A wireless network communication interaction method and system based on cloud computing

Abstract

The present invention discloses a wireless network communication interaction method and system based on cloud computing, which relates to the cross-technical field of the Internet of Things and cloud computing, including collecting real-time data of wireless channel status, network traffic demand and device energy status; storing real-time data in the distributed storage architecture of the cloud computing platform to form a real-time data set and a historical data warehouse; using the real-time data set and the historical data warehouse in the cloud computing platform to build a deep learning model and generate prediction results; implementing dynamic coding methods, modulation mode adjustments, optimized routing strategies, and cross-layer resource interaction and collaborative optimization of network slices on edge cloud computing nodes based on dynamic coding adjustment parameter sets. The present invention realizes intelligent dynamic optimization of wireless network resources by integrating cloud computing, deep learning and genetic algorithms, improves spectrum utilization and energy efficiency, ensures high performance and stability of the network, provides a strong adaptive strategy for wireless communication interaction, and promotes significant performance enhancement.

Description

Technical Field

[0001] The present invention relates to the cross-technical field of Internet of Things and cloud computing, and in particular to a wireless network communication interaction method and system based on cloud computing. Background Art

[0002] In today's rapidly developing information technology era, wireless network communications, as the cornerstone of the information society, have always been the core issues of research and practice in terms of efficiency and reliability. With the rise and popularization of cloud computing technology, its powerful data processing capabilities and resource sharing characteristics have brought revolutionary changes to wireless network communications and promoted the continuous evolution of wireless network communication interaction methods. In particular, with the deep integration of the Internet of Things, big data, and artificial intelligence technologies, wireless network communication interaction methods based on cloud computing have become a key technical direction for improving network performance and optimizing resource allocation.

[0003] Existing technologies still face many challenges in achieving efficient and intelligent wireless network management and optimization. First, traditional methods often focus on static configuration and local optimization, which makes it difficult to cope with the dynamic changes in wireless channel status, sudden demands for network traffic, and uneven distribution of device energy consumption, resulting in unreasonable resource allocation and limited network performance. Second, the processing and analysis capabilities of massive data are insufficient, and there is a lack of effective data integration and analysis mechanisms to mine wireless network behavior patterns, which affects prediction accuracy and decision quality. In addition, existing systems have barriers in cross-layer resource management and coordination, and fail to fully utilize the flexibility and computing power advantages of cloud computing platforms, resulting in insufficient coordinated optimization of network slicing, routing strategies, and resource scheduling strategies. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a wireless network communication interaction method and system based on cloud computing to solve the problems of low efficiency and poor dynamic adaptability caused by static configuration in wireless network management and optimization, insufficient data processing and analysis capabilities affecting prediction accuracy, and poor cross-layer resource coordination limiting the potential of cloud computing.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a wireless network communication interaction method based on cloud computing, which includes collecting real-time data on wireless channel status, network traffic demand and device energy status; storing the real-time data in a distributed storage architecture of a cloud computing platform to form a real-time data set and a historical data warehouse; using the real-time data set and the historical data warehouse in the cloud computing platform to build a deep learning model and generate prediction results; executing a genetic algorithm based on the prediction results to dynamically generate scheduling instructions for spectrum allocation plans, power configuration guidance and cloud resource scheduling strategies; applying the scheduling instructions to edge cloud computing nodes, executing instruction parsing to output a dynamic coding adjustment parameter set; based on the dynamic coding adjustment parameter set, implementing dynamic coding methods, modulation mode adjustments, optimized routing strategies, and cross-layer resource interaction collaborative optimization of network slices on edge cloud computing nodes.

[0008] As a preferred solution of the wireless network communication interaction method based on cloud computing of the present invention, wherein: the real-time data of the wireless channel status, network traffic demand and device energy status are collected, and the specific steps are:

[0009] Use USRP devices combined with the OAI software stack to measure the wireless channel status in real time;

[0010] Integrate Wireshark and NetFlow network probe protocol analysis, deploy at network edge and core nodes, and analyze network traffic requirements;

[0011] Integrate IoT sensors into the power management architecture to monitor the energy status of devices;

[0012] The real-time measured wireless channel status, analyzed and monitored network traffic demand, and equipment energy status data are cleaned and formatted in a unified manner.

[0013] As a preferred solution of the wireless network communication interaction method based on cloud computing of the present invention, wherein: the real-time data is stored in the distributed storage architecture of the cloud computing platform to form a real-time data set and a historical data warehouse, the specific steps are:

[0014] Use the Kafka cluster to receive real-time data on wireless channel status, network traffic demand, and device energy status after data cleaning and format unification, and create independent KafkaTopic data for different types of real-time data;

[0015] Create an S3Bucket on the cloud computing platform, deploy and configure the S3Sink plug-in, and establish a data flow pipeline between KafkaTopic data and S3Bucket;

[0016] Integrate Parquet conversion logic in the S3Sink plugin configuration and convert Kafka Topic data flowing into S3Bucket into ApacheParquet format;

[0017] Build a table structure for each KafkaTopic data converted into ApacheParquet format;

[0018] Deploy Redshift cluster scale based on table structure architecture to form real-time data sets and historical data warehouses.

