5G edge communication method and system based on big data

Abstract

The invention relates to the technical field of edge communication, and particularly discloses a 5G edge communication method and system based on big data, and the method comprises the following steps: determining a communication demand of a target communication region, and extracting an index parameter; calculating an echo coefficient based on the index parameter; querying a communication environment corresponding to the echo coefficient, and extracting a frequency band parameter; dynamic anomaly detection is carried out on the frequency band parameters, and an aggregation data set is generated after abnormal frequency bands are removed; uploading the aggregated data set to an edge node, extracting an aggregation factor, and selecting a data transmission protocol; adjusting a data transmission path of the 5G communication cloud platform according to the selected data transmission protocol, and calculating a data bandwidth; and based on equipment parameters of the data bandwidth configuration edge communication equipment, extracting center power and resonance frequency, calculating response frequency through an environment prediction model, and generating a self-adaptive edge communication scheme. According to the method, the problems of calculation redundancy, adaptability rigidity and resource waste in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the field of edge communication technology, and in particular to a 5G edge communication method and system based on big data. Background Art

[0002] With the explosive growth of IoT devices, traditional edge communication solutions face three core contradictions:

[0003] 1. Low latency and high energy efficiency cannot be achieved at the same time:

[0004] While existing protocol optimization technologies (such as patent application CN119109948A) can statically adjust communication frequency based on device power levels, they cannot dynamically balance latency and energy consumption under burst traffic. Actual measurements show that burst latency fluctuates by over 150ms, while the energy efficiency ratio is only 2.1 TOPS / W.

[0005] 2. Insufficient adaptability to heterogeneous environments:

[0006] In 5G / Industrial IoT multi-protocol scenarios, fixed path selection algorithms (such as shortest path first) lack real-time network awareness, resulting in link failure recovery times exceeding 500ms. For example, Chinese patent publication number CN115515149B uses reinforcement learning to optimize node parameters, but fails to address the model bloat issue—DNNs with more than 10 layers experience inference delays exceeding one second on edge chips.

[0007] 3. The natural conflict between safety and efficiency:

[0008] Traditional encryption protocols (such as TLS) increase overhead, while lightweight security solutions reduce their ability to resist attacks. Summary of the Invention

[0009] In order to solve the problems faced by traditional edge communication solutions in the existing technology, such as the inability to achieve both low latency and high energy efficiency, insufficient adaptability to heterogeneous environments, and the natural conflict between security and efficiency, the present invention provides a 5G edge communication method and system based on big data.

[0010] The technical solution adopted in the present invention is:

[0011] A 5G edge communication method based on big data, comprising the following steps:

[0012] Determine the communication demand of the target communication area, analyze key communication indicators in the communication demand through a pre-trained demand prediction model, and extract indicator parameters;

[0013] Based on the indicator parameters, an echo coefficient is calculated using a neural network model, wherein the input of the neural network model includes the communication frequency, path loss, data transmission distance and signal-to-noise ratio of the indicator path, and the output is an approximate echo coefficient;

[0014] Querying the communication environment corresponding to the echo coefficient, collecting communication parameters in the environment and converting them into a communication frequency band curve, and extracting frequency band parameters;

[0015] Performing dynamic anomaly detection on the frequency band parameters through an online clustering algorithm, and generating an aggregated data set after eliminating abnormal frequency bands;

[0016] Uploading the aggregated data set to an edge node, extracting the aggregation factor, and dynamically selecting a data transmission protocol through a reinforcement learning model based on real-time network status data;

[0017] Adjust the data transmission path of the 5G communication cloud platform according to the selected data transmission protocol, monitor the data transmission rate and frequency in real time, and calculate the data bandwidth based on the energy efficiency constraint factor;

[0018] The device parameters of the edge communication device are configured based on the data bandwidth, the center power and the resonant frequency are extracted, the response frequency is calculated through the environmental prediction model, and an adaptive edge communication solution is generated.

[0019] Preferably, the demand forecasting model is a regression model of the TinyML architecture, whose training data includes historical communication area characteristics, device density and environment type, and the output is a communication demand target value, which is used to replace manual configuration of communication parameters.

