Cloud computing environment-oriented communication equipment collaborative modeling and data transmission optimization method and system

Abstract

The invention belongs to the field of cloud computing, and discloses a cloud computing environment-oriented communication equipment collaborative modeling and data transmission optimization method and system, and the method comprises the steps: carrying out the collection and modeling of feature data of communication equipment; optimizing an intelligent data transmission path; preprocessing data based on edge calculation; carrying out cooperative scheduling on calculation and communication resources; and deploying a self-adaptive network optimization strategy. Through intelligent collaborative modeling, dynamic data scheduling optimization, edge computing and SDN control, efficient collaborative modeling and data transmission optimization of communication equipment in a cloud computing environment are realized. The method has remarkable advantages in the aspects of improving the network utilization rate, reducing the data transmission time delay and optimizing computing resource allocation, and provides technical support for cloud computing and intelligent network optimization.

Description

Technical Field

[0001] The present invention belongs to, but is not limited to, the field of cloud computing technology, and particularly relates to a method and system for collaborative modeling and data transmission optimization of communication equipment for cloud computing environments. Background Art

[0002] In a cloud computing environment, the collaborative modeling and data transmission optimization of communication equipment mainly involve core technologies such as distributed computing, network topology optimization, data scheduling, and traffic control. The most relevant existing technology currently is the distributed data transmission optimization technology based on software-defined networking (SDN). This technology utilizes the centralized control characteristics of SDN to improve the efficiency and reliability of data transmission. However, this technology still faces the following key problems that need to be urgently solved:

[0003] (1) Lack of intelligent adaptive optimization ability

[0004] Existing SDN solutions mostly adopt static or history data-based traffic scheduling strategies, which are difficult to cope with sudden network traffic changes, resulting in low bandwidth utilization and even possible data congestion and packet loss problems.

[0005] (2) Insufficient collaborative modeling between communication equipment

[0006] Existing methods usually independently optimize the communication strategies of each device, lacking collaborative modeling from a global perspective and unable to fully utilize the computing and storage capabilities between devices for optimization.

[0007] (3) Imperfect collaborative scheduling of computing and communication resources

[0008] Existing methods cannot fully coordinate computing resources and communication resources, resulting in unbalanced resource allocation between computing tasks and data transmission and affecting the overall performance.

[0009] In view of the above analysis, the technical problems that urgently need to be solved in the existing technology are:

[0010] The existing technology has problems such as lack of intelligent adaptive optimization ability, insufficient collaborative modeling between communication equipment, and imperfect collaborative scheduling of computing and communication resources. Summary of the Invention

[0011] In view of the problems existing in the existing technology, the present invention provides a method and system for collaborative modeling and data transmission optimization of communication equipment for cloud computing environments.

[0012] The present invention is implemented as follows. A method for collaborative modeling and data transmission optimization of communication equipment for cloud computing environments is characterized in that the method for collaborative modeling and data transmission optimization of communication equipment for cloud computing environments specifically includes:

[0013] S1: Collection and Modeling of Communication Equipment Feature Data. Collect the key parameters of each communication equipment, use a deep neural network (DNN) to model the equipment features, and dynamically update its status;

[0014] S2: Optimization of Intelligent Data Transmission Path. Use reinforcement learning (RL) to train an optimization model for the data transmission path, dynamically adjust the data transmission path according to the real-time network status, and combine with the SDN controller to dynamically adjust the traffic allocation strategy;

[0015] S3: Data Preprocessing Based on Edge Computing. Perform data preprocessing and compression on edge devices, and perform distributed computing according to task requirements;

[0016] S4: Cooperative Scheduling of Computing and Communication Resources. Adopt a resource-aware task scheduling algorithm, consider network bandwidth and computing resources simultaneously when allocating computing tasks, predict the load situation by combining historical data and real-time status, and achieve adaptive adjustment;

[0017] S5: Deployment of Adaptive Network Optimization Strategy. Combine the centralized control characteristics of SDN, deploy the optimization strategy to the network control plane, and use the federated learning method to achieve collaborative optimization between multiple devices.