[0019] As a preferred solution of the wireless network communication interaction method based on cloud computing of the present invention, wherein: the real-time data set and the historical data warehouse in the cloud computing platform are used to build a deep learning model to generate prediction results, and the specific steps are as follows:

[0020] Extract characteristics of wireless channel status, network traffic demand, and device energy status from real-time data sets and historical data warehouses;

[0021] These features are sorted and merged in time series to form a high-dimensional time series dataset D, which is expressed as:

[0022] ;

[0023] ;

[0024] Among them, t i is the timestamp, x i is a vector containing multiple features of wireless channel status, network traffic demand and device energy status;

[0025] Based on the high-dimensional time series dataset D and the recursive feature elimination strategy, a feature importance evaluation function is constructed. , the expression is:

[0026] ;

[0027] in, is the jth feature of the i-th sample The mutual information between the target variable y, is the regularization coefficient, is the variance of the jth feature of the i-th sample;

[0028] Evaluate function based on feature importance , calculate the importance scores of all features, select the top N features with the highest scores, and generate a set of feature indexes selected after importance evaluation The expression is:

[0029] ;

[0030] ;

[0031] Among them, p i Feature index for filtering output;

[0032] Based on the selected feature index set after importance evaluation , capturing long-term dependencies in time series through LSTM networks;

[0033] Add an attention layer on top of the LSTM layer and incorporate the weight distribution function , the expression for constructing the LSTM-Attention model is:

[0034] ;

[0035] in, is the hidden state at the i-th moment after LSTM processing;

[0036] According to each time step i in the LSTM-Attention model, the hidden state weighted by the attention mechanism , forming a weighted hidden state set ;

[0037] Weight distribution function , the expression is:

[0038] ;

[0039] in, is the weight matrix of the attention layer, is the bias term;

[0040] Using the mean square error and incorporating the L2 regularization term, the loss function L is defined as:

[0041] ;

[0042] in, is the parameter vector of the loss function;

[0043] In the process of optimizing the loss function L, the Adam optimizer is used to iteratively update the model parameters and continuously adjust the model parameters. , to minimize the loss function L;

[0044] Use the trained LSTM-Attention model to predict new time series data and output the prediction results .

[0045] As a preferred solution of the wireless network communication interaction method based on cloud computing of the present invention, wherein: the genetic algorithm is executed based on the prediction results to dynamically generate scheduling instructions for spectrum allocation plan, power configuration guidance and cloud resource scheduling strategy, and the specific steps are:

[0046] The initial population expression is:

[0047] ;

[0048] in, For the The genetic code of each individual, G is the genetic code, For the Individuals’ first priority in spectrum allocation Gene loci, For the The kth gene position of an individual in dynamically generating spectrum allocation, For the The individual's lth gene position in power allocation guidance, For the The mth gene position of the individual in terms of resource allocation strategy;

[0049] Based on The genetic code of an individual Calculate the output fitness value The expression is:

[0050] ;

[0051] Among them, SE(G) is the spectrum efficiency, EE(G) is the energy efficiency, CRI(G) is the cloud resource utilization, STB(G) is the strategy stability measure, is the spectrum efficiency weight factor, is the energy efficiency weight factor, is the cloud resource utilization weight factor, is the weight factor for measuring the stability of the strategy;

[0052] Calculate the output fitness value Retain elite individuals and generate new populations through fitness-influenced selection, multi-point crossover, partial matching crossover and mutation ;

[0053] For new populations Calculating the adaptive mutation rate , the expression is:

[0054] ;

[0055] in, is the basic mutation rate, is the mutation rate adjustment coefficient;

[0056] In the iterative process of the genetic algorithm, the weight adaptive adjustment strategy based on time series analysis dynamically adjusts the weight factor of the fitness value , the expression is:

[0057] ;

[0058] in, At time t, The weight of the indicator, is the learning rate, In the current iteration cycle, The contribution of each indicator to the overall fitness improvement.

[0059] As a preferred solution of the wireless network communication interaction method based on cloud computing of the present invention, wherein: the scheduling instruction is applied to the edge cloud computing node, and the execution instruction is parsed to output the dynamic coding adjustment parameter set, and the specific steps are:

[0060] Obtaining a scheduling instruction set for spectrum allocation plans, power configuration guidance, and cloud resource scheduling strategies based on cloud computing;

[0061] Decrypt and integrity check the instruction set, convert it into JSON structured data format, and output spectrum allocation, power configuration, and cloud resource scheduling JSON structured instruction data;

[0062] The acquired structured instruction data is subjected to encoding strategy and dynamic encoding adjustment parameter analysis, and a dynamic encoding adjustment parameter set is output.

[0063] As a preferred solution of the wireless network communication interaction method based on cloud computing described in the present invention, wherein: the dynamic coding method, modulation mode adjustment, optimized routing strategy, and cross-layer resource interaction and collaborative optimization of network slices on the edge cloud computing node are implemented according to the dynamic coding adjustment parameter set, and the specific steps are:

[0064] Based on the dynamic coding adjustment parameter set, the coding adjustment parameters parsed are applied through real-time wireless channel status prediction and real-time network load of edge computing nodes to dynamically optimize the coding method of data transmission;

[0065] Combine the generated prediction results to output the channel conditions of signal-to-noise ratio and bandwidth availability, and dynamically configure the modulation mode;

[0066] Using the routing strategy optimized by genetic algorithm, the path selection algorithm of edge router is updated;

[0067] Automatically adjust the network slice resource allocation of the dynamic allocation of CPU, memory and storage resources on the edge computing nodes according to the cloud resource scheduling strategy;

[0068] Implement dynamic isolation between network slices and network function virtualization to dynamically increase or decrease the number of network function instances;

[0069] Based on the cross-layer resource interaction and collaborative optimization of dynamic adjustment of coding methods, modulation modes, routing strategies and network slicing, a cross-layer collaborative framework is constructed to enable the mutual coordination of wireless resource management physical layer and routing and load balancing network layer strategies, and to achieve cross-layer resource interaction and collaborative optimization of upper and lower layer information exchange and policy execution feedback through API interfaces or northbound interfaces.