[0020] Preferably, the neural network model is a one-dimensional convolutional neural network, and its calculation process satisfies:

[0021]

[0022] in, is the approximate echo coefficient, is the communication frequency, is the path loss, is the data transmission distance, is the signal-to-noise ratio;

[0023] The neural network model generates training data by simulating an existing echo coefficient formula.

[0024] Preferably, the online clustering algorithm is an incremental K-means algorithm, which dynamically adjusts the abnormal frequency band detection threshold, specifically including:

[0025] Receive frequency band parameter stream data in real time and update cluster centers in a sliding window manner;

[0026] Calculate the distance between each data point and the cluster center, and it will exceed the dynamic threshold The frequency band is marked as abnormal;

[0027]

[0028] in, For the The mean of the class, is the standard deviation.

[0029] Preferably, the reinforcement learning model is a Q-learning algorithm, whose state space includes:

[0030] Real-time signal-to-noise ratio (SNR) level;

[0031] Network congestion index;

[0032] Aggregation factor type;

[0033] The action space is a set of optional data transmission protocols, and the reward function is defined as:

[0034]

[0035] in, To improve bandwidth resource utilization efficiency, For data transmission delay, and is the weight coefficient.

[0036] Preferably, the energy efficiency constraint factor is calculated using the following formula to calculate the data bandwidth:

[0037]

[0038] in, represents the data bandwidth corresponding to the data in the aggregated data set, Represents the data item index corresponding to the data in the aggregated dataset, Indicates the number of data items corresponding to the data in the aggregated dataset, Indicates the data size, used to represent the amount of each data in the aggregated data set, represents the data transmission rate, represents the data frequency, The energy efficiency factor is dynamically calculated based on the remaining power of the device and historical power consumption data:

[0039]

[0040] in, is the current power consumption, is the average power consumption, is the maximum allowable power consumption.

[0041] Preferably, the environmental prediction model is a long short-term memory network, the input is a historical environmental parameter sequence, and the output is a response frequency adjustment value. , the final response frequency is calculated as:

[0042]

[0043] in, Calculated from the center power and resonant frequency using a simplified formula.

[0044] A 5G edge communication system based on big data, for executing any of the above 5G edge communication methods based on big data, comprising:

[0045] Intelligent demand analysis module: executes demand forecasting model and outputs communication demand targets;

[0046] AI calculation module: integrates neural network model to calculate approximate echo coefficient;

[0047] Dynamic aggregation module: realizes abnormal frequency band detection and data aggregation through online clustering algorithm;

[0048] Adaptive protocol engine: Dynamically selects data transmission protocols based on reinforcement learning models;

[0049] Energy efficiency optimization module: injects energy efficiency constraint factors to calculate data bandwidth and generate device configuration parameters;

[0050] Environmental prediction module: Generates response frequency adjustment through long short-term memory network;

[0051] Solution generation module: integrates the outputs of each module to generate an adaptive edge communication solution.

[0052] Preferably, the AI ​​computing module and energy efficiency optimization module are deployed on the edge node, and the remaining modules are deployed on the central cloud platform, and the two communicate through the 5G NR low-latency interface.

[0053] Preferably, the adaptive protocol engine includes a real-time monitoring unit, which continuously collects network congestion index and signal-to-noise ratio data and updates the Q-learning state space once per second.

[0054] The beneficial effects of the present invention are:

[0055] Demand forecasting models and neural network models are used to replace the traditional manual calculation or complex formula derivation of echo coefficients, reducing the computing power requirements of edge devices and achieving efficient deployment in resource-constrained environments. By dynamically eliminating abnormal frequency bands, high-quality aggregated data sets are generated, reducing the storage and processing load of edge nodes. The optimal transmission protocol is dynamically selected using real-time SNR level and network congestion index as the state space, reducing transmission delay and improving bandwidth utilization. Combined with the device resonant frequency generation solution, the communication bit error rate is reduced and the link stability is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Schematic diagram of the process of a 5G edge communication method based on big data in an embodiment of the present invention;

[0057] Figure 2 Schematic diagram of a module of a 5G edge communication system based on big data in an embodiment of the present invention;

[0058] Figure numerals: 10. Intelligent demand analysis module, 20. AI computing module, 30. Dynamic aggregation module, 40. Adaptive protocol engine, 41. Real-time detection unit, 50. Energy efficiency optimization module, 60. Environmental prediction module, 70. Solution generation module. DETAILED DESCRIPTION

[0059] The present invention is described in detail below with reference to the accompanying drawings and embodiments.

