Data acquisition method based on edge computing and satellite communication technology

Abstract

The invention discloses a data acquisition method based on edge computing and satellite communication technologies, and relates to the technical field of communication data processing. The method comprises the following steps: constructing an edge cloud service platform by fusing cloud computing, edge computing and 5G communication technologies; optimizing a satellite communication protocol, a modulation and demodulation technology and a signal processing technology, and performing satellite communication data transmission; and intelligent data acquisition and processing are carried out. According to the invention, by fusing the high-performance calculation and storage capability of 5G edge cloud and the wide coverage advantage of satellite communication, more efficient, real-time and wide-coverage data acquisition is realized; the technical problems of high propagation delay, bandwidth limitation, expensive spectrum resources, path loss, multipath effect, Doppler frequency shift, orbit parameter uncertainty, edge computing resource limitation and the like encountered in data acquisition of the 5G edge cloud and satellite communication technology are effectively solved.

Description

Technical Field

[0001] The present invention relates to the field of communication data processing technology, and in particular to a data acquisition method based on edge computing and satellite communication technology. Background Art

[0002] With the deepening of digital transformation, various industries have put forward higher requirements for the real-time, accuracy and wide coverage of data collection. Traditional data collection methods are limited by network infrastructure, communication technology and geographical coverage, and it is difficult to meet current and future application needs. Therefore, exploring new data collection technologies and improving data collection efficiency and quality have become the hot spots and difficulties of current research.

[0003] As a new generation of mobile communication technology, 5G technology has the characteristics of high bandwidth, low latency and large number of connections, providing new possibilities for data collection. At present, the data collection system of 5G network can provide higher data transmission rate and lower latency, which is particularly important for application scenarios that require fast response and processing of large amounts of data. For example, in the field of industrial automation and intelligent manufacturing, 5G edge computing can realize local processing of data, reduce data transmission time and bandwidth requirements, thereby improving production efficiency and system response speed.

[0004] However, although 5G technology performs well in terrestrial networks, its effectiveness is limited in some remote or under-covered areas, which requires the assistance of satellite communication technology. Satellite communication can provide global coverage, especially in places that terrestrial networks cannot reach. Combining 5G technology and satellite communication technology can achieve wider network coverage and more stable data transmission, which is critical for scenarios such as disaster monitoring, telemedicine and global IoT applications.

[0005] In addition, the integration of 5G and satellite communications also faces some technical challenges. For example, the shortcomings of 5G edge cloud and satellite communication technology in data collection mainly include the following aspects:

[0006] 1. High propagation delay: One of the main disadvantages of satellite communication is high propagation delay. Due to the long distance between the satellite and the ground station, the signal needs to travel a long path to reach the destination, which leads to a large delay.

[0007] 2. Bandwidth limitation: Although satellite technology has improved in recent years, the bandwidth of satellite communications is still limited compared to ground networks. This is a major limitation for application scenarios that require large amounts of data transmission (such as high-definition video streaming, big data analysis, etc.).

[0008] 3. Expensive spectrum resources: The spectrum resources used by satellite communications are relatively precious, which not only increases costs but also limits the available spectrum resources. This is a problem for data-intensive applications that require a large amount of spectrum resources.

[0009] 4. Path loss and multipath effect: Satellite signals will be severely affected by path loss during transmission and are easily affected by multipath effects, which will affect the quality and reliability of the signal.

[0010] 5. Doppler frequency shift: Due to the rotation of the earth and the high-speed movement of satellites around the earth, satellite signals will produce Doppler frequency shift, which requires complex correction processing at the receiving end, increasing the complexity and cost of the system.

[0011] 6. Uncertain orbital parameters: The position and speed of satellites are affected by many factors, such as weather, earth's gravity, etc. These factors may lead to uncertainty in the satellite position, thereby affecting the stability and reliability of communications.

[0012] 7. Limited edge computing resources: Although applying the concept of edge computing to satellite communications can bring performance improvements, traditional mobile communication satellites only serve as information transfer stations and do not have strong data processing capabilities. This means that in some cases, satellite networks may not be able to effectively handle complex computing tasks.

[0013] 8. Resource constraints and operational isolation: Low-orbit satellite communication systems are limited by onboard capabilities and on-orbit computing capabilities. In adapting to the future integrated space-ground network architecture, there are problems such as resource constraints and operational isolation.

[0014] Therefore, it is necessary to design a data acquisition method that can effectively solve the technical problems encountered by 5G edge cloud and satellite communication technologies in data acquisition, such as high propagation delay, bandwidth limitation, expensive spectrum resources, path loss and multipath effect, Doppler frequency shift, orbit parameter uncertainty, and limited edge computing resources. Summary of the invention

[0015] In view of the problems existing in the prior art, the present invention provides a data collection method based on edge computing and satellite communication technology.

[0016] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0017] A data collection method based on edge computing and satellite communication technology includes the following steps: step 1, building an edge cloud service platform by integrating cloud computing, edge computing and 5G communication technology; the construction process is: step 11, designing and deploying network architecture; step 12, deploying computing nodes at the edge of the network; step 13, collecting and processing raw data; step 14, data storage and access management; step 15, deploying real-time computing tasks for data processing and analysis; step 16, monitoring and optimizing data; step 2, optimizing satellite communication protocols, modulation and demodulation technologies and signal processing technologies, and performing satellite Communication data transmission; the process is: step 21, optimize satellite communication protocol; step 22, optimize modulation technology and demodulation technology; step 23, optimize signal processing technology; step 24, perform system integration and testing; step 3, perform intelligent data collection and processing; the process is: step 31, build an intelligent data collection model; step 32, optimize the data processing algorithm; step 33, build a data mining model based on the data mining algorithm, and train and optimize it; step 34, integrate the intelligent data collection model, data processing algorithm and data mining capabilities; step 35, adaptive adjustment and optimization.

[0018] Based on the above technical solution, further, in step 11, a network architecture including 5G base stations, edge computing nodes and cloud computing centers is designed, and a 5G network is deployed; wherein, the shortest path algorithm is used for design and deployment.

[0019] Based on the above technical solution, further, in step 12, a multi-objective optimization algorithm is used to deploy computing nodes, wherein the multi-objective optimization algorithm is any one of a scalarization method, a goal programming method, an ε-constraint method, and a Pareto optimal solution.