[0070] In the second aspect, the present invention provides a wireless network communication interaction system based on cloud computing, including a data acquisition and storage module, a prediction module, a scheduling module, a network optimization instruction parsing module, and a cross-layer resource interaction and collaborative optimization module; the data acquisition and storage module is used to collect real-time data of wireless channel status, network traffic demand and device energy status, and store the real-time data in the distributed storage architecture of the cloud computing platform to form a real-time data set and a historical data warehouse; the prediction module is used to use the real-time data set and the historical data warehouse in the cloud computing platform to build a deep learning model and generate prediction results; the scheduling module is used to execute a genetic algorithm based on the prediction results to dynamically generate scheduling instructions for spectrum allocation plans, power configuration guidance and cloud resource scheduling strategies; the network optimization instruction parsing module is used to apply scheduling instructions to edge cloud computing nodes, execute instruction parsing to output a dynamic coding adjustment parameter set; the cross-layer resource interaction and collaborative optimization module is used to implement dynamic coding methods, modulation mode adjustments, optimized routing strategies, and cross-layer resource interaction and collaborative optimization of network slices on edge cloud computing nodes based on the dynamic coding adjustment parameter set.

[0071] In a third aspect, an embodiment of the present invention provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the wireless network communication interaction method based on cloud computing as described in the first aspect of the present invention is implemented.

[0072] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the cloud computing-based wireless network communication interaction method as described in the first aspect of the present invention.

[0073] The beneficial effects of the present invention are as follows: the present invention monitors the wireless environment in real time, collects channel status, traffic demand and equipment energy information, and uses cloud-based distributed storage to ensure data security and enhance processing capabilities; with the help of deep learning models, historical and real-time data are analyzed to accurately predict future network conditions and energy consumption trends; genetic algorithms are used to dynamically optimize spectrum allocation, power configuration and cloud resource management based on predictions to achieve flexible and efficient resource utilization; edge cloud nodes execute instructions accordingly, adjust encoding methods, modulation modes and routing strategies, optimize network slice resource coordination, and comprehensively improve network performance: speed up data transmission, improve energy efficiency and system stability, greatly optimize user experience, reduce costs, increase service flexibility, and innovate wireless network management and optimization practices. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0075] Figure 1 This is a flow chart of the wireless network communication interaction method based on cloud computing in Example 1.

[0076] Figure 2 This is a flow chart of executing instruction parsing and outputting a dynamic encoding adjustment parameter set in Example 1. DETAILED DESCRIPTION

[0077] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0078] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0079] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0080] Example 1, reference Figure 1 and Figure 2, which is the first embodiment of the present invention, and provides a wireless network communication interaction method based on cloud computing, comprising the following steps:

[0081] S1. Collect real-time data on wireless channel status, network traffic demand, and device energy status, and store the real-time data in the distributed storage architecture of the cloud computing platform to form a real-time data set and a historical data warehouse.

[0082] Furthermore, the USRP device is combined with the OAI software stack to measure the wireless channel status in real time;

[0083] Preferably, the USRP device is a general-purpose software radio peripheral that allows users to implement a variety of wireless communication functions in a software-defined manner. It can receive and send radio frequency signals, so it is very suitable for real-time measurement of wireless channel status, such as signal strength, frequency response, multipath propagation effects, etc. When selecting a USRP device, factors such as frequency band coverage, sampling rate, and processing power should be considered to ensure compatibility with the target wireless communication standard (such as LTE, 5G NR). For example, for 5G research, you may need to consider the USRP X310 or N3xx series because they provide higher bandwidth and a wider frequency range; the OAI software stack is an open source software stack that supports mobile communication standards such as LTE and 5G. When used with the USRP, OAI can drive the USRP device to perform physical layer signal processing and high-level protocol stack implementation, thereby achieving comprehensive and accurate measurement of the wireless channel. When configuring OAI (Open Air Interface), you need to plan the parameters of the eNodeB or gNodeB in detail, such as carrier frequency, bandwidth, MIMO configuration, etc., to ensure matching with the hardware, and adjust the high-level protocol stack parameters such as scheduling algorithms and power control strategies according to experimental or monitoring requirements;

[0084] Integrate Wireshark and NetFlow network probe protocol analysis, deploy at network edge and core nodes, and analyze network traffic requirements;

[0085] It should be noted that when deploying Wireshark, you can focus on specific types of data packets or protocols by writing custom display filters, such as analysis of application layer traffic such as HTTP and DNS, and combine the IO Graphs function to intuitively display the trend of traffic changes over time; NetFlow data collection should set a reasonable sampling rate to balance accuracy and resource consumption. NetFlow can also be integrated with flow analysis tools such as ntopng or Flowmon to provide more advanced traffic visualization and anomaly detection capabilities;

[0086] Integrate IoT sensors into the power management architecture to monitor the energy status of devices;