[0018] Furthermore, in S1, the key parameters of each communication equipment include network status, CPU / GPU load, storage occupancy, data throughput rate, etc.

[0019] Furthermore, in S2, use deep reinforcement learning to train an intelligent scheduling model:

[0020] State (S): Current network topology, task load;

[0021] Action (A): Select the best data transmission path;

[0022] Reward (R): Minimize transmission delay and maximize network bandwidth utilization;

[0023] Use Dueling DQN to optimize the learning efficiency.

[0024] Furthermore, in S3, edge devices include intelligent gateways, edge servers, etc.

[0025] Furthermore, in S5, use the SDN controller for global traffic scheduling. The control plane is responsible for calculating the optimal data path and issuing flow tables; the data plane is responsible for executing data forwarding; combine with the OpenFlow protocol to dynamically adjust the packet forwarding strategy; use federated learning to train local models on different edge nodes and summarize them in the cloud.

[0026] Another object of the present invention is to provide a communication equipment collaborative modeling and data transmission optimization system for a cloud computing environment, which specifically includes:

[0027] A collaborative modeling layer, which adopts distributed device feature modeling, uses deep learning methods to extract features such as the operating status, bandwidth occupancy, and computing power of communication equipment, forms a dynamically updated collaborative modeling system, and optimizes the modeling process through reinforcement learning to enable it to adapt to different computing loads and data transmission requirements;

[0028] An intelligent optimization layer, which adopts an adaptive data scheduling strategy based on "reinforcement learning (RL) and deep neural network (DNN)", adjusts the data transmission path and traffic priority in real time, and combines edge computing technology to perform preprocessing and splitting at nodes close to the data source;

[0029] A resource collaborative scheduling layer, which adopts a joint optimization scheduling strategy for computing and communication resources, considers both computing resources and communication bandwidth during task allocation, improves resource utilization, and combines the characteristics of 5G / 6G networks to achieve efficient and low-latency data transmission.

[0030] Furthermore, the collaborative modeling layer includes distributed device feature modeling and reinforcement learning optimization modeling:

[0031] (1) Adopt a distributed data acquisition module to obtain the operating parameters of communication equipment:

[0032] Network status parameters: bandwidth occupancy rate, data throughput, packet loss rate, etc.;

[0033] Computing resource parameters: CPU / GPU occupancy rate, storage occupancy rate;

[0034] Task load conditions: current task execution status, data traffic characteristics;

[0035] Use a deep neural network (DNN) for feature extraction, map the state of the device to a feature vector, and build a dynamically updated device model;

[0036] (2) Adopt reinforcement learning (RL) + collaborative learning methods to enable the system to adapt to dynamically changing computing loads and data transmission requirements, and optimize the collaborative model of the device through the Policy Gradient algorithm.

[0037] Furthermore, in the intelligent optimization layer, the adaptive data scheduling strategy uses a deep Q network (DQN) for data scheduling decisions:

[0038] Input: current network topology, task load, device status;

[0039] Output: optimal data transmission path and traffic priority;

[0040] In combination with the SDN controller, the traffic scheduling policy is adjusted in real time to dynamically change the data packet forwarding path;

[0041] Data preprocessing is performed at the edge nodes, including: data compression, task shunting, and data aggregation; the edge layer is used for preprocessing data, and the cloud computing layer is used for performing complex calculations, such as deep learning inference tasks.

[0042] Furthermore, the resource collaborative scheduling layer includes:

[0043] (1) Joint optimization scheduling of computing and communication resources

[0044] Reinforcement learning (Actor-Critic model) is used to optimize the computing task allocation:

[0045] Agent: Task scheduling management module;

[0046] Environment: Current network status, computing resource occupancy;

[0047] Action: The allocation method of tasks among different nodes;

[0048] Reward Function: Minimize computing delay & data transmission delay;

[0049] Adopt a task priority adaptive allocation strategy: High-priority tasks are preferentially allocated to low-load computing nodes, and low-priority tasks are scheduled in batches to reduce system resource occupancy;

[0050] (2) 5G / 6G high-speed data transmission optimization

[0051] Combined with the 5G / 6G network characteristics, an ultra-low latency path selection algorithm is adopted to optimize data transmission:

[0052] Intelligent routing optimization: Predict traffic changes through AI and adjust the routing in advance;

[0053] Multipath transmission: Adopt the MPTCP (Multipath TCP) technology to improve the transmission bandwidth;

[0054] Network Slicing: Allocate different network resources for different task types.

