing continuous iteration of engine performance. The system has advantages such as significantly improved parameter mapping accuracy, strong dynamic adaptive capability, and outstanding intelligent and self-evolving characteristics.

Description

Technical Field

[0001] This invention relates to the field of wireless communication simulation and digital twin technology, and in particular to a scenario-channel parameter mapping system and method based on multi-source fusion and dynamic learning. Background Technology

[0002] In the design, verification, and simulation training process of wireless communication systems, the channel simulator is the core device for reproducing the real propagation environment. Its simulation accuracy directly depends on the accuracy of the input channel parameters (such as path loss, shadowing fading, multipath delay spread, Doppler spectrum, interference-to-noise ratio, etc.).

[0003] Currently, the mainstream methods for configuring channel parameters in the industry have the following significant bottlenecks, making it difficult to meet the needs of high-fidelity and high-dynamic scenarios:

[0004] (i) The model is too simple and has poor scene adaptability.

[0005] Most solutions rely on only a single type of model, specifically:

[0006] 1. When using traditional empirical models (such as the Okumura-Hata model), the accuracy of parameter prediction drops sharply in complex and heterogeneous scenarios (such as alternating urban canyons and open areas with strong electromagnetic interference).

[0007] 2. While using a geometric stochastic model (such as the RTK model) can improve accuracy, the computational load is too large, making it difficult to meet the real-time requirements of rapid node movement scenarios in communication scenarios.

[0008] (ii) Elements are isolated and lack coupling effect.

[0009] Existing methods generally treat factors affecting the channel, such as terrain, weather, and electromagnetic interference, as independent variables, or simply perform linear superposition. However, in real-world scenarios, these factors exhibit strong nonlinear coupling effects, for example:

[0010] 1. Rainfall alters the surface reflection characteristics, thereby indirectly affecting the intensity and distribution of multipath components;

[0011] 2. Dense interference can alter the statistical distribution of background noise, reduce receiver sensitivity, and further degrade channel quality.

[0012] Existing technologies lack the ability to effectively model such nonlinear coupling relationships, resulting in a large deviation between parameter mapping results and the real environment.

[0013] (iii) Static configuration, lacking dynamic feedback mechanism

[0014] Traditional parameter mapping processes are mostly static and offline. When the scenario changes dynamically (such as node maneuver trajectory adjustment, interference pattern switching, and gradual change in weather conditions), the system cannot adjust the channel parameters in real time and smoothly, resulting in the simulation environment being out of sync with the actual scenario and making it difficult to support closed-loop testing and simulation training in highly dynamic environments.

[0015] (iv) Reliance on expert experience, low degree of automation

[0016] Channel parameter configuration heavily relies on engineers' understanding of the principles and scenario background of the channel model, which presents the following problems:

[0017] 1. The configuration process is cumbersome and inefficient, and cannot adapt to the needs of rapid simulation in large-scale, multi-scenario applications;

[0018] 2. Differences in experience among different engineers lead to poor consistency in parameter configuration results, making it difficult to guarantee the repeatability and comparability of simulation data. Summary of the Invention

[0019] The technical problem to be solved by the present invention is how to provide a scenario-channel parameter mapping system and method based on multi-source fusion and dynamic learning, which has significantly improved parameter mapping accuracy, strong dynamic adaptive capability, and outstanding intelligent and self-evolutionary characteristics.

[0020] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a scenario-channel parameter mapping system based on multi-source fusion and dynamic learning, the system comprising:

[0021] Multi-source information perception and standardization layer: used to receive and standardize heterogeneous hypothetical input data, and output a standardized set of quantized feature vectors;

[0022] Intelligent mapping and coupling effect modeling layer: This layer serves as the core computing unit of the engine and is responsible for mapping standardized feature vectors to preliminary channel parameters.

[0023] Parameter synthesis and interface adaptation layer: This layer synthesizes the basic channel parameters output by the intelligent mapping layer into a complete parameter set required by the channel simulator, and completes format adaptation and interface compatibility.

[0024] Verification and Dynamic Learning Optimization Layer: Real-time monitoring of simulation results, comparison with expected targets, forming parameter optimization feedback, and realizing continuous iteration of engine performance.

[0025] Furthermore, the multi-source information perception and standardization layer includes:

[0026] High-precision geographic information integration module: used to deeply integrate GIS data and real-time remote sensing data to extract refined environmental feature vectors;

[0027] Electromagnetic situation awareness module: used to import electromagnetic spectrum monitoring data or interference scenario scripts, analyze the spatiotemporal-frequency domain characteristics of interference sources, and generate dynamic interference basis;

[0028] Node Behavior Analysis Module: Used to analyze node motion trajectories and platform attitude, providing basic data for antenna gain pattern calculation and dynamic Doppler effect analysis.

