A network connection optimization method, device, medium and program product for cellular communication module

Abstract

The present application provides a network connection optimization method, device, medium and program product for a cellular communication module, which collects parameter information representing the current network connection status, performs feature processing on the parameter information, and constructs a multidimensional feature vector; for multiple candidate connection strategies, calculates a score value for each candidate connection strategy based on the multidimensional feature vector and a preset weight value; normalizes each score value to obtain a probability distribution of the candidate connection strategy; based on the probability distribution, periodically performs probability sampling from multiple candidate connection strategies to determine the target connection strategy of the current cycle and determine whether to perform network switching; based on the performance evaluation results of the current connection status, dynamically adjusts the temperature parameter of the calculated probability distribution, and optimizes the sampling behavior of the connection strategy of the next cycle. The present application introduces the temperature parameter as a control factor to improve the connection stability, strategy switching flexibility and overall network access quality of the cellular communication module.

Description

Technical Field

[0001] The present application relates to the field of wireless communication technology, and in particular to a network connection optimization method, device, medium and program product for a cellular communication module. Background Art

[0002] With the rapid development of cellular communication technology, communication modules supporting standards such as LTE (4G) and NR (5G) have been widely used in scenarios such as the Industrial Internet of Things (IIoT), the Internet of Vehicles (IoV), and remote control. However, in complex or dynamic network environments, the connection stability of communication modules still faces many challenges. Frequent disconnections and access failures are common, making it difficult to meet the actual demand for continuous and highly reliable communication.

[0003] In actual applications, the disconnection problem of cellular communication modules is usually caused by the following factors:

[0004] Frequent cell handovers: In areas with high mobility or at the edge of network coverage, the module frequently performs cell handovers. If handovers fail or are delayed, connection interruption may occur.

[0005] Severe signal quality fluctuations: Key wireless metrics such as reference signal received power (RSRP) and signal-to-noise ratio (SNR) are significantly affected by factors such as obstruction, interference, and reflection in real-world environments. If these metrics remain low or fluctuate dramatically for a long time, they will directly impact link stability and data transmission quality.

[0006] Improper frequency band selection: Cellular communication systems support multiple frequency bands, each with different coverage, interference, and communication capacity. If the module fails to dynamically select the optimal frequency band based on the current scenario (for example, relying solely on the default configuration to use a high-capacity but weak-penetration frequency band), connection performance may degrade or even lead to disconnection.

[0007] Abnormal carrier network or IP channel: Some areas may have insufficient coverage from a single carrier, or the connected IP channel may have problems such as high load, severe packet loss, and slow connection establishment, which may affect communication stability or cause frequent disconnections.

[0008] Therefore, existing technologies urgently need to be improved to achieve dynamic evaluation and optimization of communication module access strategies and improve connection stability in complex wireless environments. Summary of the Invention

[0009] In response to the shortcomings of the existing technology, the present application provides a network connection optimization method, device, medium and program product for a cellular communication module, which is at least used to solve the problems of unstable connection, delayed policy response and lack of adaptive decision-making mechanism of the cellular communication module in the existing technology.

[0010] In order to achieve the above objectives and other advantages, some embodiments of the present application provide the following aspects:

[0011] In a first aspect, some embodiments of the present application provide a network connection optimization method for a cellular communication module, comprising:

[0012] Real-time collection of parameter information used to characterize the current network connection status, and feature processing of the parameter information to construct a multi-dimensional feature vector. The parameter information includes: base station cell switching frequency, operating frequency band type, received signal power, signal-to-noise ratio, operator network service quality indicators, and connection stability indicators;

[0013] For the plurality of candidate connection strategies that have been constructed, respectively calculating a score value of each candidate connection strategy based on the multidimensional feature vector and a preset weight value;

[0014] Normalizing each of the score values ​​to obtain a probability distribution of the candidate connection strategies;

[0015] According to the probability distribution, periodically perform probability sampling from the plurality of candidate connection strategies to determine a target connection strategy for the current period;

[0016] determining whether the target connection strategy is consistent with the current connection strategy of the cellular communication module; if not, controlling the cellular communication module to disconnect the current connection and establish a network access path corresponding to the target connection strategy;

[0017] According to the performance evaluation result of the current connection state, the temperature parameter of the probability distribution calculated in the normalization process is dynamically adjusted to dynamically balance exploration and utilization in the connection strategy selection, thereby optimizing the sampling behavior of the connection strategy in the next cycle.

[0018] In a second aspect, some embodiments of the present application further provide an electronic device, comprising:

[0019] One or more processors; and a memory storing computer program instructions, wherein when the computer program instructions are executed, the processor executes any one of the network connection optimization methods for a cellular communication module as described above.

[0020] On the third aspect, some embodiments of the present application also provide a computer-readable storage medium on which a computer program and / or instructions are stored, and when the computer program and / or instructions are executed by a processor, a network connection optimization method for a cellular communication module as described in any one of the above is implemented.

[0021] In a fourth aspect, some embodiments of the present application further provide a computer program product, comprising a computer program and / or instructions, which, when executed by a processor, implements a network connection optimization method for a cellular communication module as described in any one of the above.

[0022] Compared with related technologies, the solution provided in the embodiment of the present application quantitatively evaluates multiple candidate connection strategies by introducing a weighted scoring model and performing normalization processing to form a probability distribution. A temperature parameter is introduced as a control factor to adjust the discrete degree of sampling probability between each candidate connection strategy during the scoring normalization process, thereby achieving a dynamic balance between exploration and utilization in the connection strategy selection process. When the connection status is good, the system strengthens the sampling tendency of high-scoring strategies to improve connection stability. When the environment fluctuates or the signal degrades, the range of strategy selection is expanded to improve robustness and recovery capabilities, thereby significantly improving the connection stability, strategy switching flexibility and overall network access quality of the cellular communication module. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other implementation methods can be obtained based on these drawings without paying any creative work.

[0024] Figure 1 This is a flow chart of a method for optimizing network connection of a cellular communication module provided in an embodiment of the present application;

[0025] Figure 2 This is a system interaction flow chart of the cellular communication module network connection optimization method provided in an embodiment of the present application;

[0026] Figure 3 This is a flow chart of the adaptive parameter update control logic under continuous disconnection triggering provided by an embodiment of the present application;

[0027] Figure 4 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] To make the purpose, technical solutions, and advantages of the embodiments of this application more clear, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0029] First embodiment

[0030] The first embodiment of the present application relates to a network connection optimization method for a cellular communication module, referring to Figure 1 、 Figure 2 As shown, the method may include the following steps:

[0031] Step S1: Real-time collection of parameter information used to characterize the current network connection status, and feature processing of the parameter information to construct a multi-dimensional feature vector. The parameter information includes: base station cell switching frequency, operating frequency band type, received signal power, signal-to-noise ratio, operator network service quality index, and connection stability index.