[0055] Combined with the above technical solutions and the technical problems solved, the advantages and positive effects of the technical solution to be protected by the present invention are:

[0056] First, through intelligent path optimization, dynamic scheduling, and reinforcement learning algorithms, the present invention has achieved a significant improvement in network transmission efficiency, thereby reducing the risk of data congestion. In cloud computing and edge computing environments, this technical solution can not only significantly improve the reliability and response speed of data transmission between devices but also reduce the costs of enterprises in infrastructure construction and operation and maintenance. Its commercial value lies in its ability to support larger-scale data traffic and business requirements, meet the rapid development of future intelligent networks, the Internet of Things, and big data applications, and bring significant economic benefits and competitive advantages to enterprises.

[0057] Second, currently, although there are various scheduling and traffic control strategies in the field of network transmission optimization, for scenarios involving complex cloud and edge computing collaboration, most solutions lack comprehensive consideration of device-to-device collaborative modeling and dynamic adaptability. The strategy proposed in the present invention, which combines distributed device feature modeling with reinforcement learning, innovatively realizes device-to-device collaborative optimization and real-time scheduling, thus filling the domestic and international gaps in network resource collaborative scheduling and intelligent adaptive optimization technologies and providing a new idea and practical path for technological breakthroughs in related fields.

[0058] Third, traditional network transmission systems often struggle to balance efficient data transmission and system self-adaptability. Especially when facing sudden traffic changes and diverse computing requirements, problems such as uneven resource scheduling and decreased system stability are likely to occur. The present invention utilizes deep learning and reinforcement learning technologies to achieve dynamic traffic control and intelligent resource allocation, successfully solving this long-standing problem that has plagued the industry. Through multi-level scheduling strategies, the system can quickly self-adjust in the face of various complex network environments, ensuring transmission stability and response speed to meet the needs of future high-frequency data interactions.

[0059] Fourth, in previous technical solutions, there has often been a reliance on traditional static scheduling models and a lack of trust in new intelligent algorithms, resulting in a lack of flexibility in the system when dealing with changing environments. The present invention introduces federated learning and SDN technologies to achieve an optimization strategy that combines centralized management with distributed collaboration, breaking through the limitations brought by technical biases. Its innovation lies in the integration of deep learning and real-time scheduling, successfully transforming the thinking mode of traditional network architectures, pointing the way for the development of future network transmission technologies, and setting a new benchmark in the field of intelligent network optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 is a flowchart of a method for collaborative modeling and data transmission optimization of communication equipment for a cloud computing environment provided by an embodiment of the present invention;

[0061] Figure 2 is a module diagram of a system for collaborative modeling and data transmission optimization of communication equipment for a cloud computing environment provided by an embodiment of the present invention. Detailed implementation mode

[0062] In order to make the objectives, technical solutions and advantages of the present invention more clear and understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0063] As Figure 1 shown, the embodiment of the present invention provides a communication equipment collaborative modeling and data transmission optimization method for a cloud computing environment. The method specifically includes:

[0064] S1: Collection and modeling of communication equipment characteristic data, collecting key parameters of each communication equipment, using a deep neural network (DNN) to model the equipment characteristics, and dynamically updating its state;

[0065] S2: Optimization of intelligent data transmission path, using reinforcement learning (RL) to train a data transmission path optimization model, dynamically adjusting the data transmission path according to the real-time network state, combining with an SDN controller, and dynamically adjusting the traffic allocation strategy to avoid network congestion;

[0066] S3: Data preprocessing based on edge computing, performing data preprocessing and compression on edge devices, reducing the cloud computing burden, and performing distributed computing according to task requirements to accelerate data analysis and transmission;

[0067] S4: Collaborative scheduling of computing and communication resources, using a resource-aware task scheduling algorithm, considering network bandwidth and computing resources simultaneously when allocating computing tasks, optimizing resource utilization, combining historical data and real-time status to predict the load situation, and achieving adaptive adjustment;

[0068] S5: Deployment of an adaptive network optimization strategy, combining the centralized control characteristics of SDN, deploying the optimization strategy to the network control plane to achieve efficient traffic scheduling, using the federated learning method to achieve collaborative optimization between multiple devices, and improving the intelligent collaboration ability between communication equipment.

