Method and system for dynamically adjusting and optimizing network performance of low-orbit satellite communication terminal

Abstract

The invention provides a low-orbit satellite communication terminal network performance dynamic tuning method and system, and relates to the technical field of low-orbit satellite communication. Comprehensive data such as relative movement of a satellite and a terminal and signal transmission loss are acquired; preprocessing the integrated data, inputting the preprocessed integrated data into a time sequence prediction model containing LSTM and Transform, generating a signal quality and intensity index sequence and a link stability level in a future preset time window, and carrying out dynamic optimization based on the generated indexes, including adjusting a terminal modulation mode according to the signal quality; combining the link stability and the signal strength to aggregate the link, and optimizing the switching opportunity and path; allocating resources according to the signal strength and the service characteristics; and optimizing the energy consumption according to the relative motion and the business characteristics. Through accurate prediction and multi-dimensional adjustment and optimization, the communication efficiency and reliability of the low-orbit satellite communication terminal network are improved, energy consumption and delay are reduced, and good communication quality is guaranteed.

Description

Technical Field

[0001] This invention relates to a method and system for dynamic optimization of network performance of low-Earth orbit satellite communication terminals, belonging to the field of low-Earth orbit satellite communication technology. Background Technology

[0002] With the rapid development of low-Earth orbit (LEO) satellite constellations, they have become crucial for building integrated air-space-ground networks due to their advantages such as low transmission latency and wide coverage. However, the high-speed movement of LEO satellites relative to ground terminals causes continuous and rapid changes in communication link characteristics (such as signal strength and channel quality), accompanied by frequent inter-satellite handovers. This poses a significant challenge to maintaining stable and efficient network communication services.

[0003] In existing technologies, such as the Chinese invention patent application with publication number CN119135246B, a resource scheduling method for integrated low-Earth orbit satellite communication and navigation is disclosed. This patent primarily relies on the static position information of users and satellites, constructs a priority sequence by calculating the geometrical precision factor (GDOP), and selects a set of service satellites for the user and performs beamforming design based on coalition game theory. The core of this patent lies in static or quasi-static optimization based on relative position; however, the optimization dimension is singular, mainly relying on spatial geometric relationships, failing to fully consider multi-dimensional dynamic factors such as signal transmission loss, dynamic interference, link interruption risks, and terminal service characteristics. Furthermore, its decision-making lacks foresight, being a passive, reactive scheduling method. It cannot make forward-looking resource allocation and switching decisions based on predicted trends in link status, making it difficult to adapt to the rapidly changing time-varying characteristics of links in low-Earth orbit satellite scenarios, easily leading to problems such as untimely switching and mismatched resource allocation.

[0004] Therefore, there is an urgent need for a method to optimize the performance of low-orbit satellite communication terminal networks that can integrate multi-dimensional dynamic data, have forward-looking prediction capabilities, and achieve closed-loop dynamic tuning. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention proposes a method and system for dynamic optimization of the network performance of low-orbit satellite communication terminals.

[0006] The technical solution of the present invention is as follows: On one hand, the present invention provides a method for dynamic optimization of the network performance of a low-Earth orbit satellite communication terminal, the method comprising: Acquire comprehensive data on low-Earth orbit satellite communication links and services, including relative motion data between satellites and terminals, signal transmission loss data, link interference data, link interruption risk data, and terminal service characteristic data; Based on comprehensive data of low-Earth orbit satellite communication links and services, a time-series prediction model is used to generate comprehensive satellite communication link indicators within a future preset time window, including signal quality indicator sequence, signal strength indicator sequence, and link stability level. Based on comprehensive satellite communication link indicators, and combined with comprehensive data on low-Earth orbit satellite communication links and services, dynamic optimization of low-Earth orbit satellite communication terminal network performance is performed. This includes: adjusting terminal modulation mode based on signal quality index sequences; aggregating links and optimizing switching timing and communication paths based on link stability levels and signal strength index sequences; allocating communication resources based on signal strength index sequences and terminal service characteristic data; and optimizing terminal power consumption based on satellite-terminal relative motion data and terminal service characteristic data.

[0007] Preferably, the relative motion data between the satellite and the terminal is used to calculate the relative motion speed and signal frequency deviation between the satellite and the terminal, including frequency offset data output by the phase-locked loop, phase error data of the I / Q signal, auxiliary positioning and velocity estimation data provided by the satellite navigation system, real-time orbital position and velocity prediction data of the satellite generated from ephemeris data, and frequency offset measurement data of pilot symbols in the channel sounding signal.

[0008] Preferably, the signal transmission loss data is used to quantify the degree of signal attenuation from the satellite to the terminal, including theoretical data of basic path loss, dynamic environmental loss data, measured received signal strength index (RSSI) and signal-to-noise ratio (SNR) historical data, wherein the dynamic environmental loss data includes real-time distance and elevation angle related loss, weather-related loss, and antenna polarization mismatch loss.

[0009] Preferably, the link interference data is used to assess the impact of various types of interference on link quality, including co-channel interference data, adjacent-channel interference data, cross-polarization interference data, and ground system interference data. The co-channel interference data includes the strength of the co-channel satellite signal, the power ratio of the interference to the useful signal, and the azimuth and elevation angles of the interference source. The adjacent-channel interference data includes the adjacent channel leakage ratio, receiver adjacent channel selectivity, and filter suppression characteristics. The cross-polarization interference data includes cross-polarization discrimination and polarization antenna alignment error.