[0020] Based on the above technical solution, further, in step 21, the optimization process is: step 211, define quantitative indicators of error rate, protocol complexity and compatibility; step 212, set constraints: including complexity constraints and compatibility constraints; step 213, formulate optimization strategy: optimize the error rate, and optimize the balance between complexity and compatibility; step 214, implement and evaluate based on the optimization strategy.

[0021] Based on the above technical solution, further, in step 22, the optimization process is: step 221, define the objective function: including minimizing the signal distortion function and maximizing the transmission rate function; step 222, set the constraints: including algorithm complexity and power consumption; step 223, select the optimization algorithm, and formulate a solution strategy; step 224, implement and evaluate according to the solution strategy.

[0022] Based on the above technical solution, further, in step 23, the optimization process is: step 231: define the target processing function; step 232, set constraints: including processing delay and algorithm stability; step 233, select the optimization algorithm, and formulate a solution strategy; step 234, evaluate and verify according to the solution strategy.

[0023] Based on the above technical solution, further, in step 24, the system test process is: first, data transmission rate test: the test goal is to ensure that the data transmission rate R' meets the preset minimum requirement R' min ; R′=data volume / time, verify R′≥R′ min ; Second, bit error rate test: The test goal is to ensure that the bit error rate BER' is lower than the preset maximum tolerance value BER' max ; BER′=number of error bits / total number of transmitted bits, verify BER′≤BER′ max ; Third, system stability test: Fourth, coverage test.

[0024] Based on the above technical solution, further, in step 31, the construction process includes the following steps: Step 311, data collection demand analysis: set the data set to be collected, A1 is a data source set composed of multiple data sources A1′; for each data source A1′∈A1, define its data contribution c(A1′) and collection difficulty d(A1′); the goal is to minimize the collection difficulty while maximizing the data contribution: min∑ A1′∈A1 d(A1′)-λ∑ A1′∈A1 c(A1′), λ is the weight coefficient; Step 312, data acquisition model design: let N be the data acquisition model, take the data source A1′ as the input of the data acquisition model, and take the collected data B as the model output; the goal of the data acquisition model is to minimize the acquisition error: min N ∑ A1′∈A1 ‖M(A1′)-y true (A1′)‖ 2 , where y true (A1′) is the real data.

[0025] Based on the above technical solution, further, in step 32, the optimization process includes: step 321, algorithm selection: for algorithm a, define its processing speed v(a) and processing accuracy p(a); the goal is to minimize the processing time and maximize the processing accuracy: min a∈A [1 / v(a)]-μ×p(a), where μ is the weight coefficient and A is the data processing algorithm set; Step 322, algorithm implementation and optimization: Let f(b) be the data processing function, b be the input data, and d be the output data; the optimization goal is to minimize the error of the output data: min f ∑b∈D1 ‖f(b)-d expected (b)‖ 2 , where d expected (b) is the expected output, and D1 is the given data set.

[0026] Based on the above technical solution, further, in step 33, the process includes: step 331, constructing a data mining model: let G be the data mining model, X be the input feature set, and Y be the target variable; the goal of the model is to minimize the prediction error: min G ∑ (x,y)∈(X,Y) ‖G(x)-y‖ 2 ; Step 332, model training and optimization: Use the optimization algorithm to train the model G, set L(G) as the loss function, and the optimization goal is min G L(G).

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] The present invention discloses an intelligent data collection method based on 5G edge cloud and satellite communication technology, which realizes more efficient, real-time and wide-coverage data collection by integrating the high-performance computing and storage capabilities of 5G edge cloud and the wide coverage advantage of satellite communication. The successful implementation of this method is expected to solve many challenges faced by traditional data collection methods and promote the intelligent transformation and upgrading of related industries. It can effectively solve the technical problems encountered by 5G edge cloud and satellite communication technologies in data collection, such as high propagation delay, bandwidth limitation, expensive spectrum resources, path loss and multipath effect, Doppler frequency shift, orbit parameter uncertainty and limited edge computing resources. Specifically, it is reflected in:

[0029] 1. High propagation delay: One of the typical challenges of satellite communication systems is large path loss and delay. To solve this problem, dynamic spectrum sharing technology and downlink time-frequency synchronization research and prototype verification of low-orbit satellite mobile communication systems can be used to reduce dynamic Doppler frequency shift and dynamic delay caused by high-speed relative motion.

[0030] 2. Bandwidth limitation: Traditional satellite networks have disadvantages such as expensive spectrum bandwidth resources. Through the application of edge computing technology, especially the introduction of edge computing architecture in mobile satellite communication networks, business processing can be carried out in edge network nodes closer to the user side, thereby effectively shortening service response time and saving precious backhaul bandwidth resources in satellite networks.

[0031] 3. Expensive spectrum resources: The scarcity and cost of spectrum resources are one of the main challenges facing 5G systems. Dynamic spectrum sharing technology in the satellite-ground integrated network can alleviate the spectrum scarcity problem and improve spectrum utilization.

[0032] 4. Path loss and multipath effects: These problems can be solved by adopting new physical layer technologies, such as massive MIMO technology and beamforming algorithms, as well as using software-defined networking and virtual switching technologies to optimize the signal transmission and reception process and reduce the impact of path loss and multipath effects.

[0033] 5. Doppler frequency shift: In the low-orbit satellite mobile communication channel link, the high-speed relative motion between the satellite and the ground terminal will cause a large dynamic Doppler frequency shift. This problem can be solved by designing a broadband low-orbit satellite mobile communication ground experimental system integrated with 5G and adopting the corresponding time-frequency synchronization algorithm.

[0034] 6. Orbital parameter uncertainty: This can be mitigated by improving spectrum sharing technologies for satellite-5G integrated networks and algorithms that use energy detection for spectrum sensing. These technologies can help better manage the uncertainty of satellite positions and orbital parameters.

[0035] 7. Limited edge computing resources: In order to solve the problem of limited edge computing resources, lightweight virtualization, software-defined networking, virtual switching and other technologies can be used, combined with mobile edge caching and computing strategies to make full use of the storage computing resources at the edge of the network, user mobility and dynamic renewable energy supply to provide users with caching and computing services. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0037] The present invention is further described and illustrated below in conjunction with the accompanying drawings and specific embodiments. The technical features of each embodiment of the present invention can be combined accordingly without conflicting with each other.