[0029] Furthermore, the intelligent mapping and coupling effect modeling layer includes

[0030] Hybrid rule base: used to cover parameter association logic in different scenarios;

[0031] Coupling effect compensation module: Introduces a lightweight neural network or gradient boosting decision tree model to learn and predict the amount of correction to the output parameters of the basic model under the coupling effect of multiple factors;

[0032] Real-time interpolation mechanism module: used to perform smooth interpolation filtering between channel parameters at adjacent locations or time points for node mobile communication scenarios.

[0033] This invention also discloses a scenario-channel parameter mapping method based on multi-source fusion and dynamic learning, the method comprising the following steps:

[0034] It receives and standardizes heterogeneous hypothetical input data, and outputs a standardized set of quantized feature vectors.

[0035] Map the standardized feature vectors to preliminary channel parameters;

[0036] The basic channel parameters output by the intelligent mapping layer are synthesized into a complete parameter set required by the channel simulator, and format adaptation and interface compatibility are completed.

[0037] Real-time monitoring of simulation results, comparison with expected targets, and feedback for parameter optimization are generated to achieve continuous iteration of engine performance.

[0038] The beneficial effects of adopting the above technical solution are as follows: 1) The accuracy of parameter mapping is significantly improved: Through high-precision geographic information fusion technology, the environmental features are extracted in a refined manner; combined with a dedicated coupling effect compensation model, the nonlinear superposition problem in complex multi-element scenarios is effectively solved, and the overall parameter mapping error is greatly reduced.

[0039] 2) Strong dynamic adaptive capability: It integrates real-time interpolation technology and standardized online interface, which can seamlessly respond to dynamic changes in the scenario (such as node maneuvering and interference switching), and provide high-fidelity channel environment support for high dynamic scenario simulation such as mobility management and emergency response.

[0040] 3) Prominent intelligent and self-evolving characteristics: The built-in machine learning module and dynamic optimization closed loop enable the system to break through the limitations of the traditional "static lookup table" and become an intelligent agent that can learn from data and continuously improve itself, greatly reducing the dependence on expert experience and enhancing long-term use value.

[0041] 4) High practicality and feasibility: The solution design fully considers compatibility with existing simulation platforms: it provides flexible API interfaces that can be quickly connected to commercial / open source channel simulators; it adopts an "offline training-online inference" mechanism to ensure real-time processing speed while maintaining accuracy, and is easy to deploy and apply in existing simulation systems. Attached Figure Description

[0042] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0043] Figure 1 This is a schematic block diagram of the system described in the embodiment of the present invention;

[0044] Figure 2 This is a flowchart of the coupling effect compensation module in the system described in the embodiment of the present invention;

[0045] Figure 3 This is a flowchart of the method described in an embodiment of the present invention. Detailed Implementation

[0046] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0047] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0048] Example 1

[0049] like Figure 1 As shown, this embodiment of the invention discloses a scenario-channel parameter mapping system based on multi-source fusion and dynamic learning, the system comprising:

[0050] 1) Multi-source information perception and standardization layer:

[0051] The multi-source information perception and standardization layer is used to receive and standardize heterogeneous hypothetical input data, and output a standardized set of quantized feature vectors to provide high quality for subsequent mapping processes. The input data breaks through the limitations of traditional "drop-down menu" scene classification and has high-precision, multi-dimensional data fusion capabilities.

[0052] The multi-source information perception and standardization layer includes:

[0053] High-precision geographic information integration module: used to deeply integrate GIS data (including digital elevation model DEM and building outline data DSM) with real-time remote sensing data, automatically extract refined environmental feature vectors such as terrain relief, building density and height, and vegetation index, replacing traditional coarse-grained classification such as "city / mountain".

[0054] Electromagnetic situation awareness module: used to import electromagnetic spectrum monitoring data or interference scenario scripts, analyze the spatiotemporal-frequency domain characteristics of interference sources (such as pulse repetition intervals and frequency hopping patterns), and generate dynamic interference basis;

[0055] Node Behavior Analysis Module: Used to analyze node motion trajectories and platform attitudes (such as UAV pitch and roll angles), providing basic data for antenna gain pattern calculation and dynamic Doppler effect analysis.

[0056] 2) Intelligent mapping and coupling effect modeling layer:

[0057] The intelligent mapping and coupling effect modeling layer serves as the "core computing unit" of the engine, responsible for mapping standardized feature vectors to preliminary channel parameters. Its innovation lies in the deep integration of rules and models, and the accurate modeling of nonlinear coupling effects.