[0060] Example

[0061] A 5G edge communication method based on big data, comprising the following steps:

[0062] Determine the communication demand of the target communication area, analyze key communication indicators in the communication demand through a pre-trained demand prediction model, and extract indicator parameters.

[0063] The demand forecasting model is a regression model based on the TinyML architecture. Its training data includes historical communication area characteristics, device density, and environment type. The output is a target communication demand value, which replaces manual configuration of communication parameters. This lightweight demand forecasting model analyzes communication demand in the target communication area, avoiding the high latency caused by complex calculations, improving adaptability and reducing resource consumption.

[0064] Based on the index parameters, the echo coefficient is calculated using a neural network model, wherein the input of the neural network model includes the communication frequency, path loss, data transmission distance and signal-to-noise ratio of the index path, and the output is an approximate echo coefficient.

[0065] The neural network model is a one-dimensional convolutional neural network, and its calculation process satisfies:

[0066]

[0067] in, is the approximate echo coefficient, is the communication frequency, is the path loss, is the data transmission distance, is the signal-to-noise ratio;

[0068] The neural network model generates training data by simulating an existing echo coefficient formula.

[0069] By training lightweight neural network models such as one-dimensional convolutional neural networks (1D CNN), using relevant input data to output approximate echo coefficients, the defects of existing technical formulas are avoided, replacing the traditional manual calculation of echo coefficients or complex formula derivation. The model calculation error is lower, the computing power requirements of edge devices are reduced, and efficient deployment in resource-constrained environments is achieved.

[0070] The communication environment corresponding to the echo coefficient is queried, the communication parameters in the environment are collected and converted into a communication frequency band curve, and the frequency band parameters are extracted.

[0071] Dynamic anomaly detection is performed on the frequency band parameters using an online clustering algorithm, and an aggregated data set is generated after removing abnormal frequency bands.

[0072] The online clustering algorithm is an incremental K-means algorithm that dynamically adjusts the abnormal frequency band detection threshold, specifically including:

[0073] Receive frequency band parameter stream data in real time and update cluster centers in a sliding window manner;

[0074] Calculate the distance between each data point and the cluster center, and it will exceed the dynamic threshold The frequency band is marked as abnormal;

[0075]

[0076] in, For the The mean of the class, is the standard deviation.

[0077] By dynamically removing abnormal frequency bands through the incremental K-means algorithm and generating high-quality aggregated data sets, the amount of invalid data transmission can be reduced, and the storage and processing load of edge nodes can be lowered.

[0078] The aggregated data set is uploaded to the edge node, the aggregation factor is extracted, and the data transmission protocol is dynamically selected through the reinforcement learning model based on the real-time network status data.

[0079] The reinforcement learning model is a Q-learning algorithm, and its state space includes:

[0080] Real-time signal-to-noise ratio (SNR) level;

[0081] Network congestion index;

[0082] Aggregation factor type;

[0083] The action space is a set of optional data transmission protocols, and the reward function is defined as:

[0084]

[0085] in, To improve bandwidth resource utilization efficiency, For data transmission delay, and is the weight coefficient.

[0086] Through the Q-learning model, the optimal transmission protocol is dynamically selected with the real-time SNR level and network congestion index as the state space, which can significantly reduce transmission delay, improve bandwidth utilization, and adapt to network fluctuation scenarios (such as burst traffic, channel interference, etc.).

[0087] According to the selected data transmission protocol, the data transmission path of the 5G communication cloud platform is adjusted, the data transmission rate and frequency are monitored in real time, and the data bandwidth is calculated in combination with the energy efficiency constraint factor.