[0038] In order to make the above-mentioned purpose, features and advantages of the present invention more obvious and easy to understand, the specific implementation mode of the present invention is described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below. The technical features in each embodiment of the present invention can be combined accordingly without conflicting with each other.

[0039] Example

[0040] Combination Figure 1 As shown, this embodiment provides a data collection method based on edge computing and satellite communication technology, including the following steps:

[0041] Step 1: Build an edge cloud service platform by integrating cloud computing, edge computing, and 5G communication technologies. Specifically, adopt 5G edge cloud fusion technology. The core is to utilize the high bandwidth and low latency advantages of the 5G network to migrate computing and data storage capabilities to the edge of the network, and achieve efficient construction and operation of the edge cloud. By deeply integrating cloud computing, edge computing, and 5G communication technologies, an edge cloud service platform that is both efficient and low-latency has been built. This platform provides powerful computing and storage support for data collection tasks, ensures the real-time and accuracy of data collection, and effectively reduces the delay and cost of data transmission. The specific construction process is as follows:

[0042] Step 11: Design and deploy the network architecture to ensure high bandwidth and low latency connections at the edge of the network. Specifically, design a network architecture that includes 5G base stations, edge computing nodes, and cloud computing centers, and deploy a 5G network. The shortest path algorithm is used for design and deployment. In this embodiment, the shortest path algorithm selects the Dijkstra algorithm. It should be noted that the vertices and origins of the algorithm can be called nodes. The algorithm steps are as follows:

[0043] Step 111, perform initialization processing: set the source point s. Create a distance array dist[], where dist[i] represents the storage of the current shortest distance estimate from the source point s to the vertex i. Initially, dist[s] = 0. For all other vertices i, dist[i] = ∞ (infinity, indicating unknown). Create a set visited to store the vertices for which the shortest path has been found. Initially, the set is empty.

[0044] Step 112, loop until the shortest path to all vertices is found: among all unprocessed vertices (i.e., vertices not in the visited set), find the vertex u that is closest to the source point s (i.e., the value of dist[u] is the smallest). Add vertex u to the visited set, indicating that the shortest path from the source point to vertex u has been found. Update the distances of all vertices adjacent to vertex u. For each vertex v adjacent to vertex u, if the distance from vertex u to vertex v (i.e., dist[u]+weight(u,v)) is smaller than the currently known dist[v], update dist[v] to dist[u]+weight(u,v); where weight(u,v) is used to represent the weight of the edge between vertex u and vertex v.

[0045] Step 113, end of algorithm: When all vertices are added to the visited set, the algorithm ends. At this point, the dist[] array stores the shortest path length from the source point to all other vertices.

[0046] Step 12: Deploy computing nodes at the edge of the network; configure the corresponding hardware and software resources, integrate the cloud computing platform, and realize cloud-edge collaboration and resource scheduling; this process involves multi-objective optimization problems. For this problem, the processing ideas are as follows: Clarify the problem: determine the domain of all objective functions, constraints, and variables. Select the method: Select the appropriate multi-objective optimization method according to the specific characteristics and requirements of the problem. Build a model: Convert the problem into a mathematical model, including the objective function and constraints. Solve: Use appropriate algorithms or software tools to solve the model. Result analysis: Analyze the quality of the solution, including the Pareto frontier, the distribution of the solution, and the trade-off of objectives.

[0047] Specifically, a multi-objective optimization algorithm is used to deploy computing nodes, wherein the multi-objective optimization algorithm is any one of a scalarization method, a weighted sum method, a target programming method, an ε-constraint method, and a Pareto optimal solution. In this embodiment, the process of deploying computing nodes is as follows: Let U be a decision variable, f1(U), f2(U),…, f n (U) is the objective function to be optimized. The scalarization method is used for deployment calculation. The process is: convert the multi-objective optimization problem into a single-objective optimization problem. This is achieved by assigning a weight to each objective function and summing them up. That is, construct a new objective function:

[0048] Among them, w k Each objective function f k (U) weight, and w k ≥ 0. The choice of weights usually depends on the specific context of the problem and the preferences of the decision maker. n is the number of objective functions, k = 1, 2, 3...n.

[0049] Then, solve the following single-objective optimization problem:

[0050] minimize F(U);

[0051] subject to g i (U)≤0, i=1,2,...m; where m represents the upper limit of the value of the formula and has no special meaning;

[0052] h j (U)=0,j=1,2,…,q;where q represents the upper limit of the value of the formula and has no special meaning. i (U) and h j (U) is a constraint condition.

[0053] Step 13: Raw data collection and processing flow: Based on the development of data collection module, collect data from various sensors and devices. Implement data processing logic, including data cleaning, compression and encryption, etc. For example, data cleaning methods can be used to process missing values ​​and data conversion. Specifically, data smoothing: Let h1, h2, …, h z is the original data, and the smoothed data is expressed as: t =[h (t-1) v+h t +h (t+1 )] / 3,t=2,3,…,z-1; where the current element h t , the previous element h (t-1) and the next element h (t+1) , get the new value y by setting the weighted average t Specifically, the new value y t is the previous element h (t-1) Multiply by a certain weight v, the current element h t And the next element h (t+1) One third of the sum of these three.

[0054] Among them, the new value can be specially processed, let y1 = h1, y z =h z .

[0055] Step 14, data storage and access management: configure local storage and remote storage solutions to ensure data security and availability, implement data access control mechanisms, and protect sensitive data from unauthorized access. In this embodiment, access is performed using an indexing method, where the indexing process is as follows: set a balanced multi-way search tree, the relationship between its height h′ and the node degree m′ and the number of elements n′ in the tree is expressed as: h′=log m′ ((n′+1) / 2)+1; where log m′ Represents the logarithm with base m′.

[0056] Step 15: Deploy real-time computing tasks for data processing and analysis; deploy real-time computing tasks on edge computing nodes, use local computing power to process and analyze data; feed back the processing results to users in real time or upload them to the cloud computing center for further processing. In this embodiment, a sliding window model is used for data processing and analysis. The specific process is as follows: Let W be the window size, that is, the time range that defines the window, that is, the time span of the data contained in the window; S is a data stream, which contains multiple data items s, and each data item has an associated timestamp s.timestamp; t1 is the current timestamp, which indicates the time point of the current data processing; the sliding window S W The data in (t) is expressed as: S W(t)={s∈S∣t1_W≤s.timestamp≤t1}.