[0010] Preferably, based on integrated data of low-Earth orbit satellite communication links and services, a time-series prediction model is used to generate integrated satellite communication link indicators for a future preset time window. Specifically: Outlier removal and missing value imputation preprocessing are performed on the integrated data of low-Earth orbit satellite communication links and services. The preprocessed integrated data of low-Earth orbit satellite communication links and services is then input into a time series prediction model, which includes a Long Short-Term Memory (LSTM) network and a Transformer model. The LSTM extracts local time series features of the integrated data of low-Earth orbit satellite communication links and services, while the Transformer model models the global dependencies of the data. The output includes a signal quality index sequence including a signal-to-noise ratio (SNR) sequence, a signal strength index sequence including an RSSI sequence, and a link stability level. The link stability level is determined as high, medium, or low based on a comprehensive assessment of signal quality, interference level, and outage risk data.

[0011] Preferably, the step of adjusting the terminal modulation mode based on the signal quality index sequence includes establishing a mapping relationship between the modulation mode and the signal quality index, obtaining a predicted value for a future preset time window based on the historical signal quality index sequence, inputting the predicted value into the mapping relationship to obtain the corresponding modulation mode, and triggering the terminal to switch between different modulation modes based on a preset hysteresis threshold and the corresponding modulation mode, wherein the hysteresis threshold for switching from low-order modulation to high-order modulation is greater than the hysteresis threshold for switching from high-order modulation to low-order modulation.

[0012] Preferably, the link aggregation based on link stability level and signal strength index sequence, and the optimization of handover timing and communication path are as follows: In the overlapping areas of the satellite constellation's visibility period, the bandwidth of multiple links is aggregated through the multipath transmission control protocol, and traffic is dynamically allocated; The satellite handover timing is optimized based on a reinforcement learning algorithm. The reinforcement learning algorithm uses link quality indicators, network status, terminal context and time prediction as the state space and handover actions as the action space. The optimal handover timing is calculated by constructing a reward function, which includes throughput, handover cost, packet loss penalty and stability reward. A time-varying graph model is constructed based on ephemeris data of satellites with future preset time windows and inter-satellite link models. The optimal communication path is calculated by combining the shortest path algorithm and heuristic search algorithm.

[0013] Preferably, the allocation of communication resources based on signal strength index sequences and terminal service characteristic data specifically includes: A centralized scheduling mechanism based on software-defined networking (SDN) is adopted to generate a time-frequency resource allocation table based on signal strength index sequences and terminal service characteristic data. When the utilization rate of satellite resources or the delay of the service queue reaches the preset scheduling threshold, cross-satellite resource scheduling is triggered.

[0014] Preferably, the optimization of terminal energy consumption based on satellite-terminal relative motion data and terminal service characteristic data includes: A satellite overpass schedule is generated based on the relative motion data between the satellite and the terminal, and the terminal sleep cycle is adjusted based on the satellite overpass schedule. Based on the terminal service characteristic data, differential coding and compressed sensing technology are used to reduce the transmission volume of uplink data, and downlink high-frequency demand data is pre-cached to ground station nodes to reduce real-time transmission demand.

[0015] On the other hand, the present invention also provides a dynamic performance optimization system for low-Earth orbit satellite communication terminal networks, the system comprising: The data acquisition module is used to acquire comprehensive data on low-orbit satellite communication links and services, including relative motion data between satellites and terminals, signal transmission loss data, link interference data, link interruption risk data, and terminal service characteristic data. The indicator prediction module, based on the comprehensive data of low-Earth orbit satellite communication links and services, generates comprehensive satellite communication link indicators within a preset future time window through a time-series prediction model. These indicators include signal quality indicator sequences, signal strength indicator sequences, and link stability levels. The performance optimization module dynamically optimizes the network performance of low-Earth orbit (LEO) satellite communication terminals based on comprehensive satellite communication link indicators and combined with comprehensive data on LEO satellite communication links and services. This includes: adjusting the terminal modulation mode based on signal quality indicator sequences; aggregating links and optimizing handover timing and communication paths based on link stability levels and signal strength indicator sequences; allocating communication resources based on signal strength indicator sequences and terminal service characteristic data; and optimizing terminal power consumption based on satellite-terminal relative motion data and terminal service characteristic data.

[0016] The present invention has the following beneficial effects: 1. This invention relates to a method and system for dynamic optimization of network performance of low-Earth orbit (LEO) satellite communication terminals. By preprocessing the comprehensive data of LEO satellite communication links and services and inputting it into a time-series prediction model, the local time-series features and global dependencies of the data are effectively extracted. The signal quality index sequence, signal strength index sequence, and link stability level output by the time-series prediction model provide an accurate basis for subsequent dynamic optimization of network performance. This helps to plan the modulation mode, link aggregation, and resource allocation of the terminal in advance, and avoids communication quality degradation due to sudden signal changes during actual communication. 2. This invention is a method and system for dynamic optimization of network performance of low-Earth orbit satellite communication terminals. By establishing a mapping relationship between modulation mode and signal quality index, and triggering the terminal to switch between different modulation modes based on the predicted signal quality index, the modulation mode of the terminal is matched with the actual signal quality. A larger hysteresis threshold is set when switching from low-order modulation to high-order modulation to avoid frequent modulation mode switching, reduce switching overhead and communication instability, and help to improve communication efficiency and spectrum utilization as much as possible while ensuring communication quality. 3. This invention provides a method and system for dynamic performance optimization of low-Earth orbit (LEO) satellite communication terminal networks. It aggregates bandwidth from multiple links and dynamically allocates traffic through a multi-path transmission control protocol, fully utilizing the resources of the satellite constellation to improve communication bandwidth and reliability. Furthermore, this invention utilizes reinforcement learning algorithms to optimize satellite handover timing, ensuring communication continuity while reducing handover overhead and packet loss rate. In addition, this invention calculates the optimal communication path based on time-varying graph models and shortest path algorithms, effectively reducing communication latency, improving communication efficiency, ensuring fast and stable data transmission, and enhancing the overall performance of the LEO satellite communication terminal network. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] It should be understood that the step numbers used in the text are for ease of description only and are not intended to limit the order in which the steps are performed.