[0069] First, the key signal data of each communication equipment, including bandwidth occupancy rate, data throughput, packet loss rate, CPU / GPU and storage utilization rate, etc., are obtained in real time through a distributed data collection module. After the collected signal data is preprocessed, it is input into a deep neural network (DNN), and the operating states of each device are mapped into high-dimensional feature vectors, and a preliminary device model is constructed. This model supports dynamic update and can reflect the device state changes in real time to ensure that subsequent scheduling decisions are based on the latest data information.

[0070] Based on real-time collection of network status data, a data transmission path optimization model is trained using reinforcement learning (RL) technology. The input data includes the current network topology, link latency, traffic load, and historical transmission conditions, etc. After model prediction, an optimal transmission path and traffic priority scheme are generated. The SDN controller receives the decision signal of the optimization model and dynamically adjusts the data packet forwarding path and traffic allocation strategy to ensure efficient and low-latency data transmission even when the network conditions change.

[0071] At the edge nodes close to the data source, the system preprocesses the collected raw signal data, including data compression, format conversion, and aggregation operations. Through preprocessing, the edge nodes can eliminate redundant information and initially split the task data to achieve local distributed computing. While reducing the amount of data transmitted, the preprocessed data also provides more refined and accurate data information for subsequent in-depth analysis and scheduling, reducing the computing load on the cloud.

[0072] At the system scheduling level, through a resource-aware task scheduling algorithm, a comprehensive analysis is carried out on the computing resource occupancy and network bandwidth status reflected in the signal data. Using historical data and real-time monitoring results, the load change trend is predicted, and a task allocation strategy is formulated to achieve coordinated scheduling of computing and communication resources. This process ensures that high-priority tasks can obtain low-load computing nodes first, while avoiding network bandwidth bottlenecks, thus effectively balancing the computing load and transmission requirements and improving the overall system efficiency.

[0073] Combined with the centralized control characteristics of SDN, the system deploys the network optimization strategy optimized by reinforcement learning and federated learning to the network control plane. At this time, the status of each communication device, task execution situation, and network load information are continuously uploaded through the feedback channel to form a closed-loop control. Each device realizes parameter sharing and collaborative update through federated learning, and collaborates with each other among multiple devices to optimize resource allocation, thereby further improving the system self-adaptability and overall operation stability.

[0074] The entire system, from data collection, edge preprocessing, to intelligent path optimization and resource collaborative scheduling, constitutes a closed-loop control process. Information is continuously exchanged among modules through real-time signal data, and the decision-making model is continuously iteratively updated using reinforcement learning. Through this dynamic feedback mechanism, the system can quickly respond to network state changes, achieve continuous adaptive optimization, and ultimately ensure that the collaborative modeling and data transmission of communication devices are always in the best state, providing a solid technical support for intelligent network applications in the cloud computing environment.

[0075] In S1, the key parameters of each communication device include network status, CPU / GPU load, storage occupancy, data throughput rate, etc. For the device status modeling, let the set of communication devices be:

[0076] E = {e1, e2, …, e n}

[0077] Each device e i has the following state vector at time t:

[0078] S i (t) = [B i (t), C i (t), M i (t), L i (t)]

[0079] where B i (t) represents the device bandwidth occupancy rate; C i (t) represents the computing resource utilization rate (CPU / GPU); M i (t) represents the storage resource occupancy rate; L i (t) represents the current data transmission task load.