[0057] Step 16: Monitor and optimize data; implement a monitoring mechanism to track the performance and resource usage of edge computing nodes, optimize based on the monitoring data, and ensure the efficient operation of the edge cloud service platform. In this embodiment, an AR model (time series prediction model) is used to analyze and predict time series data. It uses past data points of the same time series to predict future values. In the AR model, the value at the current time point is assumed to be a linear combination of the values ​​at the past few time points plus a random error term (white noise), thereby achieving optimization. The process is: let R T is the value of the time series at time T, p is the model order, and the AR model is expressed as: R T =φ1R T_1 +φ2R T_2 +…+φ p R T_p +∈ t ; where φ1, φ2, …, φ p is the model parameter, ∈ t is white noise, representing random errors or unpredictable components at time T. It is usually assumed that these error terms are independent and identically distributed, with a mean of 0 and a variance of a constant σ 2 ; p is the order of the model, that is, the number of past observations used in the model.

[0058] For example, the complete process of data collection, processing, real-time calculation, and result storage and upload based on the 5G edge cloud fusion environment is shown in pseudo code as follows:

[0059]

[0060]

[0061]

[0062] Please note that the above pseudo code is a simplified example to illustrate the core implementation steps of 5G edge cloud fusion technology. In actual applications, detailed system design and programming implementation are required according to specific needs and scenarios.

[0063] Explain the specific meaning of the above pseudo code:

[0064] Explanation 1. initialize5GEdgeCloudEnvironment(config) is explained as: This step is responsible for initializing the 5G edge cloud fusion environment according to the provided configuration (config). This may include configuring network connections, initializing edge computing resources, setting security policies, etc.

[0065] Explanation 2, setupEdgeComputingNodes(nodeConfigs), which is explained as: This step sets up and configures edge computing nodes according to the provided node configuration (nodeConfigs). This may include configuring hardware resources, installing necessary software, setting up communication between nodes, etc.

[0066] Explanation 3, rawData = collectDataFromSensors(), which is explained as: This function collects raw data from connected sensors. These sensors may be deployed in various environments such as factories, farmlands, and city streets to monitor various types of data such as temperature, humidity, pressure, and images.

[0067] Explanation 4: processedData=preprocessData(rawData), which is explained as: in the preprocessing stage, the raw data undergoes steps such as cleaning (removing noise, outliers, etc.), compression (reducing storage and transmission costs) and encryption (ensuring data security).

[0068] Explanation 5, results = executeRealTimeTasks (processedData), which is explained as: for the preprocessed data, the system executes a series of real-time computing tasks. These tasks may include machine learning model reasoning, data analysis, real-time alarms, etc. Each task is executed independently and the results are collected.

[0069] Explanation 6, storeResults(results, storageConfig), which is explained as: According to the storage configuration (storageConfig), the results can be stored in the local or remote storage system. This depends on factors such as the sensitivity of the data, access frequency, and backup requirements.

[0070] Explanation 7, if needToUploadToCloud:

[0071] uploadResultsToCloud(results,cloudConfig), which means: if the results need to be uploaded to the cloud computing center for further analysis or backup, the system will use the 5G network for efficient transmission. This ensures the timeliness and reliability of the data.

[0072] Explanation 8, waitForNextCycle(), which is explained as: after completing a round of data collection, processing, calculation and storage, the system waits for the arrival of the next data collection cycle. This waiting time may be adjusted according to actual needs to ensure that the system can run continuously and stably.

[0073] Step 2: Optimize satellite communication protocols, modulation and demodulation technologies, and signal processing technologies to perform satellite communication data transmission; fully utilize the wide coverage and high reliability of satellite communications to achieve data collection and transmission in remote and remote areas, greatly improving the efficiency and stability of data transmission. It not only ensures global coverage and seamless connection of data collection, but also successfully breaks geographical restrictions, providing solid support for data collection work around the world. The specific process is:

[0074] Step 21, optimize satellite communication protocol: analyze the performance bottleneck of the existing satellite communication protocol; design and implement a more efficient data encapsulation and parsing mechanism; integrate error detection and correction mechanism to improve the reliability of data transmission. The optimization process of this step is as follows: let P1 represent the existing satellite communication protocol, and P1′ represent the optimized satellite communication protocol;

[0075] Step 211, define quantitative indicators of error rate, protocol complexity and compatibility; in this embodiment, error rate (ErrorRate(P1′)): This can usually be estimated through experimental testing or based on theoretical models; it may depend on multiple factors, such as signal strength, noise level, coding efficiency, etc. Protocol complexity can be measured by the number of lines of code, algorithm complexity (such as time complexity and space complexity), or hardware resources required to implement the protocol. Compatibility usually refers to the interoperability of the optimized new protocol P1′ with existing systems (such as old versions of protocols, hardware devices, etc.), which may involve compatibility at the protocol level (such as message format, control commands, etc.) and compatibility at the physical level (such as frequency, modulation method, etc.).

[0076] Step 212: Set constraints: including complexity constraints and compatibility constraints. Specifically, complexity constraints: ensure that the complexity of the new protocol P1′ does not exceed a preset upper limit to avoid excessive implementation costs or operating costs. Compatibility constraints: ensure that the new protocol P1′ maintains sufficient compatibility with the existing system to enable smooth transition and upgrade.

[0077] Step 213: Formulate an optimization strategy;

[0078] Specifically, (1) Optimize the error rate: ① Improve coding technology: Use more efficient channel coding and error correction codes to reduce transmission errors. ② Optimize modulation methods: Select appropriate modulation methods according to channel conditions to improve the reliability and efficiency of signal transmission. ③ Enhance signal processing capabilities: Reduce the impact of noise and interference by improving signal processing technology (such as filtering, equalization, etc.).

[0079] (2) Balance complexity and compatibility: ① Modular design: Design the protocol as configurable modules so that complexity and compatibility can be adjusted according to specific needs. ② Gradual upgrade: Adopt a phased upgrade strategy, first implement some optimized functions, and then gradually expand and improve them. ③ Compatibility testing: Conduct extensive compatibility testing during the development process to ensure the interoperability of the new protocol with the existing system.

[0080] Step 214: Implement and evaluate based on the optimization strategy. Specifically, ① prototype development: Based on the above optimization strategy, develop a prototype system of the new protocol P1′. ② Experimental testing: Test the performance of the new protocol in a simulated or real environment, including error rate, complexity and compatibility. ③ Performance evaluation: Evaluate whether the new protocol meets the optimization objectives and constraints based on the test results. If not, return to the optimization strategy stage for adjustment.