[0020] It should be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.

[0021] The terms “comprising” and “including” indicate the presence of the described feature, whole, step, operation, element and / or component, but do not exclude the presence or addition of one or more other features, wholes, steps, operations, elements, components and / or collections thereof.

[0022] The term “and / or” refers to any combination of one or more of the associated listed items, as well as all possible combinations, and includes these combinations.

[0023] Example 1: like Figure 1 As shown in the figure, this embodiment provides a method for dynamic optimization of the network performance of a low-Earth orbit satellite communication terminal, the method comprising: S1. Acquire comprehensive data on low-Earth orbit satellite communication links and services, including relative motion data between satellites and terminals, signal transmission loss data, link interference data, link interruption risk data, and terminal service characteristic data; S11. The relative motion data between the satellite and the terminal is used to calculate the relative motion speed and signal frequency deviation between the satellite and the terminal, including the frequency offset data output by the phase-locked loop, the phase error data of the I / Q signal, the auxiliary positioning and velocity estimation data provided by the satellite navigation system, the real-time orbital position and velocity prediction data of the satellite generated by the ephemeris data, and the frequency offset measurement data of the pilot symbols in the channel sounding signal. Preferably, in this embodiment, the satellite navigation system prioritizes the BeiDou system to adapt to domestic power scenarios. If the terminal is in a multi-system coverage area, GPS data can be integrated to improve positioning accuracy and control the positioning error within 10 meters. Ephemeris data is received every second through the satellite broadcast channel to synchronously update the satellite's real-time orbital position (longitude, latitude, and altitude) and velocity prediction data, thereby matching the typical high-speed motion characteristics of low-orbit satellites at 7.8 km / s and avoiding data deviation caused by orbital position lag. The phase-locked loop outputs frequency offset data every 10 ms, and the I / Q signal phase error is sampled every 5 ms by the baseband signal processing module. The average value of 10 consecutive samples is taken as the effective data to reduce instantaneous noise interference. At the same time, 8 pilot symbols are inserted into each frame of communication signal. The Doppler frequency offset generated by the relative motion between the satellite and the terminal is corrected by measuring the frequency offset of the pilot symbols to ensure that the measurement accuracy can reach ±1 Hz. S12. The signal transmission loss data is used to quantify the attenuation of the signal from the satellite to the terminal, including theoretical basic path loss data, dynamic environmental loss data, measured received signal strength index (RSSI) and signal-to-noise ratio (SNR) historical data. The theoretical basic path loss data is calculated based on a free-space path loss model and is expressed by the following formula: ; In the formula, Based on theoretical data of path loss; For communication frequencies, the typical 1.2-1.6 GHz band for low-Earth orbit satellites was selected; The straight-line distance between the satellite and the terminal is calculated jointly from ephemeris data and terminal positioning data. The dynamic environmental loss data includes real-time distance and elevation angle related loss, weather-related loss (such as rain attenuation and cloud attenuation), and antenna polarization mismatch loss. In this embodiment, when the satellite elevation angle is <10°, the loss will increase by an additional 5-10dB due to ground obstruction and enhanced multipath effect. The loss decreases by 1dB for every 5° increase in elevation angle. Weather-related losses are obtained by combining the terminal's built-in miniature meteorological sensor (measuring temperature, humidity, and rainfall) with broadcast data from ground meteorological stations. Rain attenuation is calculated according to the ITU-RP.838 standard (e.g., when the rainfall is 5mm / h, the rain attenuation in the 1.5GHz band is about 0.5dB / km), and cloud attenuation is calculated at 0.1dB / km. Antenna polarization mismatch loss will occur when the polarization direction of the terminal antenna deviates from the polarization direction of the satellite signal by >10°, resulting in a loss of 0.5-2dB. This needs to be corrected in real time by the antenna attitude adjustment module to control the deviation within 5°. The measured RSSI and SNR are sampled every 2ms, storing nearly 1 minute of historical data for subsequent training and verification of the time series prediction model, ensuring the continuity of the model input data. S13. The link interference data is used to assess the impact of various types of interference on link quality, including co-channel interference data, adjacent-channel interference data, cross-polarization interference data, and ground system interference data. In this embodiment, co-channel interference data is acquired by scanning a 20MHz co-channel band every 100ms using the terminal spectrum monitoring module, recording the co-channel satellite signal strength, the power ratio (J / N) of the interference to the useful signal, and the azimuth and elevation angles of the interference source. The typical threshold for J / N is <-15dB; exceeding this threshold indicates severe interference, requiring the triggering of an interference avoidance process. Adjacent-channel interference data mainly monitors the adjacent channel leakage ratio (ACLR) and the receiver adjacent channel leakage ratio. Selective ACS and filter suppression characteristics require ACLR ≥ 30dB and ACS ≥ 40dB. If the monitored values ​​do not meet the requirements, filter gain adjustment needs to be initiated to improve the suppression characteristics by 2-5dB to reduce adjacent channel interference. Cross-polarization interference data is obtained by monitoring cross-polarization discrimination XPD and polarization antenna alignment error. XPD needs to be ≥ 25dB. If it is lower than this value, the terminal polarization antenna angle is adjusted (in 1° increments) until XPD meets the requirements, while the polarization antenna alignment error is controlled within 3°. Ground system interference mainly comes from ground microwave communication systems. By monitoring the frequency and power of ground interference signals, if the interference power exceeds -100dBm, an interference avoidance request is sent to the satellite gateway, and the satellite communication frequency is adjusted (offset by 500kHz) to avoid the ground interference source. S14. The link interruption risk data includes satellite obstruction probability, atmospheric attenuation prediction, and hardware failure indicators. The risk level is calculated comprehensively using a Bayesian network model and is divided into high, medium, and low levels. In this embodiment, the satellite obstruction probability uses 30-meter resolution digital elevation model (DEM) data, combined with satellite orbital parameters (elevation and azimuth) and terminal location. A ray tracing algorithm is used to determine whether the satellite signal is obstructed by terrain (such as mountains) or buildings. An obstruction probability >60% is considered high risk. Atmospheric attenuation prediction is obtained by receiving the ionospheric scintillation index (IoS) broadcast by the ionospheric monitoring station. (index) to achieve, When ionospheric scintillation causes signal attenuation >10dB, it is considered high risk. A decay of 5-10 dB when the value is between 0.2 and 0.5 is considered medium risk. If the time attenuation is <5dB, it is judged as low risk; the main hardware failure indicators are the operating temperature (threshold -40℃~70℃) and operating time (lifetime 10,000 hours) of the terminal amplifier, which are monitored in real time by temperature sensors. If the temperature exceeds the threshold or the operating time approaches 90% of the lifetime, the hardware failure risk is upgraded to medium risk, and a failure warning needs to be sent to the terminal management platform; the Bayesian network model takes the occlusion probability, atmospheric attenuation risk, and hardware failure risk as input nodes and the link interruption risk as output node. The network parameters are trained by historical failure data, and the final output risk level includes high risk (interruption probability >30%), medium risk (interruption probability between 10% and 30%), and low risk (interruption probability <10%). S15. Terminal service characteristic data includes service type (voice, video, and data), quality of service requirements (latency limit, bandwidth requirement, packet loss rate requirement), data volume, and service priority. In this embodiment, there is a clear correspondence between service type and quality of service requirements. For example, for emergency command voice services, the latency limit is ≤100ms, bandwidth requirement is ≥128kbps, and packet loss rate requirement is ≤0.1%; for real-time video surveillance services, the latency limit is ≤500ms, bandwidth requirement is ≥2Mbps, and packet loss rate requirement is ≤1%; for voice call services, the latency limit is ≤200ms, bandwidth requirement is ≥64kbps, and packet loss rate requirement is ≤0.5%; for non-real-time data (such as meter reading) services, the latency limit is ≤30s, bandwidth requirement is ≥10kbps, and packet loss rate requirement is ≤5%. Each terminal transmits… Each business request is initiated, and the data volume is recorded synchronously. For example, the data volume of a single emergency command voice request is less than 10KB, and the data volume of each frame of real-time video is approximately 50KB, providing a basis for subsequent resource allocation. Business priorities are dynamically adjusted based on application scenarios. When a power failure occurs, the priority of the emergency command business is automatically increased to the highest level (weight 1.0), while the priority of other businesses is reduced (e.g., the weight of real-time video is reduced to 0.6). After the failure is recovered, the priorities are restored to the default values ​​(emergency command 1.0, real-time video 0.8, voice 0.7, non-real-time data 0.3), ensuring the service quality of critical businesses.