[0088] The energy efficiency constraint factor is calculated using the following formula to calculate the data bandwidth:

[0089]

[0090] in, represents the data bandwidth corresponding to the data in the aggregated data set, Represents the data item index corresponding to the data in the aggregated dataset, Indicates the number of data items corresponding to the data in the aggregated dataset, Indicates the data size, used to represent the amount of each data in the aggregated data set, represents the data transmission rate, represents the data frequency, The energy efficiency factor is dynamically calculated based on the remaining power of the device and historical power consumption data:

[0091]

[0092] in, is the current power consumption, is the average power consumption, is the maximum allowable power consumption.

[0093] Introducing dynamic energy efficiency factors Optimize data bandwidth (based on device remaining power and power consumption history) Allocation can extend device battery life and avoid resource overload in high power consumption states.

[0094] The device parameters of the edge communication device are configured based on the data bandwidth, the center power and the resonant frequency are extracted, the response frequency is calculated through the environmental prediction model, and an adaptive edge communication solution is generated.

[0095] The environmental prediction model is a long short-term memory network, the input is a historical environmental parameter sequence, and the output is a response frequency adjustment value. , the final response frequency is calculated as:

[0096]

[0097] in, It is calculated by the simplified formula from the center power and resonant frequency:

[0098]

[0099] in, is the inductor, is the capacitance, For resistance, simplified according to the device One third of the vibration circuit, Related to the resonant frequency, Related to center power.

[0100] Using the Long Short-Term Memory (LSTM) model to predict the response frequency adjustment , combined with the device resonant frequency to generate an adaptive solution, which can reduce the communication bit error rate and improve the link stability under sudden environmental changes (such as weather interference).

[0101] A 5G edge communication system based on big data, used to execute the 5G edge communication method based on big data as described in any of the above embodiments, comprising:

[0102] Intelligent demand analysis module 10: executes the demand forecasting model and outputs the communication demand target;

[0103] AI calculation module 20: integrates a neural network model to calculate the approximate echo coefficient;

[0104] Dynamic aggregation module 30: realizes abnormal frequency band detection and data aggregation through online clustering algorithm;

[0105] Adaptive protocol engine 40: Dynamically selects a data transmission protocol based on a reinforcement learning model; the adaptive protocol engine includes a real-time monitoring unit 41, which continuously collects network congestion index and signal-to-noise ratio data and updates the Q-learning state space once per second;

[0106] Energy efficiency optimization module 50: injects energy efficiency constraint factors to calculate data bandwidth and generates device configuration parameters;

[0107] Environmental prediction module 60: generating a response frequency adjustment through a long short-term memory network;

[0108] Solution generation module 70: integrates the outputs of each module to generate an adaptive edge communication solution.

[0109] Among them, the AI ​​computing module 20 and the energy efficiency optimization module 50 are deployed on the edge node, and the remaining modules are deployed on the central cloud platform, and the two communicate through the 5G NR low-latency interface.

[0110] This invention solves the problems of computational redundancy, adaptive rigidity, and resource waste in existing technologies by injecting three core innovations: lightweight AI computing, dynamic feedback mechanism, and energy efficiency constraint, and achieves step-by-step progress in "improved computing efficiency", "enhanced dynamic response", and "optimized energy efficiency ratio".

[0111] The above-described embodiments merely represent specific implementations of the present invention. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, and all such variations and improvements fall within the scope of protection of the present invention.

Claims

1. A 5G edge communication method based on big data, characterized in that: The following steps are involved: Determine the communication demand of the target communication area, analyze key communication indicators in the communication demand through a pre-trained demand prediction model, and extract indicator parameters; Based on the indicator parameters, an echo coefficient is calculated using a neural network model, wherein the input of the neural network model includes the communication frequency, path loss, data transmission distance and signal-to-noise ratio of the indicator path, and the output is an approximate echo coefficient; Querying the communication environment corresponding to the echo coefficient, collecting communication parameters in the environment and converting them into a communication frequency band curve, and extracting frequency band parameters; Performing dynamic anomaly detection on the frequency band parameters through an online clustering algorithm, and generating an aggregated data set after eliminating abnormal frequency bands; Uploading the aggregated data set to an edge node, extracting the aggregation factor, and dynamically selecting a data transmission protocol through a reinforcement learning model based on real-time network status data; Adjust the data transmission path of the 5G communication cloud platform according to the selected data transmission protocol, monitor the data transmission rate and frequency in real time, and calculate the data bandwidth based on the energy efficiency constraint factor; The device parameters of the edge communication device are configured based on the data bandwidth, the center power and the resonant frequency are extracted, the response frequency is calculated through the environmental prediction model, and an adaptive edge communication solution is generated.