[0081] The above whole process can systematically optimize the satellite communication protocols P1 to P1′ to minimize the error rate of data transmission and simultaneously meet the constraints of protocol complexity and compatibility.

[0082] Step 22, optimize modulation and demodulation technology: study and select appropriate modulation and demodulation schemes to adapt to the satellite communication environment; implement modulation and demodulation algorithms to ensure stability under signal attenuation and multipath effects; optimize algorithm parameters to improve data transmission rate and anti-interference ability. The optimization process of this step is: let M and D represent modulation technology and demodulation technology respectively, x1 is the input signal, y is the output signal, and the optimization goal is to minimize signal distortion and maximize transmission rate, expressed as: \min_{M,D}\text{Distortion}(x1,D(M(x1)))\]\[\max_{M,D}\text{TransmissionRate}(M,D); constraints include algorithm complexity and power consumption.

[0083] Specifically, step 221, define the objective function: minimize the signal distortion function and maximize the transmission rate function.

[0084] Specifically, distortion is usually defined as the difference between the original signal x and the signal D(M(x)) that is recovered after modulation and demodulation. This can be quantified by mean square error (MSE), signal-to-noise ratio (SNR), or other distortion metrics. It can be expressed as:textDistortion(x,D(M(x)))=textMSE(x,D(M(x)))quadtext orquadtextDistortion(x,D(M(x)))=frac1textSNR(x,D(M(x))).

[0085] The transmission rate is the amount of data transmitted per unit time, which depends on the efficiency of the modulation technology and the channel conditions. textTransmissionRate(M,D)=R'; where R' is the transmission rate, usually in bits per second (bps).

[0086] Step 222: Set constraints: including algorithm complexity and power consumption.

[0087] Specifically, algorithm complexity: ensure that the complexity of the modulation and demodulation algorithms is within an acceptable range to avoid excessively high computational costs, which can be achieved by limiting the time complexity and space complexity of the algorithm.

[0088] textComplexity(M)leqCM,quadtextComplexity(D)leqC D , where C M and C D They are the upper limits of the complexity of the modulation and demodulation algorithms respectively.

[0089] Power consumption is to ensure that the power consumption of the modulation and demodulation process does not exceed the power supply capability of the device.

[0090] textPower(M,D)leqP textmax Where P max is the maximum power consumption allowed by the system.

[0091] Step 223: Select an optimization algorithm and formulate a solution strategy; wherein the optimization algorithm includes but is not limited to hierarchical optimization, Pareto frontier method and weighted sum method. Exemplarily, the hierarchical optimization idea is to first optimize one goal (e.g., minimize distortion), and then optimize another goal (e.g., maximize transmission rate) on the premise of satisfying the goal.

[0092] Step 224: Implement and evaluate according to the solution strategy. Specifically, algorithm implementation: implement the modulation and demodulation algorithms according to the selected solution strategy. Simulation test: test the performance of the algorithm in a simulated environment, including distortion, transmission rate, complexity and power consumption. Performance evaluation: evaluate whether the algorithm meets the optimization objectives and constraints based on the test results; if not, return to adjust the algorithm or solution strategy.

[0093] Step 23, optimize signal processing technology: develop advanced signal filtering and enhancement algorithms to reduce noise and interference; implement adaptive signal processing technology to cope with different channel conditions; integrate signal synchronization mechanism to ensure accurate data reception. The optimization process of this step is: let S1 represent the signal processing algorithm, n1 is the noise signal, and x is the original signal; the optimization goal is to minimize the impact of noise on the signal, expressed as: \min_{S1}\text{NoiseImpact}(S1 (x+n1) ,x)\]; the constraints include processing delay and algorithm stability.

[0094] Specifically, step 231: define the target processing function: minS1textNoiseImpact(S1(x+n1),x); wherein S1(x+n1) is the output of the signal processing algorithm S1 after being applied to the noise-contaminated signal x+n1;

[0095] Step 232, set constraints: including processing delay and algorithm stability; where (1) processing delay is to ensure that the delay of the signal processing algorithm is within an acceptable range; it usually means that the algorithm should complete the processing within a certain time in order to be applied in real time or meet specific time requirements; textDelay(S1)≤Δ, where Δ is the maximum allowed delay. (2) Algorithm stability is to ensure that the algorithm can work stably when processing various inputs without excessive fluctuations or errors. Stability can be measured in many ways, such as algorithm robustness, convergence, etc.

[0096] textStability(S1)≥StabilityThreshold, where StabilityThreshold is the minimum threshold that the algorithm needs to reach for stability.

[0097] Step 233, select an optimization algorithm and develop a solution strategy; specifically, since this is an optimization problem and involves the complexity and nonlinearity of the signal processing algorithm, one of the following strategies or a combination of multiple strategies can be used to solve it: 1. Gradient descent method (if the objective function is differentiable): by calculating the gradient of the objective function with respect to the algorithm parameters (if the algorithm is parameterized) and updating the parameters in the opposite direction of the gradient to gradually reduce the impact of noise. 2. Heuristic search: such as genetic algorithms, simulated annealing, etc., these algorithms do not rely on the derivative of the objective function and are suitable for complex and nonlinear optimization problems. 3. Expert system or rule-based method: If there is prior knowledge about a specific type of signal and noise, a rule-based algorithm can be designed to select or adjust the signal processing parameters. 4. Machine learning: Train a machine learning model to predict the optimal signal processing parameters or directly output the processed signal to minimize the impact of noise.

[0098] Step 234: Evaluate and verify the solution strategy. Specifically, generate various noise signals in a simulated environment, and apply the signal processing algorithm to evaluate its performance, thereby implementing simulation testing. Use actual collected data to verify the effectiveness and robustness of the algorithm, thereby implementing actual data testing. Evaluate the performance of the algorithm based on indicators such as noise impact, processing delay, and algorithm stability.

[0099] Step 24, perform system integration and testing: integrate the optimized communication protocol, modulation and demodulation technology, and signal processing technology into the satellite communication system; perform system-level testing to verify the efficiency, stability, and global coverage of data transmission; and make necessary adjustments and optimizations based on the test results. Specifically, let H represent the entire satellite communication data transmission system, which is integrated with the optimized communication protocol P1′, modulation technology M, demodulation technology D, and signal processing technology S′. The goal of system integration and testing is to verify the performance and stability of S under various conditions.