[0024] S2. Based on the comprehensive data of low-orbit satellite communication links and services, a time-series prediction model is used to generate comprehensive satellite communication link indicators within a future preset time window, including signal quality indicator sequence, signal strength indicator sequence and link stability level. S21. Perform outlier removal and missing value filling preprocessing on the low-orbit satellite communication link and service integrated data. In this embodiment, outlier removal adopts the Z-score algorithm, and missing value filling adopts the linear interpolation method to ensure data continuity. Since the preprocessing methods are all commonly used in the field, they will not be described in detail here. S22. Input the preprocessed low-orbit satellite communication link and service integrated data into the time series prediction model. In this embodiment, the time series prediction model consists of a Long Short-Term Memory (LSTM) network and a Transformer model. The LSTM is used to extract local time series features of the data (such as short-term signal fluctuations), and the Transformer model is used to model the global dependencies of the data (such as the influence of long-term orbital periods). The model input is the historical data of the past 30 seconds (time series window step size T=60, 500ms per step), and the output is the predicted value for the next 10 seconds (step size 20, 500ms per step).

[0025] In the overall structure of the time series prediction model, the input layer receives data in the format (B, T, N) (B is the batch size of 32, and N is the feature dimension of 32), and the format is preserved by the data preprocessing layer. The LSTM local feature extraction module contains two bidirectional LSTM layers. The first layer contains 128 neurons, with Dropout=0.2 and return_sequences=True, and the output format is (B, T, 256). The second layer contains 64 neurons, with Dropout=0.2 and return_sequences=False, and the output format is (B, 128). The Transformer global dependency module first uses an embedding layer to process the 32... The 128-dimensional features are mapped to a 128-dimensional matrix, and positional encoding is added (preserving temporal information through sine and cosine functions). The data is then processed by a 4-layer encoder. Each encoder layer contains an 8-head self-attention mechanism, residual connections, a LayerNorm layer, and a feedforward network. The 8-head self-attention mechanism projects the input data into multiple subspaces and computes attention in parallel to capture the multifaceted dependencies of the data. After the self-attention computation is complete, the input and self-attention output are added through residual connections, and then the features are normalized by the LayerNorm layer to stabilize the training process. Finally, the data enters the feedforward network, a two-layer fully connected structure with a 512-dimensional intermediate layer for further feature extraction and transformation.