[0080] The device state is modeled by a deep neural network (DNN) and defined as:

[0081]

[0082] where X i (t) is the historical state data of the device; f θ is the deep neural network model with parameters θ; it is optimized by backpropagation and gradient descent:

[0083]

[0084] where L(·) is the loss function that measures the error between the predicted state and the true state.

[0085] In S2, a deep reinforcement learning (DRL) is used to train the intelligent scheduling model:

[0086] State (S): The current network topology, task load;

[0087] Action (A): Select the best data transmission path;

[0088] Reward (R): Minimize the transmission delay and maximize the network bandwidth utilization rate;

[0089] The Dueling DQN is used to optimize the learning efficiency and accelerate the model convergence speed.

[0090] The data scheduling problem is modeled as a Markov decision process (MDP):

[0091] MDP = (S, A, P, R, γ)

[0092] Among them:

[0093] State space S:

[0094] Action space A:

[0095] State transition probability

[0096] Reward function

[0097] Discount factor γ: Controls the weight of long-term rewards.

[0098] The goal is to learn an optimal policy:

[0099]

[0100] Optimize using a Deep Q-Network (DQN):

[0101]

[0102] Update formula:

[0103]

[0104] In S3, the edge device includes an intelligent gateway, an edge server, etc.

[0105] In S5, use an SDN controller for global traffic scheduling. The control plane is responsible for calculating the optimal data path and issuing flow tables; the data plane is responsible for performing data forwarding to improve network flexibility; combined with the OpenFlow protocol, dynamically adjust the packet forwarding strategy to improve transmission efficiency; use Federated Learning (FL) to train local models on different edge nodes and aggregate them in the cloud to reduce the risk of privacy leakage, reduce data transmission costs, and improve model training efficiency.

[0106] Task scheduling is modeled as a multi-objective optimization problem, and the goal is to minimize the task completion time and resource consumption:

[0107]

[0108] Among them:

[0109] T i Is the completion time of task i;

[0110] C i Is the computing resource consumption;

[0111] B i Is the communication resource consumption;

[0112] ω1, ω2, ω3 are weight factors.

[0113] Task execution time model: The completion time of task i consists of the computing time T c and the communication time T t as follows:

[0114]

[0115] where:

[0116] D i is the data volume of the task;

[0117] f i is the processing capacity (GHz) of the computing node;

[0118] S i is the data transmission size;

[0119] R i is the network transmission rate.

[0120] Reinforcement learning + linear programming is adopted to optimize task scheduling:

[0121]

[0122] where:

[0123] R t = -(ω1T t + ω2C t + ω3B t )

[0124] Gradient descent optimization is adopted:

[0125]

[0126] where J(θ) is the loss function of the task scheduling strategy.

[0127] As Figure 2 shown, a communication equipment collaborative modeling and data transmission optimization system for a cloud computing environment provided by an embodiment of the present invention specifically includes:

[0128] A collaborative modeling layer, which adopts distributed device feature modeling, uses deep learning methods to extract features such as the operating state, bandwidth occupancy, and computing power of communication equipment, forms a dynamically updated collaborative modeling system, and optimizes the modeling process through reinforcement learning to enable it to adapt to different computing loads and data transmission requirements.

[0129] An intelligent optimization layer, which adopts an adaptive data scheduling strategy based on "reinforcement learning (RL) and deep neural network (DNN)", adjusts the data transmission path and traffic priority in real time. Combining edge computing technology, preprocessing and splitting are performed at nodes close to the data source to reduce the computing load on the cloud.

[0130] The resource collaborative scheduling layer adopts a joint optimization scheduling strategy for computing and communication resources, taking into account both computing resources and communication bandwidth during task allocation to improve resource utilization. Combining the characteristics of 5G / 6G networks, it realizes efficient and low-latency data transmission.

[0131] The collaborative modeling layer includes distributed device feature modeling and reinforcement learning optimization modeling:

[0132] (1) A distributed data acquisition module is used to obtain the operating parameters of communication equipment:

[0133] Network status parameters: bandwidth occupancy rate, data throughput, packet loss rate, etc.;

[0134] Computing resource parameters: CPU / GPU occupancy rate, storage occupancy rate;

[0135] Task load conditions: current task execution status, data traffic characteristics;

[0136] Deep neural network (DNN) is used for feature extraction, mapping the status of the device into a feature vector, and constructing a dynamically updated device model.