[0032] Specifically, regarding step S1, the cellular communication module periodically collects key parameter information related to the network connection status through its open wireless parameter interface and performs feature engineering on the collected raw data. Feature engineering includes: uniformly mapping raw data with physical dimensions (such as dBm, dB, ms, etc.) to a standardized numerical range of 0-1 to facilitate unified scoring, and ultimately combining them to form a multidimensional feature vector for scoring decisions; mapping certain parameters using piecewise linear or sigmoid functions to enhance the model's ability to identify boundary intervals; introducing an exponential decay function for cell handover frequency to generate a mobility index; and applying a sliding average to the number of historical disconnections to form a stability coefficient.

[0033] Furthermore, in step S1, the step of performing feature processing on the parameter information to construct a multi-dimensional feature vector includes:

[0034] The original collected values ​​of the parameter information are processed by feature engineering and converted into standardized values ​​that can be used for subsequent scoring calculations. The standardized values ​​are combined into a multidimensional feature vector. The feature dimensions of the multidimensional feature vector can be expanded or tailored according to the business scenario. The standardized values ​​range from 0 to 1, indicating the performance of the candidate connection strategy in the parameter information dimension.

[0035] Parameter information includes the following six core features:

[0036] Base station cell switching frequency: refers to the ratio of the number of cell reselection or switching events that occur in the cellular communication module within a set time window to the duration, which is used to reflect the mobility or connection stability characteristics of the device in a specific network environment.

[0037] First, the system calculates the cell switching frequency per unit time (i.e., switching rate) and normalizes it so that the result is limited to the interval [0,1] to prevent extreme values ​​from interfering with the score. The normalized switching rate can be expressed as:

[0038]

[0039] Among them, switch_rate is the normalized switching rate; switch_count is the number of base station cell switches that occur in the cellular communication module within the set statistical period (i.e., window_size); window_size is the statistical time window of the switching behavior.

[0040] On this basis, to further enhance the model's sensitivity to high-frequency handoffs (high mobility), the system introduces an exponential decay function to construct a mobility index (mobility_index). This index gradually approaches 1 as the handoff rate increases, reflecting the instability of the connection state. The constructor of the mobility index is:

[0041]

[0042] Wherein, λ is an adjustable attenuation coefficient, which controls the response sensitivity of the function to high-frequency switching, and its preferred value is 0.1.

[0043] Operating frequency band type: refers to the wireless communication frequency band that the cellular communication module is currently connected to or can access, such as Band 3 and Band 8, which affects signal penetration and interference characteristics.

[0044] Specifically, the basic priority table can refer to the frequency band performance characteristics defined by 3GPP. The system pre-sets a frequency band basic priority table based on the capacity, interference, and coverage characteristics of each frequency band as specified in the 3GPP standard. For example: Band 3 (1800MHz): basic priority 0.8; Band 8 (900MHz): basic priority 0.6; Band 7 (2600MHz): basic priority 0.9. For each frequency band used by a candidate connection strategy, its basic priority value, base_priority, is searched from the priority table.

[0045] To improve the dynamics of priority expression, the system further adjusts the priority based on the module's historical connection success rate suc_rate (value range is 0-1) in the frequency band. The corrected frequency band adjustment factor adjusted_priority is calculated as follows:

[0046]

[0047] Received Signal Power (RSRP): refers to the received power measurement of the base station downlink signal in the current connection state of the cellular communication module, measured in dBm, and is used to evaluate the signal strength of the wireless link.

[0048] To effectively incorporate this parameter into the connection strategy scoring model, the system performs segmented normalization on the original RSRP data to enhance the ability to identify different signal intervals and suppress the interference of extremely weak signals on the scoring. The processing method is as follows:

[0049] When RSRP is greater than or equal to –85dBm, the module's current signal quality is good. The normalized result is directly set to 1.0, indicating the highest score for this dimension.

[0050] When RSRP is between –100dBm and –85dBm, a linear function is used to map the signal strength to the middle of the [0,1] interval to maintain the discrimination of the score within the medium intensity range. The normalization calculation method is:

[0051]

[0052] When RSRP is less than -100dBm, it indicates that the module is in a weak signal area. To prevent the scoring model from being overly sensitive to extremely low values, the system introduces a scaling factor for compression mapping to weaken the impact of the scoring. The processing method is:

[0053]

[0054] The processed received signal power normalized eigenvalue rsrp_norm is limited to the interval [0,1] and serves as a one-dimensional input in the normalized eigenvector to reflect the signal strength performance of the candidate connection strategy in the current environment.

[0055] Signal-to-noise ratio (SNR): refers to the ratio of the useful signal power to the background noise power measured at the receiving end of a cellular communication module. It is measured in dB and is an important parameter for evaluating wireless channel quality and anti-interference capabilities.

[0056] To incorporate this metric into the scoring model and improve its discrimination, the system linearizes the SNR and maps it to a normalized channel capacity value to quantify the performance of different connection strategies in terms of channel utilization efficiency. The specific processing method is as follows:

[0057] Convert the original dB value of SNR to a linear value using the following formula:

[0058]

[0059] Referring to the simplified form of Shannon's channel capacity formula, the theoretical communication capacity under unit bandwidth is calculated, and the maximum capacity is set as the normalized upper limit (preferably 6), and the normalized channel capacity is obtained:

[0060]

[0061] The obtained channel capacity value capacity is used as a one-dimensional input in the standardized feature vector to participate in the scoring calculation of candidate connection strategies, which is used to distinguish the advantages and disadvantages of different strategies in terms of anti-interference and channel resource availability.

[0062] Operator network service quality index: This index reflects the service performance of the operator (such as China Mobile, China Unicom, or China Telecom) to which the current cellular communication module is connected in the network environment. This index comprehensively considers two dimensions: communication latency and the operator's global quality evaluation. After feature engineering processing, it forms the two-dimensional input features of the scoring model.