[0026] The data processed by the 4-layer encoder is output in (B, 128) format through global average pooling. The feature fusion and output layer concatenates the outputs of LSTM and Transformer to (B, 256), and after passing through two fully connected layers (128 neurons ReLU activated and 64 neurons ReLU activated), it is output in three branches. The three branches include SNR prediction branch (fully connected layer output T=20 steps, no activation), RSSI prediction branch (fully connected layer output T=20 steps, no activation), and stability classification branch (fully connected layer output 3D, Softmax activated).

[0027] Furthermore, during the training of the time series prediction model, the loss function is a weighted sum of the mean squared error (MSE) of SNR / RSSI prediction and the cross-entropy loss of stability classification (each with a weight of 0.5). The optimizer used is the Adam optimizer (initial learning rate 1e-4, decaying to 0.8 every 100 epochs). The training dataset contains 10,000 samples (each containing 30 seconds of historical data combined with 10 seconds of real data), which are divided into training, validation, and test sets in a 7:2:1 ratio. Training is stopped when the total loss of the validation set decreases by <0.001 for 10 consecutive epochs to ensure the convergence of the time series prediction model. S23. The signal quality index sequence output by the time-series prediction model is the SNR sequence (in dB) for the next 10 seconds, and the signal strength index sequence is the RSSI sequence (in dBm) for the next 10 seconds. The link stability level is determined based on a comprehensive assessment of signal quality, interference level, and outage risk data. High stability requires SNR > 20 dB, interference level < 0.1 (J / N < -10 dB), and outage risk < 10%. Medium stability requires 10 dB ≤ SNR ≤ 20 dB, 0.1 ≤ interference level ≤ 0.3 (-10 dB ≤ J / N ≤ 5 dB), and 10% ≤ outage risk ≤ 30%. Low stability requires SNR < 10 dB, interference level > 0.3 (J / N > -5 dB), or outage risk > 30%. The output format is to output the predicted sequence for the next 10 seconds every 100ms. For example, the SNR sequence is [18.2, 18.5, 19.1, ..., 20.3] (100 data points in total), and the RSSI sequence is [-52.3, -51.8, -51.2, ..., -49.5]. The stability level is "Medium" (probability 0.75). To ensure the accuracy of the indicator, the predicted value is compared with the measured value every 5 seconds. If the SNR prediction error is >2dB or the RSSI prediction error is >3dB, the model parameters are fine-tuned (the learning rate is updated to 1e). -5 (fine-tuning over 10 epochs) to continuously optimize prediction performance; S3. Based on comprehensive satellite communication link indicators, and combined with comprehensive data on low-Earth orbit satellite communication links and services, dynamically optimize the network performance of low-Earth orbit satellite communication terminals. This includes: adjusting the terminal modulation mode based on signal quality index sequences; aggregating links and optimizing switching timing and communication paths based on link stability levels and signal strength index sequences; allocating communication resources based on signal strength index sequences and terminal service characteristic data; and optimizing terminal power consumption based on satellite-terminal relative motion data and terminal service characteristic data. S31. In this embodiment, adjusting the terminal modulation mode includes establishing a mapping relationship between the modulation mode and the signal quality index, obtaining a predicted value for a future preset time window based on a historical signal quality index sequence, inputting the predicted value into the mapping relationship to obtain the corresponding modulation mode, and triggering the terminal to switch between different modulation modes based on a preset hysteresis threshold and the corresponding modulation mode.

[0028] Furthermore, the mapping relationship is based on the MCS and SNR threshold mapping table in the low-Earth orbit satellite communication scenario, as shown in Table 1: Table 1. MCS to SNR Threshold Mapping Table Based on Low-Earth Orbit Satellite Communication Scenarios