[0137] (2) The reinforcement learning (RL) + collaborative learning method is adopted to enable the system to adapt to dynamically changing computing loads and data transmission requirements. The collaborative model of the device is optimized through the Policy Gradient algorithm, enabling it to continuously adjust its own resource allocation strategy and improve data transmission efficiency.

[0138] The intelligent optimization layer adopts the method of reinforcement learning (RL) + deep neural network (DNN) to achieve adaptive data scheduling, optimize the data transmission path, and reduce data transmission delay.

[0139] The adaptive data scheduling strategy uses the Deep Q-Network (DQN) for data scheduling decisions:

[0140] Input: current network topology, task load, device status;

[0141] Output: optimal data transmission path and traffic priority;

[0142] Combined with the SDN controller, the traffic scheduling strategy is adjusted in real time, dynamically changing the packet forwarding path to avoid network congestion.

[0143] Data preprocessing is performed at the edge node, including:

[0144] (1) Data compression (reducing the amount of data transmitted);

[0145] (2) Task offloading (reducing the computing pressure on the cloud);

[0146] (3) Data aggregation (to improve bandwidth utilization);

[0147] Adopt a hierarchical data processing architecture:

[0148] Edge layer: Preprocess data to reduce the burden on the cloud;

[0149] Cloud computing layer: Perform complex calculations, such as deep learning inference tasks;

[0150] The resource collaborative scheduling layer simultaneously optimizes the allocation of computing resources and communication resources, improves the overall task scheduling efficiency of the system, and reduces resource competition between computing and communication. It includes:

[0151] (1) Joint optimization scheduling of computing and communication resources

[0152] Adopt reinforcement learning (Actor-Critic model) to optimize computing task allocation:

[0153] Agent: Task scheduling management module;

[0154] Environment: Current network status, computing resource occupancy;

[0155] Action: The allocation method of tasks between different nodes;

[0156] Reward Function: Minimize computing delay & data transmission delay;

[0157] Adopt a task priority adaptive allocation strategy:

[0158] High-priority tasks are preferentially allocated to low-load computing nodes;

[0159] Low-priority tasks are scheduled in batches to reduce system resource occupancy;

[0160] (2) 5G / 6G high-speed data transmission optimization

[0161] Combine the characteristics of 5G / 6G networks and adopt an ultra-low latency path selection algorithm to optimize data transmission:

[0162] Intelligent routing optimization: Predict traffic changes through AI and adjust the route in advance;

[0163] Multipath transmission: Adopt MPTCP (Multipath TCP) technology to improve transmission bandwidth;

[0164] Network Slicing: Allocate different network resources for different task types to improve the stability of data transmission.

[0165] The present invention is a communication equipment collaborative modeling and data transmission optimization system for cloud computing environments, applicable to fields such as 5G / 6G networks, the Internet of Things, smart cities, and industrial automation. Through distributed device modeling and adaptive data scheduling, the system provides technical support for high-speed and low-latency data transmission, and is an important product and technical solution for future network architecture upgrades.

[0166] The system is not only applicable to cloud big data processing and edge computing platforms, but can also be integrated into key infrastructures such as communication base stations, core networks, and data centers. By introducing a scheduling strategy that combines an SDN controller, reinforcement learning, and a deep neural network, the product can achieve efficient collaborative scheduling of cross-domain resources, providing comprehensive solutions for operators, enterprise users, and intelligent manufacturing.

[0167] The embodiment significantly reduces the cloud computing load and data transmission delay through data preprocessing (including data compression, task shunting, and data aggregation) at the edge node. Experimental or actual deployment results show that in complex network environments, this technical solution can effectively reduce latency by dozens of percentage points, meeting real-time requirements and demonstrating its superiority in high-speed data transmission.