[0087] Preferably, the selection of sensors should be based on accuracy, power consumption, communication protocols (such as LoRa, Zigbee, NB-IoT) and environmental adaptability; for example, for long-range low-power requirements, LoRaWAN sensors can be selected; when designing sensor networks, consider redundant deployment to enhance data reliability and network robustness. At the same time, implement security protocols (such as TLS encryption) to ensure the security of communication between sensors and with the cloud;

[0088] Clean and unify the data format of the wireless channel status measured in real time, the network traffic demand and the device energy status data analyzed and monitored, including using Python, Spark or Flink and other tools to write data cleaning scripts, automatically perform missing value filling, outlier processing, data standardization and other steps; use Schema Registry or Avro and other technologies to maintain data model consistency, ensure that the format of data is unified when transmitted between different systems, and reduce data parsing errors;

[0089] Use the Kafka cluster to receive real-time data on wireless channel status, network traffic demand, and device energy status after data cleaning and formatting, and create independent Kafka topic data for different types of real-time data;

[0090] Preferably, Kafka Topic is a core concept in the Kafka messaging system, representing a logical classification of a type of message. Each Topic can have multiple producers publishing messages and multiple consumers subscribing to consume these messages. Creating separate Topics for different types of real-time data helps keep data organized and flexible in processing;

[0091] Create an S3 Bucket on the cloud computing platform, deploy and configure the S3 Sink plug-in, and establish a data flow pipeline between Kafka Topic data and S3 Bucket;

[0092] Integrate Parquet conversion logic in the S3 Sink plugin configuration and convert Kafka Topic data flowing into the S3 Bucket into Apache Parquet format;

[0093] Preferably, S3 Bucket is the basic container for storing objects (files) in Amazon S3. By configuring the S3Sink plugin, data in Kafka can be automatically exported to S3 for persistent storage and subsequent analysis.

[0094] Build a table structure for each Kafka Topic data converted into Apache Parquet format;

[0095] Preferably, Apache Parquet is a columnar storage format that is particularly suitable for big data analysis because it can efficiently read and compress large numbers of data columns. Converting to Parquet format can optimize query performance, especially on data warehouse systems such as Redshift.

[0096] Deploy Redshift cluster scale based on table structure architecture to form real-time data sets and historical data warehouses.

[0097] Preferably, Redshift cluster is a fully managed data warehouse service provided by Amazon, which is particularly suitable for processing large-scale data analysis queries. It builds table structures based on data in Parquet format and deploys corresponding Redshift clusters, which can efficiently analyze and report on real-time data sets and historical data warehouses;

[0098] It should also be noted that by combining USRP and OAI to accurately measure the status of wireless channels, integrating Wireshark and NetFlow for in-depth analysis of network traffic, and using IoT sensors to monitor the energy status of devices, the system can collect key data in real time. Subsequently, by using data cleaning, format unification, and Kafka technologies, the real-time data can be efficiently aggregated and distributed to the cloud-based distributed storage system. The data is stored in S3 in Apache Parquet format, which optimizes the query performance in the Redshift data warehouse, thus providing a solid data foundation for network operation and maintenance, resource management, and future planning.

[0099] S2. Use the real-time data set and historical data warehouse in the cloud computing platform to build a deep learning model and generate prediction results.

[0100] Furthermore, the characteristics of wireless channel status, network traffic demand and device energy status are extracted from the real-time data set and historical data warehouse;

[0101] These features are sorted and merged in time series to form a high-dimensional time series dataset D, which is expressed as:

[0102] ;

[0103] ;

[0104] Among them, t i is the timestamp, x i is a vector containing multiple features of wireless channel status, network traffic demand and device energy status;

[0105] Based on the high-dimensional time series dataset D and the recursive feature elimination strategy, a feature importance evaluation function is constructed. , the expression is:

[0106] ;

[0107] in, is the jth feature of the i-th sample The mutual information between the target variable y, The regularization coefficient is used to penalize features with large variance to prevent the model from over-relying on features with large noise. is the variance of the jth feature of the i-th sample, and MI represents the measure of the feature The correlation between the feature and the target variable y, the higher the mutual information, the greater the information contribution of the feature to the prediction of the target y;

[0108] Evaluate function based on feature importance , calculate the importance scores of all features, select the top N features with the highest scores, and generate a set of feature indexes selected after importance evaluation The expression is:

[0109] ;

[0110] ;

[0111] Among them, p i Feature index for filtering output;

[0112] Based on the selected feature index set after importance evaluation , capturing long-term dependencies in time series through LSTM networks;

[0113] Add an attention layer on top of the LSTM layer and incorporate the weight distribution function , the expression for constructing the LSTM-Attention model is:

[0114] ;

[0115] in, is the hidden state at the i-th moment after LSTM processing;

[0116] Preferably, the weight distribution function Combining mutual information and feature variance for scoring not only considers the relevance of features to the target, but also penalizes features with large variance, helping to select the features with the greatest predictive value;

[0117] According to each time step i in the LSTM-Attention model, the hidden state weighted by the attention mechanism , forming a weighted hidden state set ;

[0118] Weight distribution function , the expression is:

[0119] ;

[0120] in, is the weight matrix of the attention layer, is the bias term, and jointly determine how to use the hidden state Calculate attention weights;

[0121] Using the mean square error and incorporating the L2 regularization term, the loss function L is defined as:

[0122] ;

[0123] in, is the parameter vector of the loss function;

[0124] In the process of optimizing the loss function L, the Adam optimizer is used to iteratively update the model parameters and continuously adjust the model parameters. , to minimize the loss function L;

[0125] Use the trained LSTM-Attention model to predict new time series data and output the prediction results .