[0063] The module regularly obtains the average round-trip delay (delay_ms) between the module and the core network through network layer commands (such as ping tests). To prevent the absolute value of the delay from amplifying linear errors in the scoring model, the system uses a sigmoid function to normalize it. The normalization formula is as follows:

[0064]

[0065] This function uses 100ms as the median reference point to ensure that low and medium latencies receive higher scores, and high latencies gradually compress scores to avoid extreme deviations.

[0066] The system obtains the autonomous system number (ASN) based on the SIM card's operator identification or IP address information, queries the preset operator network performance database, and returns the service quality evaluation score (asn_quality) of the current operator. The value range is [0,1]. The closer to 1, the more stable and reliable the operator's performance is in the current area or historical statistics.

[0067] The delay quality score delay_score and the network quality score asn_quality together constitute the operator service quality features used for policy scoring. The output is a two-dimensional standardized feature used to quantify the pros and cons of candidate connection strategies in terms of service layer performance.

[0068] Connection stability index: This refers to the master station connection stability feature, which reflects the continuity and reliability of the connection, constructed by the cellular communication module by collecting historical behavioral data during its connection to the master station. This feature is modeled from two aspects: a stability coefficient calculated based on disconnection events, and a jitter penalty term based on latency fluctuations.

[0069] Stability coefficient calculation (EWMA smoothing):

[0070] The system counts the number of offline_count disconnections of the cellular communication module in each cycle and uses the exponentially weighted moving average (EWMA) algorithm in combination with the stability score prev_stability of the previous cycle to construct a smoothed stability evaluation index with historical memory. The calculation formula is as follows:

[0071]

[0072] The discretization coefficient of 0.1 represents the update weight for the latest observation data. The maximum number of offline times is 5 times, which is the maximum deduction for the full score. This ensures that the score value is in the interval [0, 1]. The closer the value is to 1, the more stable the connection.

[0073] Delay jitter penalty calculation:

[0074] The system calculates the standard deviation of the RTT (round-trip time) between the module and the master station during the current cycle, rtt_deviation, to reflect the fluctuation of link delay during the connection process. To limit the scoring interference caused by extreme fluctuations, a normalized jitter penalty term is introduced. The calculation formula is:

[0075]

[0076] The higher the penalty value rtt_penalty is, the worse the link stability is, and it will eventually be used as a negative factor in the strategy scoring judgment.

[0077] The final output connection stability coefficient stability and penalty item value rtt_penalty together constitute the two-dimensional master station connection stability feature, which is used to evaluate the continuity and volatility of the connection, respectively, and assist in screening candidate strategies with better historical connection performance in the scoring model.

[0078] After completing feature engineering for each parameter, the system concatenates all normalized feature values ​​obtained after processing in a preset order to construct a multidimensional feature vector for candidate connection strategy scoring and decision making. In this embodiment, the feature vector contains eight dimensions, each corresponding to the following feature parameters: mobility index (mobility_index), frequency band adjustment factor (adjusted_priority), received signal power normalized feature value (rsrp_norm), channel capacity (capacity), delay score (delay_score), network quality score (asn_quality), connection stability coefficient (stability), and penalty value (rtt_penalty). These dimensional features are concatenated in a fixed order into a unified input vector structured as follows: feature_vector = [mobility_index, adjusted_priority, rsrp_norm, capacity, delay_score, asn_quality, stability, rtt_penalty].

[0079] It should be noted that the dimensions and content of the above-mentioned feature vectors can be expanded or tailored according to specific business scenarios to achieve stronger adaptability and model scalability.

[0080] Step S2: for the multiple candidate connection strategies that have been constructed, a score value of each candidate connection strategy is calculated based on the multi-dimensional feature vector and a preset weight value.

[0081] A weighted scoring mechanism is used to quantitatively evaluate the multiple candidate connection strategies constructed, combining the multidimensional feature vectors collected and processed under the current network environment. The score represents the strategy's overall adaptability and optimality under the current network conditions.

[0082] Furthermore, in step S2, multiple candidate connection strategies are constructed, specifically including:

[0083] Obtain the network access parameters that can be detected by the cellular communication module in the current network environment. The network access parameters include: cell ID, frequency band type, operator ID, and wireless access technology type;

[0084] Based on different combinations of network access parameters, multiple candidate connection strategies are constructed. The candidate connection strategies are used to represent the connection paths that the cellular communication module can use in the current network environment.

[0085] Specifically, the cellular communication module periodically scans and collects accessible network parameter information in the current network environment through the underlying communication protocol stack and commissioning interface. This parameter information includes at least: Cell ID: The unique identifier of the surrounding base station cell detected by the module; Band Type: The operating frequency band of each cell, such as Band 3 (1800MHz) and Band 8 (900MHz); Operator ID: The network operator to which the current cell belongs, such as China Mobile and China Unicom, which can be further associated with a SIM card or eSIM configuration to achieve multi-operator access; Radio Access Technology Type (RAT Type): Supported access standards, such as LTE, NR (5G), NB-IoT, GSM, etc. The above parameters can all be obtained through the wireless module's system information block decoding, signaling reports (such as measurement reports), frequency scanning, and other methods.

[0086] Base station selection dimension: such as connecting to CellA, CellB or CellC; frequency band combination dimension: such as Band3 single-frequency access, Band3+Band7 cross-frequency carrier aggregation; operator selection dimension: such as operator A corresponding to SIM1 and operator B corresponding to SIM2; RAT type dimension: such as 4G and 5G independent access, or non-independent networking (NSA) switching strategy.

[0087] Based on the various network access parameters collected above and feasible combination strategy rules, the system generates several groups of candidate connection strategies for the module to attempt. Each group of candidate connection strategies can be considered a specific network access path solution. For example, in a certain network environment, the module may detect the following optional access configurations: Strategy 1: Connect to CellA (Band 3, Carrier A); Strategy 2: Connect to CellB (Band 8, Carrier B); Strategy 3: Connect to CellA + CellC (Band 3 + Band 7 cross-band aggregation, Carrier A).

[0088] The system considers the three strategies mentioned above as candidate connection strategies for the current cycle and enters the scoring and sampling phase to evaluate the pros and cons of each strategy under the current network conditions. This candidate strategy set is dynamically updated based on the network environment, base station broadcasts, SIM configuration, and module mobility, ensuring a broad and adaptable strategy space and providing a rich basis for subsequent decision-making based on feature vector scoring.

[0089] Step S3: normalize each score value to obtain the probability distribution of the candidate connection strategies.