[0168] The system adopts an adaptive data scheduling strategy based on "reinforcement learning + deep neural network" to achieve dynamic adjustment of transmission paths and traffic priorities. The application of multi-path transmission and intelligent routing optimization technologies effectively alleviates network congestion problems and improves bandwidth utilization. Test data shows that in high-concurrency scenarios, both the stability and efficiency of data transmission are significantly improved, providing strong evidence for the technical effects.

[0169] The collaborative modeling layer realizes real-time dynamic monitoring of the operating status, bandwidth occupancy rate, and computing resources of communication equipment through distributed device feature modeling and reinforcement learning optimization. The collaborative model constructed therefrom can quickly respond to load changes and intelligently adjust resource allocation strategies. In practical applications, the system shows extremely strong adaptability in the face of sudden traffic changes, verifying the significant advantages of the overall solution in terms of stability and flexibility.

[0170] The resource collaborative scheduling layer adopts a joint optimization scheduling strategy for computing and communication resources, and combines the characteristics of 5G / 6G networks to achieve global task allocation optimization. Reinforcement learning (such as the Actor-Critic model) is used for task scheduling, enabling high-priority tasks to quickly obtain low-load resources, while low-priority tasks reduce resource contention through batch scheduling. Considering various indicators, the experimental data fully demonstrate the breakthrough progress of this solution in reducing computing latency and data transmission latency and improving the overall system response speed. It should be noted that the implementation mode of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware part can be implemented using dedicated logic; the software part can be stored in a memory and executed by an appropriate instruction execution system, such as a microprocessor or dedicated design hardware. Those of ordinary skill in the art can understand that the above-mentioned devices and methods can be implemented using computer-executable instructions and / or included in processor control code, for example, such code is provided on a carrier medium such as a disk, CD, or DVD-ROM, a programmable memory such as a read-only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The devices and modules of the present invention can be implemented by hardware circuits of programmable hardware devices such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or field programmable gate arrays, programmable logic devices, etc., can also be implemented by software executed by various types of processors, or can be implemented by a combination of the above hardware circuits and software such as firmware.

[0171] As described above, the above are only specific implementation manners of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be covered by the protection scope of the present invention.

Claims

1. A communication equipment collaborative modeling and data transmission optimization method for cloud computing environment, characterized in that: The following steps are involved: S1: Communication equipment feature data collection and modeling: obtain key parameters of each communication equipment through a distributed data collection module, use a deep neural network to model the equipment features, and build a dynamically updated equipment status model; S2: Intelligent data transmission path optimization, using deep reinforcement learning to train the data transmission path optimization model, adjust the data transmission path according to the real-time network status, and dynamically adjust the traffic distribution strategy in combination with the software-defined network controller; S3: Data preprocessing based on edge computing: preprocessing the collected data on edge devices, including data compression, format conversion, and data aggregation, and performing distributed computing based on task requirements; S4: Coordinated scheduling of computing and communication resources, using a resource-aware task scheduling algorithm, taking into account both network bandwidth and computing resources when allocating computing tasks, combining historical data and real-time status to predict load and perform dynamic optimization and adjustment; S5: Adaptive network optimization strategy deployment, combined with the centralized control characteristics of software-defined networks, deploys optimization strategies to the network control plane, and uses federated learning to achieve collaborative optimization among multiple devices.

2. The method for collaborative modeling and data transmission optimization of communication equipment for cloud computing environment according to claim 1, characterized in that: In the S1 step, the key parameters of the communication equipment include network status parameters, computing resource parameters and task load parameters. The state vector of the equipment status at a specific time point is composed of bandwidth occupancy, computing resource utilization, storage resource occupancy and current data transmission task load. All parameters are represented numerically and input into a deep neural network for training to form a dynamic equipment status model.

3. The method for collaborative modeling and data transmission optimization of communication equipment for cloud computing environment according to claim 1, characterized in that: In step S2, the data transmission path optimization is based on Markov decision process modeling, which specifically includes: The state space consists of the current network topology and task load data; The action space includes different options for data transmission paths; The reward function is calculated based on the principle of minimizing transmission delay and maximizing network bandwidth utilization; The deep reinforcement learning method is used to optimize the data transmission path, and by calculating the Q value of the current state, the path that can optimize the data transmission performance is selected.