[0100] Among them, system testing includes:

[0101] First, data transmission rate test: The data transmission rate R' can be defined as the amount of data successfully transmitted per unit time. The test goal is to ensure that the data transmission rate R' meets the preset minimum requirement R' min ; R = data volume / time, verify R′≥R′ min ; When the verification conditions are met, it proves that the performance is good.

[0102] Second, bit error rate test: The bit error rate BER' is a key indicator to measure the accuracy of data transmission, which is defined as the ratio of the number of error bits to the total number of transmitted bits. The test goal is to ensure that the bit error rate BER' is lower than the preset maximum tolerance value BER' max; BER′=number of error bits / total number of transmitted bits, verify BER′≤BER′ max ; When the verification conditions are met, it proves that the performance is good.

[0103] Third, system stability test: System stability can be measured by performance fluctuations during long-term operation. The test goal is to ensure that the fluctuation range of key performance indicators (such as R′ and BER′) is within an acceptable range after the system has been running for a long time. Let X be a certain performance indicator (such as R′ or BER′), X avg is its long-term average value, and ΔX is its fluctuation range. Verification | ΔX | ≤ ΔX max , where ΔX max It is the maximum acceptable fluctuation range; when the verification conditions are met, it proves that the performance is good.

[0104] Fourth, coverage test: Coverage test aims to verify the coverage capability of the system in the world or a specific area. Mathematical representation: Let A represent the test area, A covered Indicates the area actually covered by the system. Verification A covered ≥A min , where A min It is the preset minimum coverage requirement; when the verification conditions are met, it proves that the performance is good.

[0105] For example, the basic process of optimizing and implementing satellite communication data transmission technology is illustrated in pseudo code. The following is a detailed explanation of each function and main program, as well as possible implementation steps or considerations, as follows:

[0106]

[0107]

[0108] This pseudo code provides a high-level overview of how to implement and optimize the various key components of satellite communication data transmission technology. In actual applications, each function will contain more detailed implementation steps and algorithms.

[0109] Explanation 1. optimizeSatelliteCommunicationProtocol(), its function is to analyze the existing satellite communication protocol, design and optimize the data transmission encapsulation method. Its steps:

[0110] analyzeExistingProtocol(): Evaluates the efficiency, reliability, latency and other performance indicators of the current protocol.

[0111] designEfficientDataPackaging(): Based on the analysis results, design a more efficient data packaging format, which may include data compression, subpackaging strategies, etc.

[0112] integrateErrorDetectionAndCorrection(): Integrates error detection and correction mechanisms, such as CRC check, FEC (forward error correction) coding, etc., to improve the reliability of data transmission.

[0113] Explanation 2: optimizeModulationDemodulation(), its function is to select a suitable modulation and demodulation scheme and optimize its algorithm and parameters. Its steps are:

[0114] selectAppropriateModulationScheme(): Selects an appropriate modulation scheme, such as QPSK, QAM, etc., based on channel characteristics (such as bandwidth and noise level).

[0115] implementModulationDemodulationAlgorithm(): implements the selected modulation and demodulation algorithm.

[0116] optimizeAlgorithmParameters(): Adjust algorithm parameters such as modulation index, symbol rate, etc. through simulation or experiment to optimize performance.

[0117] Explanation 3, optimizeSignalProcessing(), its function is to develop and optimize signal processing and synchronization technology. Its steps are:

[0118] developSignalFilteringAndEnhancement():Design signal filters, such as bandpass filters, noise suppressors, etc., to enhance signal quality.

[0119] implementAdaptiveSignalProcessing(): Implement adaptive signal processing algorithms, such as adaptive equalizer, to cope with channel changes.

[0120] integrateSignalSynchronization(): Integrates the signal synchronization mechanism to ensure that the receiving end can accurately synchronize the signal of the sending end.

[0121] Explanation 4: integrateAndTestSystem(), its function is to integrate various modules and protocols together for system-level testing and optimization. The steps are:

[0122] integrateProtocolsAndTechniques(): Integrates the optimized protocols, modulation and demodulation techniques, and signal processing techniques into the entire system.

[0123] performSystemLevelTesting(): Performs system-level testing in a simulated or actual environment to evaluate overall performance.

[0124] adjustAndOptimizeBasedOnResults(): Adjust and optimize system parameters based on test results until performance requirements are met.

[0125] The main program main is used to execute all the above optimization and integration steps in sequence. The steps are:

[0126] Call optimizeSatelliteCommunicationProtocol() to optimize the communication protocol.

[0127] Call optimizeModulationDemodulation() to optimize the modulation and demodulation techniques.

[0128] Call optimizeSignalProcessing() to optimize the signal processing technique.

[0129] Call integrateAndTestSystem() to integrate the entire system and perform testing and optimization.

[0130] Step 3: Use intelligent data collection and processing technology for intelligent data collection and processing; among them, intelligent data collection and processing technology is the in-depth application of artificial intelligence and machine learning algorithms in the field of data collection and processing. By building an intelligent data collection model, optimizing data processing algorithms, and improving data mining capabilities, automatic data collection, real-time processing, and value mining can be achieved. This technology not only significantly improves the intelligence level of data collection, but also provides strong data support for decision support and business optimization. At the same time, this technology also has the ability to adaptively adjust and optimize according to actual application needs, ensuring the efficiency and accuracy of data collection. In addition, these core technologies are integrated with each other to form an efficient, intelligent, and flexible data collection and processing ecosystem. In this ecosystem, 5G edge cloud fusion technology provides strong infrastructure support for data processing, ensuring the efficiency and real-time nature of data processing; satellite communication data transmission technology breaks geographical restrictions, realizes seamless transmission of data worldwide, and ensures the comprehensiveness and accuracy of data; and intelligent data collection and processing technology, through advanced algorithms and models, realizes deep mining and value discovery of data, providing a scientific basis for business decision-making and strategic planning. The specific process is:

[0131] Step 31: Build an intelligent data collection model: Analyze data collection requirements and determine the types and sources of data to be collected; design and implement an intelligent data collection model that can adaptively collect data from various data sources; integrate data cleaning and preprocessing functions to ensure the quality of collected data.