[0058] The implementation of the intelligent mapping and coupling effect modeling layer relies on the following three key technologies:

[0059] 2-1) Hybrid rule base: Covers parameter association logic for different scenarios, and is divided into the following two categories:

[0060] Deterministic Model Library: Based on fixed formula mappings of theoretical models, suitable for directly calculable parameters such as path loss. It integrates classic models such as 3GPP TR38.901 and ITU-R recommendations, and has built-in model selection logic that can automatically call the optimal model based on the characteristics of a given communication scenario (such as frequency, distance, and environment type).

[0061] Probabilistic Statistical Library: Based on statistical mapping of measured data, applicable to random parameters such as multipath fading and bit error rate. Based on extensive measured data from 5G vehicle-to-everything (V2X) and low-altitude UAV channels, statistical distribution models for key parameters such as Rice K-factor and latency spread are constructed to improve parameter accuracy.

[0062] 2-2) Coupling Effect Compensation Module: This module introduces a lightweight neural network or gradient boosting decision tree model to specifically learn and predict the correction amount of the basic model's output parameters under the coupling effects of multiple factors (such as "rainfall + specific terrain + movement speed"). The specific processing flow is as follows... Figure 2 As shown, it includes:

[0063] Data preparation stage: Obtain accurate mapping relationship data of multi-factor input and correction output, perform preprocessing (remove outliers and fill in missing values), and divide it into training set (for model fitting), validation set (for hyperparameter tuning), and test set (for final effect evaluation).

[0064] Model training and optimization phase: The model is iteratively trained using the training set with the mean squared error (MSE) between the predicted and actual values ​​as the loss function; the accuracy is evaluated using the validation set in each iteration to avoid overfitting; the trained model is pruned (redundant neurons / tree nodes are removed) and quantized to ensure that the inference time meets the requirements of online real-time performance.

[0065] System integration phase: Achieve offline training-online inference closed loop, embed the trained model into the existing channel simulation system, open up the data flow path, and ensure the efficiency of online inference.

[0066] 2-3) Real-time interpolation mechanism: For node mobile communication scenarios, smooth interpolation filtering is performed between channel parameters at adjacent locations or time points to avoid parameter jumps and ensure the continuity of channel characteristic changes.

[0067] 3) Parameter synthesis and interface adaptation layer:

[0068] The parameter synthesis and interface adaptation layer is used to synthesize the basic channel parameters output by the intelligent mapping layer into the complete parameter set required by the channel simulator, and to complete the format adaptation and interface compatibility.

[0069] The parameter synthesis and interface adaptation layer includes:

[0070] The parameter integrity guarantee module not only outputs large-scale fading parameters such as path loss and shadow fading, but also automatically generates small-scale fading parameters (such as tap parameters of a geometry-based random channel model, including delay, power, and Doppler) according to the scenario, forming a complete channel impulse response description.

[0071] Standardized API Interface: Provides RESTful API or standardized message middleware interface, supports JSON / XML file output, and can also perform real-time online data interaction with commercial channel simulators such as Keysight Technologies, Spirent, and Kunheng Shunwei, as well as open source platforms such as NS-3 and AIRT, to meet hardware-in-the-loop simulation requirements.

[0072] 4) Validation and dynamic learning optimize the closed-loop layer:

[0073] The verification and dynamic learning optimization closed loop is used to monitor simulation results in real time, compare them with the expected goals, form parameter optimization feedback, and realize continuous iteration of engine performance.

[0074] The validation and dynamic learning optimization closed-loop layer includes:

[0075] Online performance monitoring: Real-time collection of communication performance metrics output by the simulator, including bit error rate (BER), throughput, latency, etc.

[0076] Intelligent deviation diagnosis: Preset performance deviation thresholds. When the actual indicators deviate from the expected limits, automatically analyze the main factors of the deviation (such as path loss model error under specific terrain and noise estimation deviation under interference conditions).

[0077] Online optimization of rule base / model: Using bias data and corresponding hypothetical inputs as new training samples, the coupling effect compensation module is incrementally learned through periodic triggering or event triggering, updating model parameters and rule base, so that the mapping engine has the self-evolution capability of "becoming more accurate with use".

[0078] Example 2

[0079] like Figure 3 As shown, this invention discloses a scenario-channel parameter mapping method based on multi-source fusion and dynamic learning. The method is based on the scenario-channel parameter mapping system and includes the following steps:

[0080] It receives and standardizes heterogeneous hypothetical input data, and outputs a standardized set of quantized feature vectors.