4. The method for collaborative modeling and data transmission optimization of communication equipment for cloud computing environment according to claim 1, characterized in that: In the step S3, the edge device includes an intelligent gateway and an edge server, and preprocessing the data includes: Data compression, reducing the size of data packets to reduce transmission bandwidth requirements; Format conversion, standardizing different types of data to improve compatibility; Data aggregation integrates data from the same source or similar categories to improve bandwidth utilization.

5. The method for collaborative modeling and data transmission optimization of communication equipment for cloud computing environment according to claim 1, characterized in that: In step S4, the computing task scheduling model is a multi-objective optimization problem, and the optimization objectives include: Minimize task completion time; Minimize computing resource consumption; Optimal allocation of communication resource occupancy; Among them, the task completion time is composed of computing time and communication time. The computing time is determined according to the task data volume and the processing capacity of the computing node, and the communication time is calculated according to the data transmission size and the network transmission rate.

6. The method for collaborative modeling and data transmission optimization of communication equipment for cloud computing environment according to claim 5, characterized in that: The computing task scheduling is optimized by combining reinforcement learning with linear programming, specifically including: Through intelligent task scheduling strategies, historical data and real-time status are used to predict task execution time; The gradient descent method is used to continuously optimize the task scheduling strategy and adjust the resource allocation plan to achieve the optimal utilization of computing and communication resources.

7. The method for collaborative modeling and data transmission optimization of communication equipment in a cloud computing environment according to claim 1, characterized in that: In step S5, a software-defined network controller is used to perform global traffic scheduling, wherein: The control plane selects the optimal data transmission path by calculating the current network status and dynamically adjusts the traffic scheduling strategy; The data plane performs traffic scheduling and classifies and forwards data packets according to pre-set rules; Combined with federated learning technology, local models are trained on multiple edge nodes and parameters are aggregated in the cloud to optimize the data scheduling process while reducing data transmission costs.

8. A system for collaborative modeling and data transmission optimization of communication equipment in a cloud computing environment, which implements the method for collaborative modeling and data transmission optimization of communication equipment in a cloud computing environment as described in any one of claims 1 to 7, characterized in that: include: The collaborative modeling module is used to collect the operating status parameters of each communication equipment, including network status, computing resource usage, storage resource usage and data throughput, and use deep neural networks to build a device feature model. The model supports dynamic updates to adapt to changes in computing load and data transmission requirements; Intelligent optimization module, which is used to train the data transmission path optimization model based on reinforcement learning, calculate the optimal data transmission path according to the real-time network topology and traffic conditions, and dynamically adjust the traffic distribution strategy in combination with the software-defined network controller; The resource collaborative scheduling module is used to consider the status of computing resources and communication resources when allocating computing tasks, adopt a resource-aware task scheduling algorithm to optimize resource utilization, and combine edge computing technology to perform task diversion to reduce cloud computing load and improve data transmission efficiency.

9. The communication equipment collaborative modeling and data transmission optimization system for cloud computing environment as claimed in claim 8, characterized in that: The collaborative modeling module includes: The data collection unit is used to obtain the bandwidth occupancy rate, packet loss rate, computing resource utilization rate and task load of each communication equipment in real time; A deep feature extraction unit, which is used to convert the collected data into a high-dimensional feature vector based on a deep neural network and build a device status model; The model updating unit is used to update the device state model by adopting the gradient descent optimization method so as to dynamically adapt it to different network and computing environments.

10. The communication equipment collaborative modeling and data transmission optimization system for cloud computing environment according to claim 8, characterized in that: The intelligent optimization module includes: Data transmission optimization unit, which is used to calculate the optimal data transmission path based on deep reinforcement learning and optimize the data scheduling scheme in combination with the Markov decision process; A traffic scheduling control unit, which is used to receive the scheduling decision of the optimization model and modify the packet forwarding rules through the software-defined network controller to achieve dynamic adjustment of traffic distribution; The edge computing scheduling unit is used to perform data preprocessing at the edge node, including data compression, format conversion, and task offloading, to reduce network transmission overhead and cloud computing burden.

Images