[0126] S3. Execute the genetic algorithm based on the prediction results to dynamically generate scheduling instructions for spectrum allocation plans, power configuration guidance, and cloud resource scheduling strategies.

[0127] Furthermore, the initial population expression is constructed as:

[0128] ;

[0129] in, For the The genetic code of an individual is composed of four dimensions, corresponding to different optimization objectives. G is the genetic code. For the Individuals’ first priority in spectrum allocation The gene position determines the individual's allocation strategy on spectrum resources. For the The kth gene position of an individual in dynamically generating spectrum allocation controls the power allocation and adjustment strategy. For the The individual's lth gene position in power configuration guidance reflects how cloud resources are scheduled and optimized. For the The mth gene position of an individual in terms of resource scheduling strategy involves the overall scheduling logic and strategy of network resources;

[0130] Based on The genetic code of an individual Calculate the output fitness value The expression is:

[0131] ;

[0132] Among them, SE(G) is the spectrum efficiency, EE(G) is the energy efficiency, CRI(G) is the cloud resource utilization, STB(G) is the strategy stability measure, is the spectrum efficiency weight factor, is the energy efficiency weight factor, is the cloud resource utilization weight factor, is the weight factor for measuring the stability of the strategy;

[0133] It should be noted that the fitness value Represents the fitness value of the current individual. The mutation rate of individuals with high fitness may decrease, maintaining excellent characteristics; on the contrary, the mutation rate of individuals with low fitness may increase, encouraging the exploration of new solutions;

[0134] Calculate the output fitness value Retain elite individuals and generate new populations through fitness-influenced selection, multi-point crossover, partial matching crossover and mutation ;

[0135] For new populations Calculating the adaptive mutation rate , which aims to balance exploration (finding new solutions) and exploitation (optimizing existing solutions) is expressed as:

[0136] ;

[0137] in, As the basic mutation rate, ensure that the population has a certain degree of diversity. is the mutation rate adjustment coefficient;

[0138] In the iterative process of the genetic algorithm, the weight adaptive adjustment strategy based on time series analysis dynamically adjusts the weight factor of the fitness value , the expression is:

[0139] ;

[0140] in, At time t, The weight of the indicator, The learning rate determines the magnitude of weight adjustment. If it is too large, the weight will fluctuate greatly and will not converge easily. If it is too small, the adjustment will be slow. In the current iteration cycle, The contribution of each indicator to the overall fitness improvement.

[0141] Preferably, and Respectively represent the The weight of each indicator indicates that the weight is adaptively optimized over time, which enhances the flexibility and adaptability of the algorithm.

[0142] It should be noted that in the iterative process of the genetic algorithm, the weight factor It will be dynamically adjusted according to the results of time series analysis, through the learning rate And the contribution of each indicator to the overall fitness improvement To update, the algorithm can automatically adjust the emphasis on different performance indicators according to the current network status and past optimization experience, so as to achieve better optimization effect and faster convergence speed;

[0143] It should also be noted that genetic algorithms have demonstrated flexibility and efficiency in solving complex optimization problems. Moreover, through careful coding design, construction of fitness functions, and dynamic adjustment strategies, they ensure precise control and optimization in network resource management, thereby effectively addressing the challenges of resource allocation and strategy formulation in wireless networks.

[0144] S4. Apply the scheduling instructions to the edge cloud computing nodes, execute the instruction parsing to output the dynamic coding adjustment parameter set, implement the dynamic coding method, modulation mode adjustment, optimized routing strategy, and cross-layer resource interaction and collaborative optimization of network slices on the edge cloud computing nodes.

[0145] Furthermore, the scheduling instruction set for spectrum allocation plan, power configuration guidance, and cloud resource scheduling strategy is obtained based on cloud computing;

[0146] Decrypt and integrity check the instruction set, convert it into JSON structured data format, and output spectrum allocation, power configuration, and cloud resource scheduling JSON structured instruction data;

[0147] Perform encoding strategy and dynamic encoding adjustment parameter analysis on the acquired structured instruction data, and output a dynamic encoding adjustment parameter set;

[0148] Preferably, the high-level scheduling instructions generated from the cloud are transmitted to the edge nodes, which is usually done through a secure communication protocol, such as TLS / SSL, to ensure the security and integrity of the instructions during transmission. After arriving at the edge node, the instructions need to be processed by a dedicated parsing module, which is responsible for identifying the instruction type (spectrum allocation, power configuration, cloud resource scheduling, etc.) and preparing the corresponding execution strategy.

[0149] Furthermore, based on the dynamic coding adjustment parameter set, through the real-time wireless channel state prediction and the real-time network load of the edge computing node, the parsed coding adjustment parameters are applied (the parsing process may involve complex algorithm logic to ensure that the selected parameters can maximize data transmission efficiency and adapt to real-time changes in the network), and the coding method of data transmission is dynamically optimized;

[0150] Combine the generated prediction results with the output signal-to-noise ratio and channel conditions of bandwidth availability, and dynamically configure the modulation mode (such as BPSK, QAM). The goal of the adjustment is to maximize transmission efficiency while maintaining data integrity and low latency.