[0132] Specifically, the construction process is:

[0133] Step 311, data collection requirements analysis: Set the data set to be collected, A1 is a data source set consisting of multiple data sources A1′; for each data source A1′∈A1, define its data contribution c(A1′) and collection difficulty d(A1′); the goal is to minimize the collection difficulty while maximizing the data contribution: min∑ A1′∈A1 d(A1′)-λ∑ A1′∈A1 c(A1′), λ is the weight coefficient;

[0134] Step 312, data collection model design: Let N be the data collection model, take data source A1′ as the input of the data collection model, and take the collected data B as the model output; the goal of the data collection model is to minimize the collection error: min N ∑ A1′∈A1 ‖M(A1′)-y true (A1′)‖ 2 , where y true (A1′) is the real data.

[0135] Step 32: Optimize data processing algorithms: Research and select appropriate data processing algorithms to adapt to different data processing scenarios; implement data processing algorithms and optimize them to improve processing speed and accuracy; integrate parallel processing and distributed computing technologies to meet large-scale data processing needs.

[0136] Specifically, the optimization process includes:

[0137] Step 321, select an algorithm: Let A be a set of data processing algorithms, and a∈A be a specific algorithm. For each algorithm a, define its processing speed v(a) and processing accuracy p(a). The goal is to minimize the processing time while maximizing the processing accuracy: min a∈A [1 / v(a)]-μ×p(a), where μ is the weight coefficient.

[0138] Step 322, algorithm implementation and optimization Let f(b) be the data processing function, b be the input data, and d be the output data; the optimization goal is to minimize the error of the output data: min f ∑ b∈D1 ‖f (b) -d expected (b)‖ 2 , where d expected (b) is the expected output, and D1 is the given data set. Given data set D1 = {(b1, d1), (b2, d2), …, (be, de)}, where be is the input data, de = d expected (be) is the corresponding expected output (also called the true value or label).

[0139] Step 33. Build, train and optimize data mining models based on data mining algorithms: Develop advanced data mining algorithms to discover hidden patterns and associations in data; implement data mining models, and train and optimize them to improve mining results; integrate visualization tools so that users can intuitively understand and interpret mining results.

[0140] The specific process includes:

[0141] Step 331, construct a data mining model: Let G be the data mining model, X be the input feature set, and Y be the target variable. The goal of the model is to minimize the prediction error: min G ∑ (x,y)∈(X,Y) ‖G(x)-y‖ 2 , where (x,y) is any data coordinate in the set (X,Y).

[0142] Step 332, model training and optimization: Use optimization algorithms such as gradient descent to train the model G. Let L(G) be the loss function, and the optimization goal is min GL(G).

[0143] Step 34: Integrate intelligent data collection models, data processing algorithms, and data mining capabilities into a unified data collection and processing system. Implement the automated process of the system to support continuous data collection, real-time processing, and value mining. Provide user interfaces and APIs so that users can easily access and use system functions.

[0144] The specific process includes:

[0145] Step 341, system integration: Integrate the data acquisition model N, the data processing algorithm set A and the data mining model G into a system. The output of the system is the value mining result, that is, V1 = G(A(N(S))).

[0146] Step 342: Set a scheduled task or event trigger mechanism to perform automated data collection, processing, and mining operations.

[0147] Step 35: Adaptive adjustment and optimization. Integrate monitoring and feedback mechanisms to monitor system performance and data collection and processing effects in real time. Implement adaptive adjustment algorithms to dynamically adjust and optimize the system according to actual application requirements. Provide a user configuration interface so that users can customize configuration and optimization according to business needs.

[0148] The specific process includes:

[0149] Step 351, monitoring and feedback: Let m(t') be the system performance monitoring index and t' be time. Monitor m(t') in real time and make feedback adjustments according to the preset threshold.

[0150] Step 352, adaptive algorithm implementation: dynamically adjust the parameters of the data acquisition model N, the data processing algorithm set A and the data mining model G according to the monitoring index m(t′).

[0151] For example, pseudo-code examples are used to comprehensively cover all aspects of data collection, processing, mining, automation and adaptive adjustment from demand analysis to final deployment and optimization, as follows:

[0152]

[0153]

[0154] This pseudo code provides a high-level overview of how to implement and optimize the key components of intelligent data acquisition and processing technology. In actual applications, each function will contain more detailed implementation steps and algorithms. The following is an exemplary explanation:

[0155] Explanation 1. buildSmartDataCollectionModel(), its purpose is to build a smart data collection model. The steps are:

[0156] analyzeDataCollectionRequirements(): Analyzes the requirements for data collection, including data type, frequency, source, etc.

[0157] designAndImplementDataCollectionModel(): Design and implement the data collection model according to requirements, which may include the selection of data collectors, the formulation of data collection strategies, etc.

[0158] integrateDataCleaningAndPreprocessing(): Integrates data cleaning and preprocessing functions to ensure that the quality of the collected data meets the requirements of subsequent processing.

[0159] Explanation 2: optimizeDataProcessingAlgorithms(), its purpose is to optimize the data processing algorithm. Its steps:

[0160] selectAppropriateDataProcessingAlgorithms(): Select a data processing algorithm that is suitable for the current data set and processing requirements.

[0161] implementAndOptimizeAlgorithms(): Implement these algorithms and optimize them to improve processing efficiency and accuracy.

[0162] integrateParallelAndDistributedComputing(): Integrates parallel and distributed computing technologies to meet the needs of large-scale data processing.

[0163] Explanation 3: enhanceDataMiningCapabilities() is used to enhance data mining capabilities. The steps include:

[0164] developAdvancedDataMiningAlgorithms(): Develop advanced data mining algorithms to discover hidden patterns and knowledge in data.

[0165] implementAndTrainDataMiningModels(): Implement and train data mining models to extract useful information.

[0166] integrateVisualizationTools(): Integrate visualization tools to present data mining results more intuitively.

[0167] Explanation 4: automateDataCollectionProcessingAndMining(), its purpose is to automate the data collection, processing and mining process. Its steps include:

[0168] integrateDataCollectionModelAlgorithmsAndMining(): Integrates data collection models, processing algorithms, and data mining techniques into an automated process.

[0169] implementAutomationWorkflow(): Implement automated workflow to ensure that the entire process from data collection to mining is automated.

[0170] provideUserInterfaceAndAPI(): Provides user interface and API to facilitate user interaction and system integration.