[0090] For step S3, specifically, it is assumed that there are n candidate connection strategies in the current cycle, corresponding to the score values ​​S1, S2, ..., S n, the system uses a normalization function to transform these scoring values ​​to obtain the sampling probability P of each strategy i , reflecting the possibility of being selected in the current environment.

[0091] Preferably, the system uses a Softmax function with adjustable discreteness for normalization, and its function form is as follows:

[0092]

[0093] Among them, P i is the sampling probability of the i-th candidate connection strategy; S i is the score value of the i-th candidate connection strategy; β is the temperature parameter, which is used to control the smoothness or deviation of the normalized distribution.

[0094] The temperature parameter β is a key adjustment coefficient in the Softmax function. Its value determines the probability distribution of the normalized score. The mechanism of β is as follows:

[0095] When the value of β approaches 0, the differences between scores are greatly amplified in the exponential function, and the normalized result tends to be average (entropy increases). That is, each strategy is assigned a relatively similar sampling probability, and the system exhibits uniform exploration behavior. This is suitable for use in scenarios where the connection environment is uncertain or the candidate strategy is not yet clear.

[0096] When the value of β approaches ∞, the score difference is gradually compressed during normalization, and the final probability distribution tends to concentrate on the strategy with the highest score. The system strategy selection behaves as a greedy choice, that is, it tends to strengthen the selection of the strategy with the current best score. This is suitable for scenarios with stable connection status or a clear target strategy.

[0097] Therefore, the temperature parameter β actually controls the entropy level and diversity of strategy sampling, achieving a dynamic balance between exploration and utilization in the strategy selection process.

[0098] To make the temperature regulation process controllable and stable, the system limits the value range of β to between 0.3 and 5.0, and sets a rate-of-change limit mechanism to prevent β from fluctuating too quickly and causing strategy shocks due to sudden changes in the environment or misjudgments.

[0099] Step S4: Based on the probability distribution, periodically perform probability sampling from multiple candidate connection strategies to determine the target connection strategy for the current period.

[0100] For step S4, in order to avoid single greedy behavior or path dependence in the strategy selection process, the system adopts a probability sampling mechanism and performs a sampling operation among multiple candidate connection strategies based on the probability distribution to determine the target connection strategy of the current cycle. Specifically, in the current evaluation cycle, the system samples the probability P1, P2, ..., P of each strategy according to the output of the Softmax function. n , through methods such as random number generation and cumulative distribution matching, a probability-driven sampling is performed to extract a strategy as the target connection strategy for the current cycle.

[0101] To ensure both controllability and stability in strategy selection, the system automatically triggers strategy sampling at a set interval (e.g., every 10 minutes) to adapt to dynamic changes in the network environment and improve sampling timeliness. Sampling is not based on a deterministic selection of the highest-scoring strategy, but rather on the strategy's normalized probability. This allows lower-scoring, but non-disadvantaged, strategies to be tried, maintaining the system's exploration capabilities.

[0102] Step S5: determining whether the target connection strategy is consistent with the current connection strategy of the cellular communication module; if not, controlling the cellular communication module to disconnect the current connection and establish a network access path corresponding to the target connection strategy.

[0103] For step S5, specifically, the system first determines the consistency between the target connection strategy determined in the current cycle and the actual connection strategy currently in which the cellular communication module is located, including a one-to-one comparison of fields such as whether the base station cell ID (Cell ID) is the same, whether the current frequency band combination (Band) is consistent, whether the operator ID connected has changed, and whether the wireless access technology type (RAT, such as LTE, NR, etc.) is consistent.

[0104] If the judgment result is consistent, that is, there is no difference between the target strategy and the current strategy, the system does not perform any switching operation, the current connection remains unchanged, and enters the next cycle evaluation process.

[0105] If the judgment result is inconsistent, it means that there is a deviation between the current connection status and the optimization result. The system will perform connection switching control operations, including:

[0106] Disconnect the current connection: Control the cellular communication module to safely disconnect the current network access path through the standard disconnection process (such as releasing the PDP context, deregistering the cell, etc.);

[0107] Execute reselection connection: Reconstruct network access parameters based on the target cell, frequency band, operator and access type specified in the target connection strategy, and complete the connection process such as network reselection, registration and activation;

[0108] State update and caching: After the switch is successful, the current connection state information is updated and the switch result is recorded for subsequent policy feedback and model updates.

[0109] The above judgment and switching mechanism can prevent the cellular communication module from frequently switching networks when the strategy score fluctuates slightly, thereby improving the execution stability of the strategy sampling behavior and the convergence speed of the connection path selection.

[0110] Step S6: Based on the performance evaluation results of the current connection state, dynamically adjust the temperature parameter of the probability distribution calculated in the normalization process to dynamically balance exploration and utilization in the connection strategy selection, thereby optimizing the sampling behavior of the connection strategy in the next cycle.

[0111] Specifically, regarding step S6, the system dynamically adjusts the temperature parameter used to generate the sampling probability distribution of candidate strategies during the normalization process based on the cellular communication module's connection status performance during the current cycle. This temperature parameter, acting as a control factor in the scoring normalization function, directly influences the degree of dispersion of the sampling probabilities for each candidate connection strategy, thereby impacting the bias and diversity of strategy selection. By adaptively adjusting the temperature parameter, the system dynamically balances exploration (expanding the range of candidate strategies) and exploitation (focusing on strategies with higher scores) in strategy selection, based on connection stability, signal quality, or strategy execution effectiveness in different network environments. This adjustment mechanism reduces unnecessary handoffs while maintaining flexible connection strategy response, thereby improving the effectiveness of strategy decisions.

[0112] In addition, the temperature parameter adjustment process is based on the key performance indicators of the cellular communication module during operation. It has dynamic response capabilities and environmental adaptability, and can continuously track the changing trends of connection quality and make adjustments, thereby effectively improving the judgment accuracy of policy switching and the overall network access stability of the module in a changing cellular network environment.

[0113] It is not difficult to find that compared with related technologies, the solution provided by the embodiment of the present application quantitatively evaluates multiple candidate connection strategies by introducing a weighted scoring model and performing normalization processing to form a probability distribution; the temperature parameter is introduced as a control factor to adjust the discrete degree of sampling probability between each candidate connection strategy during the scoring normalization process, thereby achieving a dynamic balance control between exploration and utilization in the connection strategy selection process. When the connection status is good, the system strengthens the sampling tendency of high-scoring strategies to improve connection stability; when the environment fluctuates or the signal degrades, the range of strategy selection is expanded to improve robustness and recovery capabilities, thereby significantly improving the connection stability, strategy switching flexibility and overall network access quality of the cellular communication module.