[0081] Map the standardized feature vectors to preliminary channel parameters;

[0082] The basic channel parameters output by the intelligent mapping layer are synthesized into a complete parameter set required by the channel simulator, and format adaptation and interface compatibility are completed.

[0083] Real-time monitoring of simulation results, comparison with expected targets, and feedback for parameter optimization are generated to achieve continuous iteration of engine performance.

[0084] It should be noted that the specific implementation methods of each step in the method can be referred to in Embodiment 2, and will not be repeated here.

[0085] Example 3

[0086] This invention discloses a method for anti-interference scenario planning for urban communication of low-altitude unmanned aerial vehicles (UAVs), the method comprising the following steps:

[0087] (a) Speculated input:

[0088] 1. Environmental parameters: GIS data (urban core area, average building height 60m, street width 20m); meteorological conditions (light rain, rainfall rate 5mm / h);

[0089] 2. Link parameters: Node A (ground control station, altitude 5m, omnidirectional antenna); Node B (UAV, altitude 150m, speed 50km / h, directional antenna pointing towards ground station); communication distance 1.5km, operating frequency 2.4GHz;

[0090] 3. Interference parameters: intermittent frequency sweep interference, center frequency 2.4GHz, bandwidth 20MHz, interference signal ratio (J / S) = 10dB.

[0091] (ii) Mapping process:

[0092] 1. Multi-source information perception and standardization layer processing:

[0093] The GIS data is parsed to identify the scene type as "non-line-of-sight (NLOS)" (due to building obstruction); additional attenuation under light rain conditions is calculated; features such as node movement speed and antenna attitude are extracted to generate a standardized feature vector set.

[0094] 2. Intelligent mapping and coupling effect modeling layer processing:

[0095] Basic path loss calculation: The system automatically selects the 3GPPUMaNLOS model to calculate the basic path loss based on the characteristics of the urban core area, 2.4GHz frequency, and NLOS scenario.

[0096] Coupling effect compensation: Call the trained coupling effect compensation model, input features such as building height 60m, street width 20m, NLOS, light rain, drone speed 50km / h, and output the correction values ​​for basic path loss (+3dB additional loss) and shadow fading variance.

[0097] Multipath parameter generation: Based on the 3GPPUMaNLOS model recommendations, multipath time delay spectrum and Rice K-factor are generated;

[0098] Interference parameter processing: Calculate the equivalent noise rise based on the J / S ratio and mark the time slots where interference exists.

[0099] 3. Parameter synthesis and interface adaptation layer processing:

[0100] By synthesizing the above parameters, configuration code for ChannelConditionModel and SpectrumModel suitable for the NS-3 simulator is generated and injected into the NS-3 simulation core in real time via RESTful API.

[0101] (III) Verification and Optimization:

[0102] 1. Simulation and Performance Monitoring: The simulation was performed in NS-3, and the actual throughput was found to be 15% lower than expected.

[0103] 2. Deviation diagnosis: Optimization of closed-loop triggering and system analysis revealed that under the conditions of "light rain + specific city geometry", the current coupled compensation model predicts insufficient additional losses, which is the dominant factor causing throughput deviation.

[0104] 3. Incremental learning and model update: Store the current target input data and performance deviation data into the sample database; during system idle periods, fine-tune the coupling effect compensation model using new samples and update the model weights;

[0105] 4. Effect verification: When encountering similar scenarios such as light rain and NLOS in the core urban area, the parameter mapping accuracy was significantly improved, and the deviation between the simulation throughput and the expectation was controlled within 5%.

[0106] The method described in this invention solves the problems of low accuracy, poor adaptability, and static rigidity in existing technologies by deeply integrating multi-source scenario information, innovating the coupling effect of modeling elements, and introducing a dynamic learning optimization closed loop. This invention also provides a methodological reference and data support for constructing high-fidelity communication digital twins, and has broad application prospects and significant practical value in fields such as military communications, autonomous driving, and the low-altitude economy.

[0107] Although the above embodiments have been described, those skilled in the art, once they understand the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the above descriptions are merely embodiments of the present invention and do not limit the scope of patent protection of the present invention. Any equivalent structural or procedural transformations made using the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A scenario-channel parameter mapping system based on multi-source fusion and dynamic learning, characterized in that, The system includes: Multi-source information perception and standardization layer: used to receive and standardize heterogeneous hypothetical input data, and output a standardized set of quantized feature vectors; Intelligent mapping and coupling effect modeling layer: This layer serves as the core computing unit of the engine and is responsible for mapping standardized feature vectors to preliminary channel parameters. Parameter synthesis and interface adaptation layer: This layer synthesizes the basic channel parameters output by the intelligent mapping layer into a complete parameter set required by the channel simulator, and completes format adaptation and interface compatibility. Verification and Dynamic Learning Optimization Layer: Real-time monitoring of simulation results, comparison with expected targets, forming parameter optimization feedback, and realizing continuous iteration of engine performance.