[0029] In this embodiment, the hysteresis threshold is set as the hysteresis threshold for switching from low-order modulation to high-order modulation. Hysteresis threshold for switching from higher-order modulation to lower-order modulation The handover trigger must meet a threshold condition for a certain period of time (50ms for upgrading to resist instantaneous fluctuations, and 30ms for downgrading to prevent rapid response degradation). For example, if the current signal is QPSK 3 / 4 (SNR≥8dB), and the predicted SNR value for the next 100ms is ≥15dB (12+3) for 5 consecutive times, then the handover to 8PSK 2 / 3 is triggered. The handover process takes <200ms, and a retransmission mechanism is used during this period to ensure that no data is lost. If the current signal is 8PSK 3 / 4 (SNR≥16dB), and the predicted SNR value for the next 100ms is ≤14dB (16-2) for 3 consecutive times, then the handover to 8PSK 2 / 3 is triggered. A notification is sent to the satellite before the handover to synchronously adjust the receiving parameters. S32. In this embodiment, link aggregation is implemented through the Multipath Transmission Control Protocol (MPTCP) in the overlapping region of the satellite constellation's visibility period. Specifically, in this embodiment: The visibility time of the two satellites is calculated using ephemeris data. If satellite A has 5 minutes of remaining visibility time and satellite B has been in its visibility window for 2 minutes, then the 3-minute visibility periods overlap. The terminal initiates TCP connections to both satellites to establish MPTCP substreams. The traffic allocation ratio is calculated based on the predicted SNR value corresponding to the substream, expressed by the formula: ; In the formula, For the first The flow allocation ratio corresponding to each sub-flow; For the first The rate of each sub-stream; This represents the total number of subflows. For example, substream 1 (satellite A, 8PSK, 3 / 4) has a rate of 8Mbps, and substream 2 (satellite B, 16APSK2 / 3) has a rate of 10.67Mbps. The traffic allocation ratio is 43:57, and the total bandwidth after aggregation is 18.67Mbps, which can meet the requirements of 4K real-time video (requiring 15Mbps). When the SNR prediction value of a substream drops to the switching threshold, the traffic ratio is automatically adjusted to avoid bandwidth waste. The handover timing optimization is based on the reinforcement learning algorithm DQN. The DQN model input is a 22-dimensional state space (including 4-dimensional current link state, 10-dimensional candidate satellite state, 5-dimensional terminal and service state, and 3-dimensional predicted state), and the output is a 3-dimensional action space (maintain connection, handover to satellite 1, handover to satellite 2). The reward function is expressed by the formula: ; In the formula, For the reward function; For throughput; For switching overhead; Penalty for packet loss; For stable rewards; for example =8Mbps =0.3 seconds =8 (packet loss rate 8%) =20 (no switching for 5 consecutive seconds and SNR>16dB), = 4.8 - 0.06 - 1.2 + 1 = 4.54.

[0030] Furthermore, the DQN model training employs empirical replay (capacity 100,000 samples), synchronization of the target network every 1000 steps, and parameter... (Represents the probability of the agent choosing a random action) Decreases linearly from 1.0 to 0.05. Training stops when the average reward fluctuation of 100 consecutive episodes is <5% and the switching success rate is >95%. When switching, the action with the highest Q value is selected to ensure the stability of the link after switching.

[0031] Communication path optimization requires constructing a time-varying graph model based on ephemeris data for future satellite time windows and inter-satellite link models, including: The satellite orbit parameters are analyzed (updated every 1 second), and the elevation angle between the terminal and the satellite is calculated (the conventional technical methods in this field will not be elaborated here; if the angle is >5°, the satellite is considered visible) and the inter-satellite distance is calculated (if the distance is <5000km, the satellite is considered to have a link). The edge weights are calculated using the following formula: ; In the formula, For nodes and The edge between The edge weight corresponding to the given time; For nodes and Between The distance at any given moment; in satellite communication, nodes can be terminals, satellites, or ground gateways, and the magnitude of the distance will affect the signal propagation time and loss. Based on the terminal's elevation angle, inter-satellite distance, and edge weights, the A* algorithm is used to calculate the optimal path, for example... At time 0, the node set includes a terminal, two satellites, and a ground gateway. The calculated edge weight between the terminal and the second satellite is 6.45. And the edge weight of 2.97 corresponding to the second satellite and the ground gateway. The total weight is 9.42. The path is superior to other paths, so the terminal → second satellite → ground gateway is selected. The path is recalculated every 1 second, and a switch is triggered when the weight increases by more than 20%. S33. In this embodiment, a centralized scheduling mechanism of Software-Defined Networking (SDN) is adopted to generate a time-frequency resource allocation table based on the signal strength index sequence and terminal service characteristic data. The centralized scheduling mechanism of SDN includes multiple stages, specifically: During the global state awareness phase, physical layer data (including predicted RSSI / SNR values ​​for each terminal and a total of 100×50 subcarriers of available time-frequency resources), service layer data (including service type, QoS requirements and priorities), and resource layer data (including resource utilization, queue length and latency) are collected every 50ms through the satellite gateway. The business priority ranking stage is based on priority weight × (1 / business waiting time). For example, emergency command (priority weight of 1.0, business waiting time of 40ms) scores 0.025, real-time video (priority weight of 0.8, business waiting time of 80ms) scores 0.01, voice (priority weight of 0.7, business waiting time of 120ms) scores 0.0058, and non-real-time data (priority weight of 0.3, business waiting time of 2000ms) scores 0.00015. The ranking result is: emergency command > real-time video > voice > non-real-time data. The time-frequency resource allocation decision-making stage satisfies the constraints of non-overlapping time-frequency resources, matching modulation method with SNR, and meeting service bandwidth requirements, and generates a time-frequency resource allocation table. For example, time slot F0 is allocated to terminal 001 (video, 8PSK3 / 4, 18dBm), F1 is allocated to terminal 003 (voice, QPSK3 / 4, 15dBm), F2 is reserved for emergency use, and F3 is allocated to terminal 005 (data, QPSK1 / 2, 12dBm). The optimization goal is to achieve a resource utilization rate ≥80% and a high-priority service latency ≤threshold. During the policy distribution and feedback phase, the allocation table is distributed to satellites and terminals every 10ms. Terminals adjust their transmission parameters, and the controller monitors the feedback data in real time. If the delay of high-priority services exceeds the threshold, the next round of scheduling is triggered. When the satellite resource utilization rate (frequency ≥ 85%, time slot ≥ 80%, power ≥ 90%) or service queue delay (emergency ≥ 50ms, real-time ≥ 100ms, non-real-time ≥ 500ms) reaches the preset threshold, or when the remaining visible time of the terminal is ≤ 10s, cross-satellite resource scheduling is triggered. For example, if the frequency utilization rate of satellite A is 92% for three consecutive cycles and the emergency service delay is 52ms, the controller requests the allocation of 5 subcarriers from satellite B (utilization rate 60%), so that the utilization rate of satellite A is reduced to 82% and the emergency service delay is reduced to 38ms. S34. In this embodiment, terminal power consumption optimization is performed from three aspects: intermittent wake-up, data compression, and pre-caching, wherein: Intermittent wake-up generates a satellite overpass schedule based on the relative motion data between the satellite and the terminal (visible window 5-15 minutes / time). For example, satellite 1 overpasses from 08:00:00 to 08:12:30 and satellite 2 overpasses from 08:18:45 to 08:29:10. The terminal goes into deep sleep during the invisible window (power consumption <10μA, RTC timing only). It wakes up and warms up 30 seconds before the overpass (including RF power-on, clock synchronization and ephemeris verification). Within the visible window, it completes link establishment (<200ms), data transmission, link maintenance (resource adjustment every 500ms) and sleep preparation (including RF power-off and data temporary storage). Based on 10 overpasses per day (total working hours 2 hours), the average daily power consumption is ≈1Wh, which is 91.7% lower than the normal mode (12Wh).