[0114] Second embodiment

[0115] The second embodiment of the present application relates to a network connection optimization method for a cellular communication module. The second embodiment is an improvement on the first embodiment. The specific improvement is that: in the second embodiment of the present application, a specific implementation method for calculating weighted scores for candidate connection strategies is provided, that is, step S2 can further include the following steps:

[0116] Step S201: For each candidate connection strategy, a set of weight values ​​is assigned based on its multidimensional feature vector related to the current network state. The weight values ​​correspond one-to-one with each standardized value in the multidimensional feature vector and are used to characterize the candidate connection strategy's sensitivity to or priority of each parameter information. The weight values ​​range from 0.01 to 1 to prevent weight failure or over-amplification.

[0117] Step S202: multiply each normalized value in the multidimensional feature vector by the corresponding weight value, and sum up all the product results to obtain a score value of the candidate connection strategy under the current network state.

[0118] For each candidate connection strategy constructed in the current cycle, the system assigns a set of weights based on its corresponding multidimensional feature vector. The multidimensional feature vector contains multiple characteristic components derived from normalized network state parameters, and each set of weights represents the candidate connection strategy's sensitivity or priority to each parameter dimension in the current network environment.

[0119] For example, each candidate strategy corresponds to a set of weight values , whose elements and eigenvectors One-to-one correspondence, where n represents the number of feature dimensions. Because in the scoring calculation, the score of each candidate connection strategy is obtained by multiplying multiple parameter eigenvalues ​​and corresponding weight values ​​and summing them up. If the weight of a feature is set to 0, then no matter what the eigenvalue is, the product will be 0, and this feature will not participate in the scoring at all, that is, the weight will be reset to zero and become invalid. On the contrary, if the weight of a feature is set very large, for example, much larger than other features, then after normalization or Softmax processing, this feature will have an excessively dominant position in the final score, that is, the weight is abnormally high, resulting in an imbalance in the score and excessive reliance on a single feature in the strategy selection, affecting the generalization ability and stability of the system. Therefore, each weight value is set to The value range of is set to between 0.01 and 1 to prevent the feature weight from being reset to zero or the weight from being abnormally high after normalization.

[0120] The above weight values ​​can be preset through static configuration. When the system is deployed, a fixed set of weight values ​​is set. For example, the signal strength is set to 0.8, indicating great concern; the frequency band type is set to 0.4, indicating general importance; and the connection stability is set to 0.6, indicating above average.

[0121] Reinforcement learning mechanisms, such as deep Q-networks, can also be used to automatically learn the importance weights of each feature under each strategy. Specifically, the system continuously tries different strategies (connecting to different network combinations) during actual operation. After each attempt, it observes the connection results, such as whether the connection is successful, whether there are any disconnections, and the speed. Based on the results of these attempts, it then assigns rewards and penalties to the weights. After repeated attempts over time, the system learns which features are more important and assigns corresponding weights. This allows for the long-term optimization of the emphasis each strategy places on different feature dimensions.

[0122] The system calculates the score of the candidate strategy by multiplying each component in the feature vector by its corresponding weight value and summing all the product results. That is:

[0123]

[0124] in, is the score of candidate connection strategy i under the current network status; is the kth component of the current eigenvector (normalized); is strategy i for feature The weight value of .

[0125] For the current strategy, if the network characteristics on a certain dimension (i.e. ) is good, and the importance weight of this dimension (i.e. ) is higher, the dimension will have a more positive impact on the final score; on the contrary, if a feature performs poorly but has a low weight, its impact on the score will be weakened, thus ensuring the balance, adjustability and interpretability of the score.

[0126] Combine the scoring structures of all candidate strategies into a scoring matrix , used for subsequent normalization and probability sampling operations. The range of the score value will be kept within a limited range due to the normalization interval of the input features (usually 0-1) and the weight constraint (such as 0.01-1.0), ensuring good stability during subsequent normalization.

[0127] It is not difficult to find that in the embodiment of the present application, by introducing a scoring calculation mechanism based on the weighted summation of feature vectors and weight vectors, this mechanism allows for configuring a set of weight values ​​corresponding to the multi-dimensional feature vectors for different candidate connection strategies, thereby achieving differentiated responses of the strategies to different network parameter dimensions. This fine-grained weight distribution method can more accurately reflect the performance differences of each strategy in different network environments, especially in scenarios where there are multiple access paths, large parameter fluctuations, or frequent dynamic changes in the environment. It can improve the resolution and stability of the score and avoid excessive averaging or bias in the strategy score.

[0128] Third embodiment

[0129] The third embodiment of the present application relates to a network connection optimization method for a cellular communication module. The third embodiment is an improvement on the first embodiment. Specifically, the improvement is as follows: In the third embodiment of the present application, a specific implementation method for dynamically adjusting temperature parameters based on the connection status evaluation result is provided, that is, step S6 can further include the following steps:

[0130] Step S601: In each preset evaluation period, the connection status indicators of the cellular communication module are counted, and the connection status indicators include the number of historical disconnections, the change rate of the comprehensive signal quality score, and the connection strategy switching success rate;

[0131] Step S602: Calculate a disconnection rate index based on the number of historical disconnections. The disconnection rate index is the normalized average of the number of historical disconnections within the evaluation period, and is used as a control factor in the calculation of the temperature parameter.

[0132] Step S603: setting the temperature parameter to the sum of a base value and a compensation term, wherein the base value is a fixed reference value and the compensation term increases as the disconnection rate indicator decreases;

[0133] Step S604: When it is detected that the rate of change of the comprehensive signal quality score has decreased by more than a preset percentage threshold compared to the previous evaluation cycle during the evaluation cycle, the temperature parameter is reduced by a preset first adjustment factor to make the sampling probability distribution after the score normalization more balanced, thereby increasing the probability of selecting a non-optimal connection strategy and enhancing the exploratory nature of the connection strategy;

[0134] Step S605: When it is detected that the connection strategy switching success rate exceeds the preset success rate threshold during the evaluation period, the temperature parameter is amplified according to the preset second adjustment factor to make the sampling probability distribution after the score normalization more concentrated, and the candidate connection strategies with high scores are sampled first to enhance the utilization of the connection strategy.