[0171] Explanation 5, adaptiveAdjustmentAndOptimization(), its purpose is to achieve adaptive adjustment and optimization. Its steps include:

[0172] integrateMonitoringAndFeedbackMechanisms(): Integrate monitoring and feedback mechanisms to monitor the performance and effectiveness of the system in real time.

[0173] implementAdaptiveAdjustmentAlgorithms(): Implements adaptive adjustment algorithms to automatically adjust system parameters and configurations based on monitoring results.

[0174] provideUserConfigurationInterface(): Provides a user configuration interface, allowing users to adjust system behavior according to actual needs.

[0175] The purpose of the main program is to execute all the above functions in sequence, build and optimize a complete intelligent data acquisition and processing system. Its steps include:

[0176] Call buildSmartDataCollectionModel() to build the data collection model.

[0177] Call optimizeDataProcessingAlgorithms() to optimize the data processing algorithm.

[0178] Call enhanceDataMiningCapabilities() to enhance data mining capabilities.

[0179] Call automateDataCollectionProcessingAndMining() to automate the data collection, processing, and mining processes.

[0180] Call adaptiveAdjustmentAndOptimization() to implement adaptive adjustment and optimization of the system.

[0181] Finally, it should be noted that the above content is only used to illustrate the technical solution of the present invention, rather than to limit the scope of protection of the present invention. Simple modifications or equivalent substitutions of the technical solution of the present invention by ordinary technicians in this field do not deviate from the essence and scope of the technical solution of the present invention.

Claims

1. A data collection method based on edge computing and satellite communication technology, characterized in that: The following steps are involved: Step 1: Build an edge cloud service platform by integrating cloud computing, edge computing and 5G communication technologies. The construction process is as follows: Step 11: Design and deploy network architecture; Step 12: Deploy computing nodes at the edge of the network. Step 13: Raw data collection and processing; Step 14: Data storage and access management; Step 15: Deploy real-time computing tasks for data processing and analysis; Step 16: Monitor and optimize data; Step 2: Optimize satellite communication protocols, modulation and demodulation technology, and signal processing technology to perform satellite communication data transmission; the process is: Step 21, optimizing satellite communication protocols; Step 22, optimizing modulation technology and demodulation technology; Step 23: Optimizing signal processing technology; Step 24: Perform system integration and testing; Step 3: Perform intelligent data collection and processing; the process is: Step 31: Building an intelligent data collection model; Step 32: Optimize the data processing algorithm; Step 33: construct a data mining model based on the data mining algorithm, and perform training and optimization; Step 34: Integrate intelligent data collection models, data processing algorithms, and data mining capabilities; Step 35: Adaptive adjustment and optimization.

2. According to claim 1, a data collection method based on edge computing and satellite communication technology is characterized in that: In step 11, a network architecture including 5G base stations, edge computing nodes, and cloud computing centers is designed, and the 5G network is deployed; the shortest path algorithm is used for design and deployment.

3. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 12, a multi-objective optimization algorithm is used to deploy computing nodes, wherein the multi-objective optimization algorithm is any one of a scalarization method, a goal programming method, an ε-constraint method, and a Pareto optimal solution.

4. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 21, the optimization process is as follows: step 211, defining quantitative indicators of error rate, protocol complexity and compatibility; Step 212: setting constraints: including complexity constraints and compatibility constraints; Step 213, formulate an optimization strategy: optimize the error rate, and optimize the balance between complexity and compatibility; Step 214: Implement and evaluate based on the optimization strategy.

5. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 22, the optimization process is: Step 221, defining the objective function: including minimizing the signal distortion function and maximizing the transmission rate function; Step 222: setting constraints: including algorithm complexity and power consumption; Step 223, selecting an optimization algorithm and formulating a solution strategy; Step 224: Implement and evaluate the solution strategy.

6. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 23, the optimization process is: Step 231: define the target processing function; Step 232, setting constraints: including processing delay and algorithm stability; Step 233: Select an optimization algorithm and formulate a solution strategy; Step 234: Evaluate and verify according to the solution strategy.

7. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 24, the system testing process is: First, data transmission rate test: The test goal is to ensure that the data transmission rate R′ meets the preset minimum requirement R′min; R′=data volume / time, verify that R′≥R′min; Second, bit error rate test: The test goal is to ensure that the bit error rate BER' is lower than the preset maximum tolerance value BER' max ; BER′=number of error bits / total number of transmitted bits, verify BER′≤BER′ max ; Third, system stability test: Fourth, coverage test.

8. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 31, the construction process includes the following steps: Step 311, data collection requirements analysis: Set the data set to be collected, A1 is a data source set consisting of multiple data sources A1′; for each data source A1′∈A1, define its data contribution c(A1′) and collection difficulty d(A1′); the goal is to minimize the collection difficulty while maximizing the data contribution: min∑ A1′∈A1 d(A1′)-λ∑ A1′∈A1 c(A1′), λ is the weight coefficient; Step 312, data collection model design: Let N be the data collection model, take data source A1′ as the input of the data collection model, and take the collected data B as the model output; the goal of the data collection model is to minimize the collection error: minN∑ A1′∈A1 ‖M(A1′)-y true (A1′)‖ 2 , where y true (A1′) is the real data.

9. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 32, the optimization process includes: Step 321, select an algorithm: for algorithm a, define its processing speed v(a) and processing accuracy p(a); the goal is to minimize the processing time while maximizing the processing accuracy: min a∈A [1 / v(a)]-μ×p(a), where μ is the weight coefficient and A is the set of data processing algorithms; Step 322, algorithm implementation and optimization: Let f(b) be the data processing function, b be the input data, and d be the output data; the optimization goal is to minimize the error of the output data: min f ∑ b∈D1 ‖f(b)-d expected (b)‖ 2 , where d expected (b) is the expected output, and D1 is the given data set.

10. The data collection method based on edge computing and satellite communication technology according to claim 1, characterized in that: In step 33, the process includes: Step 331, construct a data mining model: Let G be the data mining model, X be the input feature set, and Y be the target variable; the goal of the data mining model is to minimize the prediction error: min G ∑(x,y)∈(X,Y)‖G(x)-y‖ 2 , where (x,y) is any data coordinate in the set (X,Y); Step 332, model training and optimization: Use the optimization algorithm to train the model G, set L(G) as the loss function, and the optimization target is minGL(G).

Images