[0032] Data compression is applied to upstream data. Periodic static / quasi-static data (such as environmental monitoring temperature and humidity and mobile terminal location) uses differential encoding, and the encoding format is shown in Table 2. Table 2 Encoding Format Table

[0033] The original 12 bytes of data are compressed to 8 bytes, reducing the transmission volume by 33.3%; sparse data (such as sensor sampling data) are compressed using compressed sensing technology, which compresses N-dimensional data into 0.3N-dimensional measurement values ​​through a random Gaussian matrix, and the receiving end reconstructs the data using the OMP algorithm (error <5%). 100 bytes of data are compressed to 30 bytes, reducing the transmission volume. Pre-caching targets high-frequency downlink data, with ground stations using a weighted score based on business relevance, popularity, and timeliness. The decision to cache data is made, wherein the weighted score is expressed by the formula: ; For example, in the IoT terminal area of ​​a factory in a city's economic development zone, the V1.2.0 firmware update package (with a matching degree of 1.0, popularity of 0.82, and timeliness of 1.0, resulting in a weighted score of 0.928) is cached first. The terminal requests cached data when the satellite passes overhead, avoiding real-time retrieval from the backbone network and saving transmission latency.

[0034] Example 2: This embodiment provides a system for dynamically optimizing the network performance of a low-Earth orbit satellite communication terminal, the system comprising: The data acquisition module is used to acquire comprehensive data on low-orbit satellite communication links and services, including relative motion data between satellites and terminals, signal transmission loss data, link interference data, link interruption risk data, and terminal service characteristic data. The indicator prediction module, based on the comprehensive data of low-Earth orbit satellite communication links and services, generates comprehensive satellite communication link indicators within a preset future time window through a time-series prediction model. These indicators include signal quality indicator sequences, signal strength indicator sequences, and link stability levels. The performance optimization module dynamically optimizes the network performance of low-Earth orbit (LEO) satellite communication terminals based on comprehensive satellite communication link indicators and combined with comprehensive data on LEO satellite communication links and services. This includes: adjusting the terminal modulation mode based on signal quality indicator sequences; aggregating links and optimizing handover timing and communication paths based on link stability levels and signal strength indicator sequences; allocating communication resources based on signal strength indicator sequences and terminal service characteristic data; and optimizing terminal power consumption based on satellite-terminal relative motion data and terminal service characteristic data.

[0035] It is worth noting that the system described in this invention is based on the same inventive concept as the method described in Embodiment 1, and will not be repeated here.

[0036] In this embodiment of the invention, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, A and B simultaneously, or B alone. A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects have an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of singular or plural items. For example, at least one of a, b, and c can represent: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, and c can be single or multiple.

[0037] Those skilled in the art will recognize that the units and algorithm steps described in the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of electronic hardware and software. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0038] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0039] In several embodiments provided by this invention, any function, if implemented as a software functional unit and sold or used as an independent product, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0040] The above description is merely an embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A method for dynamically optimizing the network performance of a low-Earth orbit satellite communication terminal, characterized in that, The method includes: Acquire comprehensive data on low-Earth orbit satellite communication links and services, including relative motion data between satellites and terminals, signal transmission loss data, link interference data, link interruption risk data, and terminal service characteristic data; Based on comprehensive data of low-Earth orbit satellite communication links and services, a time-series prediction model is used to generate comprehensive satellite communication link indicators within a future preset time window, including signal quality indicator sequence, signal strength indicator sequence, and link stability level. Based on comprehensive satellite communication link indicators, and combined with comprehensive data on low-Earth orbit satellite communication links and services, dynamic optimization of low-Earth orbit satellite communication terminal network performance is performed. This includes: adjusting terminal modulation mode based on signal quality index sequences; aggregating links and optimizing switching timing and communication paths based on link stability levels and signal strength index sequences; allocating communication resources based on signal strength index sequences and terminal service characteristic data; and optimizing terminal power consumption based on satellite-terminal relative motion data and terminal service characteristic data.

2. The method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 1, characterized in that, The relative motion data between the satellite and the terminal is used to calculate the relative motion speed and signal frequency deviation between the satellite and the terminal. It includes frequency offset data output by the phase-locked loop, phase error data of the I / Q signals, auxiliary positioning and velocity estimation data provided by the satellite navigation system, real-time orbital position and velocity prediction data of the satellite generated from ephemeris data, and frequency offset measurement data of pilot symbols in the channel sounding signal.