[0135] Specifically, within each preset policy evaluation cycle (for example, every 10 minutes), the system collects real-time statistics on multiple connection status indicators of the cellular communication module, including the number of historical disconnections, the rate of change of the comprehensive signal quality score, and the success rate of connection policy switching, etc., to reflect the module's connection stability and policy response effect in the current environment.

[0136] Based on the number of dropped calls recorded during a period, a normalized drop rate metric is calculated as an important indicator of connection stability. This metric is then used as a control factor in the calculation of the temperature parameter β. Specifically, the temperature parameter β is defined as the sum of a base value and a compensation term. The base value is a fixed reference value (e.g., 1.0), while the compensation term increases as the drop rate metric decreases. In other words, the fewer dropped calls and the more stable the network, the higher the temperature parameter, thereby strengthening the sampling tendency of the high-quality strategy.

[0137] If the system detects that the current cycle's overall signal quality score has decreased by more than a preset percentage threshold (e.g., 20%) compared to the previous cycle, it determines that the current network connection quality may be deteriorating or that the current policy model is insufficiently adaptable to the environment. To improve the diversity and resilience of policy sampling, the system reduces the temperature parameter β by a preset adjustment factor (e.g., multiplying it by 0.8). This flattens the probability distribution of the normalized policy scores, increases the probability of sampling suboptimal policies or unselected policies, and enhances the exploratory nature of policy selection.

[0138] When the strategy switching success rate for the current cycle exceeds a set threshold (e.g., 95%), it indicates that the current scoring model's judgment of strategy quality is highly consistent with actual execution results, and strategy sampling has a high hit rate. At this point, the system increases the temperature parameter β (for example, by multiplying it by 1.2) to further focus the normalized results on high-scoring strategies. This prioritizes sampling of candidate connection paths with better scores, thereby enhancing the usability of strategy selection and accelerating connection path convergence and performance stabilization.

[0139] It is not difficult to find that in the embodiment of the present application, by periodically collecting key performance indicators such as the number of disconnections, signal quality change trends, and strategy switching success rates of the cellular communication module during actual operation, it is possible to accurately perceive the effectiveness of the current connection strategy and the stability of the network environment, and dynamically adjust the value range and change direction of the temperature parameter in the score normalization process accordingly, thereby affecting the sampling probability distribution of candidate strategies. This makes the strategy selection process both flexible and goal-oriented, significantly improving the connection optimization efficiency and robustness of the cellular communication module in a dynamic network environment.

[0140] It should be noted that the third embodiment of the present application may also be an improvement based on any one or more of the first to second embodiments.

[0141] Fourth embodiment

[0142] The fourth embodiment of the present application relates to a network connection optimization method for a cellular communication module. The fourth embodiment is an improvement on the first embodiment. The specific improvement is that: in the fourth embodiment of the present application, a specific implementation method for adaptively adjusting the weight values ​​in the scoring model in combination with the continuous disconnection event judgment mechanism is provided, that is, the network connection optimization method further includes:

[0143] When a continuous disconnection event is detected, in addition to dynamically adjusting the temperature parameters, the weight values ​​used in the scoring calculation process are also adaptively adjusted. The conditions for determining a continuous disconnection event include any of the following conditions:

[0144] The number of disconnections within a set time period exceeds the preset threshold, or the number of failures of the cellular communication module to receive the base station heartbeat message exceeds the preset threshold within the set time window, or the received signal power is continuously lower than the preset threshold and the duration exceeds the set time.

[0145] In the event of a continuous disconnection event, the system performs an adaptive adjustment operation of the weight value, specifically including: increasing the weight value related to the operator network service quality indicator and / or reducing the weight value related to the working frequency band type;

[0146] If the event is not determined to be a continuous disconnection event, the system enters the regular incremental learning process to maintain the continuous learning ability of the scoring model of the candidate connection strategy.

[0147] Specifically, refer to Figure 3 As shown in the figure, if a feature becomes extremely unstable or ineffective within a short period of time—for example, due to long-term interference in a certain frequency band, local anomalies in the performance of a certain operator in a certain area, or a loss of differentiation in cell handover frequency—if the scoring model maintains its original high weight, it will over-rely on a judgment basis that is no longer effective, leading to incorrect policy selection. Therefore, when the system detects that a disconnection event constitutes a series of disconnections, in addition to dynamically adjusting the temperature parameter β based on the connection performance indicator, it also makes targeted adjustments to the feature weights used in the scoring model to correct the tendency of the scoring process to over-rely on unstable parameters.

[0148] The determination conditions for the continuous disconnection event include any of the following conditions:

[0149] The number of disconnections exceeds the preset threshold (e.g., more than 3 times) within a set time period (e.g., greater than 60 seconds and less than 20 minutes);

[0150] Within the set time window, the number of failures of the cellular communication module to receive the base station heartbeat message exceeds the preset failure threshold;

[0151] The module's received signal power (RSRP) remains below the preset threshold (for example, -110dBm) for a period exceeding the set duration (for example, 30 seconds).

[0152] Once a continuous disconnection event is identified, the system will implement an adaptive adjustment strategy for the weights in the scoring model. Specifically, this includes increasing the weights associated with the operator's network service quality indicators to increase the scoring model's sensitivity to the operator's stability performance; and decreasing the weight associated with the current operating frequency band type to weaken reliance on frequency band priority configuration and avoid imbalanced policy selection in high-interference or marginal areas. This achieves adaptive correction of the scoring mechanism, making it more tailored to the current network environment, thereby improving the accuracy of the next round of connection strategy evaluation and selection.

[0153] If no continuous disconnection events are triggered during the current evaluation cycle, that is, the overall connection status of the cellular communication module is within a stable range, the system will enter the regular incremental learning process. Specifically, based on the connection performance data collected in the current cycle (including strategy scores, actual connection results, success rates, etc.), the system fine-tunes the feature weights, sampling feedback, or historical values ​​of temperature parameters in the scoring model. This update adopts a progressive learning strategy, that is, only a limited proportion of new data is introduced each time to correct the existing model parameters, thereby effectively avoiding model oscillations or unstable strategy selection due to short-term fluctuations. In this way, without interrupting the normal operation of the strategy scoring mechanism, the evaluation data accumulated in the current cycle is used to perform lightweight parameter updates on the scoring model, so as to achieve gradual optimization of the model and continuous improvement of its environmental adaptability.