[0151] Using the routing strategy optimized by the genetic algorithm, the path selection algorithm of the edge router is updated. This usually involves updating the router's routing table or FIB (Forwarding Information Base) to ensure that data packets are forwarded along the new optimized path, thereby improving network latency and throughput.

[0152] Automatically adjust the network slice resource allocation of the dynamic allocation of CPU, memory and storage resources on the edge computing nodes according to the cloud resource scheduling strategy;

[0153] Implement dynamic isolation between network slices and network function virtualization to dynamically increase or decrease the number of network function instances;

[0154] Preferably, dynamic isolation technology ensures that resource usage between different slices does not interfere with each other, while dynamically increasing or decreasing the number of network function instances (such as vEPC, vRAN) according to demand to cope with fluctuations in service demand;

[0155] Based on the cross-layer resource interaction and collaborative optimization of dynamic adjustment of coding methods, modulation modes, routing strategies and network slicing, a cross-layer collaborative framework is constructed to enable the mutual coordination of wireless resource management physical layer and routing and load balancing network layer strategies, and to achieve cross-layer resource interaction and collaborative optimization of upper and lower layer information exchange and policy execution feedback through API interfaces or northbound interfaces.

[0156] It should be noted that the purpose of building a cross-layer collaborative framework is to achieve close collaboration between the physical layer (such as wireless resource management) and the network layer (such as routing, load balancing) strategies. Through the API interface or northbound interface, high-level strategies can guide the allocation of underlying resources, and the feedback of the underlying status can affect the adjustment of upper-level strategies, forming a closed-loop control and achieving optimization of overall network performance.

[0157] The present embodiment also provides a wireless network communication interaction system based on cloud computing, including: a data acquisition and storage module, a prediction module, a scheduling module, a network optimization instruction parsing module, and a cross-layer resource interaction and collaborative optimization module; the data acquisition and storage module is used to collect real-time data of wireless channel status, network traffic demand and device energy status, and store the real-time data in the distributed storage architecture of the cloud computing platform to form a real-time data set and a historical data warehouse; the prediction module is used to use the real-time data set and the historical data warehouse in the cloud computing platform to build a deep learning model and generate prediction results; the scheduling module is used to execute a genetic algorithm based on the prediction results to dynamically generate scheduling instructions for spectrum allocation plans, power configuration guidance and cloud resource scheduling strategies; the network optimization instruction parsing module is used to apply scheduling instructions to edge cloud computing nodes, execute instruction parsing to output a dynamic coding adjustment parameter set; the cross-layer resource interaction and collaborative optimization module is used to implement dynamic coding methods, modulation mode adjustments, optimized routing strategies, and cross-layer resource interaction and collaborative optimization of network slices on edge cloud computing nodes based on the dynamic coding adjustment parameter set.

[0158] This embodiment also provides a computer device, which is suitable for the case of a wireless network communication interaction method based on cloud computing, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the wireless network communication interaction method based on cloud computing proposed in the above embodiment.

[0159] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0160] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the method for realizing wireless network communication interaction based on cloud computing proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0161] In summary, the present invention constructs an advanced, dynamic and accurate wireless network communication management system by integrating real-time monitoring of wireless channels, cloud computing platform data integration, deep learning-driven prediction models, genetic algorithm optimization strategies and edge computing dynamic execution mechanisms. It significantly improves service flexibility, optimizes wireless network performance, enhances user experience and reduces operating costs. It is of great significance and has broad application prospects.

[0162] Example 2, referring to Table 1, is the second example of the present invention. To further verify the technical solution of the present invention, experimental simulation data of a wireless network communication interaction method based on cloud computing is provided.

[0163] A simulation environment with various network topologies and load conditions was built to simulate complex wireless network scenarios in the real world. The environment configuration includes but is not limited to the following: Hardware facilities: multiple servers with similar configurations are used as edge cloud computing nodes, and each node is equipped with a high-performance GPU to accelerate the training and prediction of deep learning models; Software stack: Integrate OAI stack and USRP devices to accurately measure the status of wireless channels; Deploy Wireshark and customized NetFlow probes to accurately analyze network traffic; Integrate IoT sensor systems to monitor device energy consumption; Data stream processing: Use Kafka clusters, S3 storage services and Redshift databases to ensure efficient data collection, storage and processing; Models and algorithms: Implement the above-mentioned deep learning models and genetic algorithm processes, especially the training and optimization of the LSTM-Attention model, and the iterative strategy adjustment of the genetic algorithm.

[0164] First, to ensure the real-time and consistency of the data, a timestamp synchronization mechanism was adopted to ensure the unified time of all network nodes, sensor devices and data collection tools. Data on wireless channel status, network traffic demand and device energy status were collected every minute for one week. Tools such as OAI and Wireshark were used to ensure data accuracy and timeliness, and complete data cleaning and format standardization.

[0165] Secondly, the collected real-time data is stored in the cloud using the Kafka-S3 data pipeline, and a historical and real-time data warehouse is built based on Redshift. Based on this data set, the LSTM-Attention model is trained, and the model parameters are finally determined and predictions are made through recursive feature elimination and importance evaluation.

[0166] Next, based on the prediction results, a genetic algorithm is executed to dynamically generate spectrum allocation plans, power configuration guidance, and cloud resource scheduling strategies. Subsequently, these instructions are parsed and applied to edge cloud computing nodes to implement optimization measures such as dynamic coding adjustment.