2. The scenario-channel parameter mapping system based on multi-source fusion and dynamic learning as described in claim 1, characterized in that, The multi-source information perception and standardization layer includes: High-precision geographic information integration module: used to deeply integrate GIS data and real-time remote sensing data to extract refined environmental feature vectors; Electromagnetic situation awareness module: used to import electromagnetic spectrum monitoring data or interference scenario scripts, analyze the spatiotemporal-frequency domain characteristics of interference sources, and generate dynamic interference basis; Node Behavior Analysis Module: Used to analyze node motion trajectories and platform attitude, providing basic data for antenna gain pattern calculation and dynamic Doppler effect analysis.

3. The scenario-channel parameter mapping system based on multi-source fusion and dynamic learning as described in claim 1, characterized in that, The intelligent mapping and coupling effect modeling layer includes Hybrid rule base: used to cover parameter association logic in different scenarios; Coupling effect compensation module: Introduces a lightweight neural network or gradient boosting decision tree model to learn and predict the amount of correction to the output parameters of the basic model under the coupling effect of multiple factors; Real-time interpolation mechanism module: used to perform smooth interpolation filtering between channel parameters at adjacent locations or time points for node mobile communication scenarios.

4. The scenario-channel parameter mapping system based on multi-source fusion and dynamic learning as described in claim 3, characterized in that, The hybrid rule base includes: Deterministic Model Library: Based on fixed formula mapping of theoretical models, it is suitable for parameters that can be directly calculated. It integrates classic models and has built-in model selection logic to automatically call the optimal model according to the characteristics of the desired communication scenario. Probabilistic statistics library: Statistical mapping based on measured data, suitable for random parameters. It constructs statistical distribution models based on measured data to improve the accuracy of parameters.

5. The scenario-channel parameter mapping system based on multi-source fusion and dynamic learning as described in claim 3, characterized in that, The processing method of the coupling effect compensation module includes the following steps: Data preparation: Obtain accurate mapping relationship data of multi-factor input-correction output, perform preprocessing, and divide into training set, validation set and test set; Model training and optimization: The mean squared error between the predicted and actual values ​​of the correction is used as the loss function. The model is trained iteratively using the training set. The accuracy is evaluated through the validation set in each iteration. The trained model is pruned and quantized to ensure that the inference time meets the requirements of online real-time performance. System integration: Implement an offline training-online inference closed loop, embedding the trained model into the existing channel simulation system.

6. The scenario-channel parameter mapping system based on multi-source fusion and dynamic learning as described in claim 1, characterized in that, The parameter synthesis and interface adaptation layer includes: Parameter integrity assurance module: used to output large-scale fading parameters and automatically generate small-scale fading parameters according to the scenario to form a complete channel impulse response description; Standardized API Interface: Used to provide RESTful API or standardized message middleware interface to meet hardware-in-the-loop simulation requirements.

7. The scenario-channel parameter mapping system based on multi-source fusion and dynamic learning as described in claim 1, characterized in that, The verification and dynamic learning optimization layer includes: Online performance monitoring module: used to collect communication performance metrics output by the simulator in real time; Intelligent Deviation Diagnosis Module: Used to preset performance deviation thresholds; when the actual indicators deviate from the expected limits, it analyzes the dominant factors of the deviation. The rule base / model online optimization module is used to incrementally learn the coupling effect compensation module by taking the bias data and corresponding expected inputs as new training samples and updating the model parameters and rule base through periodic triggering or event triggering.

8. A scenario-channel parameter mapping method based on multi-source fusion and dynamic learning, wherein the method is based on the scenario-channel parameter mapping system according to any one of claims 1-7, characterized in that... The method includes the following steps: It receives and standardizes heterogeneous hypothetical input data, and outputs a standardized set of quantized feature vectors. Map the standardized feature vectors to preliminary channel parameters; The basic channel parameters output by the intelligent mapping layer are synthesized into a complete parameter set required by the channel simulator, and format adaptation and interface compatibility are completed. Real-time monitoring of simulation results, comparison with expected targets, and feedback for parameter optimization are generated to achieve continuous iteration of engine performance.

Images