[0154] It is not difficult to find that in the embodiment of this application, compared with the method based solely on scoring values ​​and temperature parameter control strategy sampling, the response path to abnormal network behavior is further refined: when the system continuously detects typical connection failure modes such as disconnection, heartbeat loss, or long-term low signal, it can accurately identify and proactively adjust the emphasis on different parameter dimensions in the scoring logic, fundamentally guiding the scoring results to be closer to the actual connection performance. This further enhances the self-repair capability of the cellular communication module in the event of network connection anomalies and the targeted adjustment of the connection strategy.

[0155] It should be noted that the fourth embodiment of the present application may also be an improvement based on any one or more of the first to third embodiments.

[0156] Fifth embodiment

[0157] The fifth embodiment of the present application relates to a network connection optimization method for a cellular communication module. The fifth embodiment is an improvement on the first embodiment. The specific improvement is that: in the fifth embodiment of the present application, a specific implementation method based on the sampling range of the candidate priority queue restriction strategy is provided. That is, before step S4, the network connection optimization method further includes:

[0158] According to the normalized score value results, the candidate connection strategies with the highest score rankings are constructed into a candidate priority queue, and the probability sampling operation is limited to be performed only in the candidate priority queue. The candidate connection strategies included in the candidate priority queue are the candidate connection strategies with the top k score rankings, where k is an adjustable parameter.

[0159] Specifically, after performing normalization processing such as the Softmax function, the system obtains the scoring probability value corresponding to each candidate connection strategy. Sort all candidate strategies in descending order, and select the top k strategies with the highest scores to form a candidate priority queue. , where k is the set priority sampling window size, which can usually be set to 3~10 and can be flexibly adjusted depending on the total number of candidate strategies and the complexity of the network environment.

[0160] At this point, subsequent probability sampling steps will no longer operate based on the complete score probability distribution of all candidate strategies. Instead, the strategy scores in the priority queue are renormalized to form a restricted sampling distribution, ensuring that sampling occurs only within the range of high-scoring strategies. Specifically, the original score proportions of the strategies in the priority queue are retained; the sampling probabilities of low-scoring strategies (those not in the candidate priority queue) are forced to zero; and the normalized distribution is recalculated within the priority sampling range to satisfy the requirement that the sum of the Softmax probabilities must be 1. This approach can prevent missampling caused by noise in the probability tails, and is particularly advantageous in situations where there is significant score variation but the long tail effect of the Softmax function (even with large score differences, low-scoring items in the tail may have non-zero sampling probabilities) persists.

[0161] In addition, to improve adaptability, the system can introduce the following adjustable mechanisms, including:

[0162] Dynamic adjustment strategy of k value: When the score distribution of candidate strategies is too balanced or fluctuates greatly, the system can dynamically increase k to retain more alternative paths; when the high-scoring strategies differ significantly, k can be reduced to improve convergence efficiency.

[0163] Candidate priority queue refresh cycle: Based on the actual policy execution cycle (such as once every 10 minutes), the priority queue is refreshed periodically to ensure coverage of the latest scoring trends.

[0164] Mandatory retention mechanism: At least one strategy with good historical performance (such as the strategy with successful connection in the previous cycle) is retained in queue construction as redundancy to improve robustness.

[0165] It is not difficult to find that in the embodiment of the present application, by introducing the candidate priority queue mechanism, the system can effectively filter out invalid strategies with significantly low scores while maintaining strategy diversity, reduce the probability of missampling, improve the reliability of strategy switching judgment, shorten the sampling convergence time, and enhance the goal orientation and effectiveness of connection path selection.

[0166] It should be noted that the fifth embodiment of the present application may also be an improvement based on any one or more of the first to fourth embodiments.

[0167] The step division of the above various methods is only for the purpose of clear description. During implementation, they can be combined into one step or some steps can be split and decomposed into multiple steps. As long as they include the same logical relationship, they are all within the scope of protection of this application; adding insignificant modifications or introducing insignificant designs to the algorithm or process without changing the core design of the algorithm and process are all within the scope of protection of this application.

[0168] In addition, some embodiments of the present application further provide an electronic device. The electronic device may be various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, etc. The electronic device may also be various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices.

[0169] The electronic device includes: one or more processors; and a memory storing computer program instructions, and when the computer program instructions are executed, the processor executes a network connection optimization method for a cellular communication module provided in any one or more of the above embodiments. Figure 4An exemplary structural diagram of the electronic device is disclosed. The electronic device includes: one or more processors 1101, a memory 1102, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. The various components are connected to each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed within the electronic device, including instructions stored in or on the memory to display graphical information of a GUI on an external input / output device (such as a display device coupled to the interface). In some other embodiments, if necessary, multiple processors and / or multiple buses can be used with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, with each device providing some of the necessary operations. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present application described and / or required herein.

[0170] The electronic device may further include: an input device 1103 and an output device 1104. The processor 1101, the memory 1102, the input device 1103 and the output device 1104 may be connected via a bus or other means. Figure 4 The bus connection is taken as an example.

[0171] Input device 1103 can receive input digital or character information and generate key signal input related to user settings and function control of the electronic device. Examples include a touch screen, keypad, mouse, trackpad, touchpad, pointing stick, one or more mouse buttons, trackball, joystick, and other input devices. Output device 1104 may include a display device, auxiliary lighting devices (e.g., LEDs), and tactile feedback devices (e.g., vibration motors). The display device may include, but is not limited to, a liquid crystal display, a light emitting diode display, and a plasma display. In some embodiments, the display device may be a touch screen.

[0172] To provide user interaction, the electronic device may be a computer. The computer includes a display device (e.g., a cathode ray tube or LCD monitor) for displaying information to the user, and a keyboard and pointing device (e.g., a mouse) through which the user can provide input to the computer. Other types of devices may also be used to provide user interaction; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback), and input from the user may be received in any form (e.g., voice input or tactile input).

[0173] In an embodiment of the present application, a computer program / instruction is stored on a computer-readable medium. When executed by a processor, the computer program / instruction implements a network connection optimization method for a cellular communication module provided in any one or more of the above-described embodiments. The computer-readable medium may be included in the electronic device described in the above-described embodiments, or may exist independently and not be incorporated into the device. The computer-readable medium carries one or more computer-readable instructions.