[0167] Finally, the network performance indicators before and after implementation were recorded and compared, including but not limited to spectrum efficiency, energy efficiency, cloud resource utilization, and policy stability.

[0168] The details are shown in the following table:

[0169] Table 1 Experimental record table

[0170]

[0171] Through the analysis of the data in the above table, it can be clearly seen that the effectiveness and superiority of the method of the present invention. The spectrum efficiency has increased by 50% (from 1.2bps / Hz to 1.8 bps / Hz), which is mainly attributed to the LSTM-Attention model's more accurate prediction of the wireless channel state, making the spectrum allocation more reasonable, reducing interference and retransmission, and thus improving the data transmission efficiency; the energy efficiency has increased by 67% (from 0.9bits / Joule to 1.5bits / Joule), thanks to the optimization of power configuration guidance, the device reduces energy consumption while meeting the communication quality, which is particularly important for mobile devices and remote nodes; the cloud resource utilization rate has increased by 23% (from 65% to 80%), indicating that through the dynamic resource scheduling strategy, the present invention can more effectively allocate and utilize computing resources, reduce resource idleness, and improve the overall service quality and response speed; the strategy stability index has increased from 0.7 to 0.9, indicating that the cloud resource scheduling and network strategy adjustment of the present invention are smoother and more stable, reducing network fluctuations and ensuring service quality.

[0172] Through the innovative combination of deep learning and genetic algorithms in the present invention, not only a significant performance improvement is achieved at the technical level, but also the problems of spectrum resource waste, energy inefficiency, uneven cloud resource allocation and unstable policy execution in the existing technology are solved, showing a high degree of innovation and practicality; this series of optimization measures jointly promote the efficient, green and intelligent development of wireless network communications, and have important reference value for the future 5G and later network architecture design.

[0173] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A wireless network communication interaction method based on cloud computing, characterized in that: include, Collect real-time data on wireless channel status, network traffic demand, and device energy status; Store real-time data in the distributed storage architecture of the cloud computing platform to form a real-time data set and historical data warehouse; Use real-time data sets and historical data warehouses in cloud computing platforms to build deep learning models and generate prediction results; Based on the prediction results, a genetic algorithm is executed to dynamically generate scheduling instructions for spectrum allocation plans, power configuration guidance, and cloud resource scheduling strategies; Apply the scheduling instructions to the edge cloud computing node, execute the instruction parsing and output the dynamic encoding adjustment parameter set; According to the dynamic coding adjustment parameter set, implement dynamic coding methods, modulation mode adjustment, optimized routing strategy, and cross-layer resource interaction and collaborative optimization of network slices on edge cloud computing nodes; The real-time data set and historical data warehouse in the cloud computing platform are used to build a deep learning model and generate prediction results. The specific steps are as follows: Extract characteristics of wireless channel status, network traffic demand, and device energy status from real-time data sets and historical data warehouses; These features are sorted and merged in time series to form a high-dimensional time series dataset D, which is expressed as: ； ； Among them, t i is the timestamp, x i is a vector containing multiple features of wireless channel status, network traffic demand and device energy status; Based on the high-dimensional time series dataset D and the recursive feature elimination strategy, a feature importance evaluation function is constructed. , the expression is: ； in, is the jth feature of the i-th sample The mutual information between the target variable y, is the regularization coefficient, is the variance of the jth feature of the i-th sample; Evaluate function based on feature importance , calculate the importance scores of all features, select the top N features with the highest scores, and generate a set of feature indexes selected after importance evaluation The expression is: ； ； Among them, p i Feature index for filtering output; Based on the selected feature index set after importance evaluation , capturing long-term dependencies in time series through LSTM networks; Add an attention layer on top of the LSTM layer and incorporate the weight distribution function , the expression for constructing the LSTM-Attention model is: ； in, is the hidden state at the i-th moment after LSTM processing; According to each time step i in the LSTM-Attention model, the hidden state weighted by the attention mechanism , forming a weighted hidden state set ; Weight distribution function , the expression is: ； in, is the weight matrix of the attention layer, b a is the bias term; Using the mean square error and incorporating the L2 regularization term, the loss function L is defined as: ； in, is the parameter vector of the loss function; In the process of optimizing the loss function L, the Adam optimizer is used to iteratively update the model parameters and continuously adjust the model parameters. , to minimize the loss function L; Use the trained LSTM-Attention model to predict the new time series data and output the prediction result Y pred ; The genetic algorithm is executed based on the prediction results to dynamically generate scheduling instructions for spectrum allocation plan, power configuration guidance and cloud resource scheduling strategy, and the specific steps are: The initial population expression is: ； in, For the The genetic code of each individual, G is the genetic code, For the Individuals’ first priority in spectrum allocation Gene loci, For the The kth gene position of an individual in dynamically generating spectrum allocation, For the The individual's lth gene position in power allocation guidance, For the The mth gene position of the individual in terms of resource allocation strategy; Based on The genetic code of an individual Calculate the output fitness value The expression is: ； Among them, SE(G) is the spectrum efficiency, EE(G) is the energy efficiency, CRI(G) is the cloud resource utilization, STB(G) is the strategy stability measure, is the spectrum efficiency weight factor, is the energy efficiency weight factor, is the cloud resource utilization weight factor, is the weight factor for measuring the stability of the strategy; Calculate the output fitness value Retain elite individuals and generate new populations through fitness-influenced selection, multi-point crossover, partial matching crossover and mutation ; For new populations Calculating the adaptive mutation rate , the expression is: ； in, is the basic mutation rate, is the mutation rate adjustment coefficient; In the iterative process of the genetic algorithm, the weight adaptive adjustment strategy based on time series analysis dynamically adjusts the weight factor of the fitness value , the expression is: ； in, At time t, The weight of the indicator, is the learning rate, In the current iteration cycle, The contribution of each indicator to the overall fitness improvement.