2. The 5G edge communication method based on big data according to claim 1, characterized in that The demand forecasting model is a regression model based on the TinyML architecture. Its training data includes historical communication area characteristics, device density, and environment type. The output is the communication demand target value, which is used to replace manual configuration of communication parameters.

3. The 5G edge communication method based on big data according to claim 2, characterized in that: The neural network model is a one-dimensional convolutional neural network, and its calculation process satisfies: ； in, is the approximate echo coefficient, is the communication frequency, is the path loss, is the data transmission distance, is the signal-to-noise ratio; The neural network model generates training data by simulating an existing echo coefficient formula.

4. The 5G edge communication method based on big data according to claim 3, characterized in that The online clustering algorithm is an incremental K-means algorithm that dynamically adjusts the abnormal frequency band detection threshold, specifically including: Receive frequency band parameter stream data in real time and update cluster centers in a sliding window manner; Calculate the distance between each data point and the cluster center, and it will exceed the dynamic threshold The frequency band is marked as abnormal; ； in, For the The mean of the class, is the standard deviation.

5. The 5G edge communication method based on big data according to claim 4, characterized in that: The reinforcement learning model is a Q-learning algorithm, and its state space includes: Real-time signal-to-noise ratio (SNR) level; Network congestion index; Aggregation factor type; The action space is a set of optional data transmission protocols, and the reward function is defined as: ； in, To improve bandwidth resource utilization efficiency, For data transmission delay, and is the weight coefficient.

6. The 5G edge communication method based on big data according to claim 5, characterized in that: The energy efficiency constraint factor is calculated using the following formula to calculate the data bandwidth: ； in, represents the data bandwidth corresponding to the data in the aggregated data set, Represents the data item index corresponding to the data in the aggregated dataset, Indicates the number of data items corresponding to the data in the aggregated dataset, Indicates the data size, used to represent the amount of each data in the aggregated data set, represents the data transmission rate, represents the data frequency, The energy efficiency factor is dynamically calculated based on the remaining power of the device and historical power consumption data: ； in, is the current power consumption, is the average power consumption, is the maximum allowable power consumption.

7. The 5G edge communication method based on big data according to claim 6, characterized in that: The environmental prediction model is a long short-term memory network, the input is a historical environmental parameter sequence, and the output is a response frequency adjustment value. , the final response frequency is calculated as: ； in, Calculated from the center power and resonant frequency using a simplified formula.

8. A 5G edge communication system based on big data, characterized in that: Used to execute the 5G edge communication method based on big data as described in any one of claims 1 to 7, comprising: Intelligent demand analysis module: executes demand forecasting model and outputs communication demand targets; AI calculation module: integrates neural network model to calculate approximate echo coefficient; Dynamic aggregation module: realizes abnormal frequency band detection and data aggregation through online clustering algorithm; Adaptive protocol engine: Dynamically selects data transmission protocols based on reinforcement learning models; Energy efficiency optimization module: injects energy efficiency constraint factors to calculate data bandwidth and generate device configuration parameters; Environmental prediction module: Generates response frequency adjustment through long short-term memory network; Solution generation module: integrates the outputs of each module to generate an adaptive edge communication solution.

9. The 5G edge communication system based on big data according to claim 8, characterized in that The AI ​​computing module and energy efficiency optimization module are deployed on the edge node, and the remaining modules are deployed on the central cloud platform. The two communicate through the 5G NR low-latency interface.

10. The 5G edge communication system based on big data according to claim 8, characterized in that The adaptive protocol engine includes a real-time monitoring unit that continuously collects network congestion index and signal-to-noise ratio data and updates the Q-learning state space once per second.

Images