3. The method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 2, characterized in that, The signal transmission loss data is used to quantify the degree of signal attenuation from satellite to terminal, including theoretical data of basic path loss, dynamic environmental loss data, measured received signal strength index (RSSI) and signal-to-noise ratio (SNR) historical data. The dynamic environmental loss data includes real-time distance and elevation angle related loss, weather-related loss, and antenna polarization mismatch loss.

4. The method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 3, characterized in that, The link interference data is used to assess the impact of various types of interference on link quality, including co-channel interference data, adjacent-channel interference data, cross-polarization interference data, and ground system interference data. The co-channel interference data includes the strength of the co-channel satellite signal, the power ratio of the interference to the useful signal, and the azimuth and elevation angles of the interference source. The adjacent-channel interference data includes the adjacent channel leakage ratio, the receiver adjacent channel selectivity, and the filter suppression characteristics. The cross-polarization interference data includes cross-polarization discrimination and polarization antenna alignment error.

5. The method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 4, characterized in that, Based on comprehensive data on low-Earth orbit satellite communication links and services, a time-series prediction model is used to generate comprehensive indicators of satellite communication links within a future preset time window. Specifically: Outlier removal and missing value imputation preprocessing are performed on the integrated data of low-Earth orbit satellite communication links and services. The preprocessed integrated data of low-Earth orbit satellite communication links and services is then input into a time series prediction model, which includes a Long Short-Term Memory (LSTM) network and a Transformer model. The LSTM extracts local time series features of the integrated data of low-Earth orbit satellite communication links and services, while the Transformer model models the global dependencies of the data. The output includes a signal quality index sequence including a signal-to-noise ratio (SNR) sequence, a signal strength index sequence including an RSSI sequence, and a link stability level. The link stability level is determined as high, medium, or low based on a comprehensive assessment of signal quality, interference level, and outage risk data.

6. The method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 5, characterized in that, The method of adjusting the terminal modulation mode based on the signal quality index sequence includes establishing a mapping relationship between the modulation mode and the signal quality index, obtaining the predicted value of the future preset time window based on the historical signal quality index sequence, inputting the predicted value into the mapping relationship to obtain the corresponding modulation mode, and triggering the terminal to switch between different modulation modes based on a preset hysteresis threshold and the corresponding modulation mode, wherein the hysteresis threshold for switching from low-order modulation to high-order modulation is greater than the hysteresis threshold for switching from high-order modulation to low-order modulation.

7. The method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 6, characterized in that, The specific steps for aggregating links and optimizing handover timing and communication paths based on link stability level and signal strength index sequence are as follows: In the overlapping areas of the satellite constellation's visibility period, the bandwidth of multiple links is aggregated through the multipath transmission control protocol, and traffic is dynamically allocated; The satellite handover timing is optimized based on a reinforcement learning algorithm. The reinforcement learning algorithm uses link quality indicators, network status, terminal context and time prediction as the state space and handover actions as the action space. The optimal handover timing is calculated by constructing a reward function, which includes throughput, handover cost, packet loss penalty and stability reward. A time-varying graph model is constructed based on ephemeris data of satellites with future preset time windows and inter-satellite link models. The optimal communication path is calculated by combining the shortest path algorithm and heuristic search algorithm.

8. A method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 7, characterized in that, The specific method for allocating communication resources based on signal strength index sequences and terminal service characteristic data is as follows: A centralized scheduling mechanism based on software-defined networking (SDN) is adopted to generate a time-frequency resource allocation table based on signal strength index sequences and terminal service characteristic data. When the utilization rate of satellite resources or the delay of the service queue reaches the preset scheduling threshold, cross-satellite resource scheduling is triggered.

9. A method for dynamic performance optimization of a low-orbit satellite communication terminal network according to claim 8, characterized in that, The optimization of terminal energy consumption based on satellite-terminal relative motion data and terminal service characteristic data includes: A satellite overpass schedule is generated based on the relative motion data between the satellite and the terminal, and the terminal sleep cycle is adjusted based on the satellite overpass schedule. Based on the terminal service characteristic data, differential coding and compressed sensing technology are used to reduce the transmission volume of uplink data, and downlink high-frequency demand data is pre-cached to ground station nodes to reduce real-time transmission demand.

10. A system for dynamically optimizing the network performance of a low-Earth orbit satellite communication terminal, characterized in that, The system includes: The data acquisition module is used to acquire comprehensive data on low-orbit satellite communication links and services, including relative motion data between satellites and terminals, signal transmission loss data, link interference data, link interruption risk data, and terminal service characteristic data. The indicator prediction module, based on the comprehensive data of low-Earth orbit satellite communication links and services, generates comprehensive satellite communication link indicators within a preset future time window through a time-series prediction model. These indicators include signal quality indicator sequences, signal strength indicator sequences, and link stability levels. The performance optimization module dynamically optimizes the network performance of low-Earth orbit (LEO) satellite communication terminals based on comprehensive satellite communication link indicators and combined with comprehensive data on LEO satellite communication links and services. This includes: adjusting the terminal modulation mode based on signal quality indicator sequences; aggregating links and optimizing handover timing and communication paths based on link stability levels and signal strength indicator sequences; allocating communication resources based on signal strength indicator sequences and terminal service characteristic data; and optimizing terminal power consumption based on satellite-terminal relative motion data and terminal service characteristic data.

Images