2. The wireless network communication interaction method based on cloud computing according to claim 1, characterized in that: The specific steps of collecting real-time data of wireless channel status, network traffic demand and device energy status are as follows: Use USRP devices combined with the OAI software stack to measure the wireless channel status in real time; Integrate Wireshark and NetFlow network probe protocol analysis, deploy at network edge and core nodes, and analyze network traffic requirements; Integrate IoT sensors into the power management architecture to monitor the energy status of devices; The real-time measured wireless channel status, analyzed and monitored network traffic demand, and equipment energy status data are cleaned and formatted in a unified manner.

3. The wireless network communication interaction method based on cloud computing according to claim 2, characterized in that: The real-time data is stored in the distributed storage architecture of the cloud computing platform to form a real-time data set and a historical data warehouse. The specific steps are as follows: Use the Kafka cluster to receive real-time data on wireless channel status, network traffic demand, and device energy status after data cleaning and formatting, and create independent Kafka topic data for different types of real-time data; Create an S3Bucket on the cloud computing platform, deploy and configure the S3Sink plug-in, and establish a data flow pipeline between KafkaTopic data and S3Bucket; Integrate Parquet conversion logic in the S3 Sink plugin configuration and convert Kafka Topic data flowing into the S3 Bucket into Apache Parquet format; Build a table structure for each KafkaTopic data converted into ApacheParquet format; Deploy Redshift cluster scale based on table structure architecture to form real-time data sets and historical data warehouses.

4. The wireless network communication interaction method based on cloud computing according to claim 3, characterized in that: The specific steps of applying the scheduling instruction to the edge cloud computing node and executing the instruction parsing to output the dynamic encoding adjustment parameter set are as follows: Obtaining a scheduling instruction set for spectrum allocation plans, power configuration guidance, and cloud resource scheduling strategies based on cloud computing; Decrypt and integrity check the instruction set, convert it into JSON structured data format, and output spectrum allocation, power configuration, and cloud resource scheduling JSON structured instruction data; The acquired structured instruction data is subjected to encoding strategy and dynamic encoding adjustment parameter analysis, and a dynamic encoding adjustment parameter set is output.

5. The wireless network communication interaction method based on cloud computing according to claim 4, characterized in that: The specific steps of implementing the dynamic coding mode, modulation mode adjustment, optimized routing strategy, and cross-layer resource interaction and collaborative optimization of network slices on edge cloud computing nodes according to the dynamic coding adjustment parameter set are as follows: Based on the dynamic coding adjustment parameter set, the coding adjustment parameters parsed are applied through real-time wireless channel status prediction and real-time network load of edge computing nodes to dynamically optimize the coding method of data transmission; Combine the generated prediction results to output the channel conditions of signal-to-noise ratio and bandwidth availability, and dynamically configure the modulation mode; Using the routing strategy optimized by genetic algorithm, the path selection algorithm of edge router is updated; Automatically adjust the network slice resource allocation of the dynamic allocation of CPU, memory and storage resources on the edge computing nodes according to the cloud resource scheduling strategy; Implement dynamic isolation between network slices and network function virtualization to dynamically increase or decrease the number of network function instances; Based on the cross-layer resource interaction and collaborative optimization of dynamic adjustment of coding methods, modulation modes, routing strategies and network slicing, a cross-layer collaborative framework is constructed to enable the mutual coordination of wireless resource management physical layer and routing and load balancing network layer strategies, and to achieve cross-layer resource interaction and collaborative optimization of upper and lower layer information exchange and policy execution feedback through API interfaces or northbound interfaces.

6. A wireless network communication interaction system based on cloud computing, based on the wireless network communication interaction method based on cloud computing according to any one of claims 1 to 5, characterized in that: It includes data collection and storage module, prediction module, scheduling module, network optimization instruction parsing module, cross-layer resource interaction and collaborative optimization module; The data acquisition and storage module is used to collect real-time data on wireless channel status, network traffic demand and device energy status, and store the real-time data in the distributed storage architecture of the cloud computing platform to form a real-time data set and a historical data warehouse; The prediction module is used to build a deep learning model using real-time data sets and historical data warehouses in the cloud computing platform to generate prediction results; The scheduling module is used to execute the genetic algorithm based on the prediction results to dynamically generate scheduling instructions for spectrum allocation plans, power configuration guidance and cloud resource scheduling strategies; The network optimization instruction parsing module is used to apply the scheduling instruction to the edge cloud computing node, execute the instruction parsing and output the dynamic encoding adjustment parameter set; The cross-layer resource interaction and collaborative optimization module is used to implement dynamic coding methods, modulation mode adjustment, optimized routing strategies, and cross-layer resource interaction and collaborative optimization of network slices on edge cloud computing nodes based on dynamic coding adjustment parameter sets.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the wireless network communication interaction method based on cloud computing described in any one of claims 1 to 5 are implemented.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the wireless network communication interaction method based on cloud computing described in any one of claims 1 to 5 are implemented.

Images