[0174] The memory 1102 can be used as a non-transitory computer-readable storage medium to store non-transitory software programs, non-transitory computer executable programs, and modules. The processor 1101 executes the non-transitory software programs, instructions, and modules stored in the memory 1102 to execute various functional applications and data processing of the server, thereby implementing the program instructions / modules corresponding to the method provided in any one or more of the above embodiments of the present application.

[0175] The memory 1102 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and applications required for at least one function; the data storage area may store data created based on the use of the electronic device, etc. In addition, the memory 1102 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some embodiments, the memory 1102 may optionally include a memory remotely located relative to the processor 1101, and these remote memories may be connected to the electronic device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0176] It should be noted that the computer-readable medium described in this application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above. Computer-readable media may be, for example, but not limited to: electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or components, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory, a read-only memory, an erasable programmable read-only memory, an optical fiber, a portable compact disk read-only memory, an optical storage device, a magnetic storage device, or any suitable combination of the above. In this application, a computer-readable medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or device.

[0177] Computer-readable media includes both permanent and non-permanent, removable and non-removable media, and can be implemented using any method or technology for information storage. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory, static random access memory, dynamic random access memory, other types of random access memory, read-only memory, electrically erasable programmable read-only memory, flash memory or other memory technology, compact discs, digital versatile discs or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0178] Computer program code for performing the operations of the present application may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as C or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network or a wide area network, or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0179] In the above embodiments, all or part of the steps or functions of the present invention may be implemented using software, hardware, firmware, or any combination thereof. For example, implementation may be achieved using a dedicated integrated circuit, a general-purpose computer, or any other similar hardware device. In some embodiments, the software program of the present application may be executed by a processor to implement the above steps or functions. Similarly, the software program of the present application (including related data structures) may be stored in a computer-readable recording medium, such as a RAM memory, a magnetic or optical drive, a floppy disk, or the like. In addition, some steps or functions of the present application may be implemented using hardware, for example, as a circuit that cooperates with a processor to perform the various steps or functions.

[0180] The computer program product provided in the embodiments of the present application includes one or more computer programs / instructions that, when executed by a processor, fully or partially produce the processes or functions described in accordance with the embodiments of the present application. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via a wired (e.g., coaxial cable, optical fiber, digital subscriber line) or wireless (e.g., infrared, wireless, microwave, etc.) method. The computer-readable storage medium may be any available medium that can be accessed by a computer or a data storage device such as a server or data center that includes one or more available media. The available medium may be a magnetic medium (e.g., a floppy disk, a hard disk, a magnetic tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state drive).

[0181] The flowcharts or block diagrams in the accompanying drawings illustrate the possible architectures, functions and operations of the devices, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or part of code, and the module, program segment or part of code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, as well as the combination of boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-specific system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0182] The scope of this application is defined by the appended claims rather than the foregoing description and is therefore intended to encompass within this application all changes that come within the meaning and range of equivalents of the claims. Any reference signs in the claims should not be construed as limiting the claims to which they relate. In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in a device claim may also be implemented by one unit or device through software or hardware. Words such as "first" and "second" are only used to distinguish the description and do not indicate any particular order, nor should they be understood as indicating or implying relative importance.

[0183] The above descriptions are merely specific embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any person skilled in the art may easily propose variations or substitutions within the technical scope disclosed in the present application, and such variations or substitutions shall be encompassed within the scope of protection of the present application. Therefore, the scope of protection of the present application shall be subject to the scope of protection of the claims, and the above descriptions shall be regarded as exemplary and non-limiting.

Claims

1. A network connection optimization method for a cellular communication module, characterized in that: include: Collect parameter information used to characterize the current network connection status in real time, perform feature processing on the parameter information, and construct Multidimensional feature vector, the parameter information includes: base station cell switching frequency, operating frequency band type, received signal power, signal-to-noise ratio, operator network service quality index, connection stability index; For the plurality of candidate connection strategies that have been constructed, respectively calculating a score value of each candidate connection strategy based on the multidimensional feature vector and a preset weight value; Build multiple candidate connection strategies, including: Obtain network access parameters detectable by the cellular communication module in the current network environment, the network access parameters including: cell identifier, frequency band type, operator identifier, and wireless access technology type; constructing a plurality of candidate connection strategies based on different combinations of the network access parameters, wherein the candidate connection strategies are used to represent connection paths that can be selected by the cellular communication module in the current network environment; Normalizing each of the score values ​​to obtain a probability distribution of the candidate connection strategies; According to the probability distribution, periodically perform probability sampling from the plurality of candidate connection strategies to determine a target connection strategy for the current period; determining whether the target connection strategy is consistent with the current connection strategy of the cellular communication module; if not, controlling the cellular communication module to disconnect the current connection and establish a network access path corresponding to the target connection strategy; Dynamically adjust the temperature parameter of the probability distribution calculated in the normalization process based on the performance evaluation result of the current connection state to dynamically balance exploration and utilization in the connection strategy selection, thereby optimizing the sampling behavior of the connection strategy in the next cycle; In each preset evaluation period, the connection status indicator of the cellular communication module is counted, and the connection status indicator includes the number of historical disconnections; Calculating a disconnection rate index based on the number of historical disconnections. The disconnection rate index is a normalized average of the number of historical disconnections within the evaluation period, and is used as a control factor in the calculation of the temperature parameter. The temperature parameter is set as the sum of a basic value and a compensation item, wherein the basic value is a fixed reference value, and the compensation item increases as the offline rate indicator decreases.

2. The network connection optimization method for a cellular communication module according to claim 1, characterized in that: The step of performing feature processing on the parameter information to construct a multi-dimensional feature vector includes: The original collected values ​​of the parameter information are subjected to feature engineering processing and converted into standardized values ​​that can be used for subsequent scoring calculations. The standardized values ​​are combined into a multidimensional feature vector. The feature dimensions of the multidimensional feature vector can be expanded or tailored according to the business scenario, wherein the value range of the standardized value is 0 to 1, indicating the performance of the candidate connection strategy in the parameter information dimension.

3. The network connection optimization method according to claim 2, characterized in that: The step of calculating the score value of each of the constructed candidate connection strategies based on the multidimensional feature vector and a preset weight value includes: For each candidate connection strategy, a set of weight values ​​is configured based on its multidimensional feature vector related to the current network state. The set of weight values ​​corresponds one-to-one to each standardized value in the multidimensional feature vector and is used to characterize the sensitivity or priority of the candidate connection strategy to each parameter information. The weight values ​​range from 0.01 to 1 to prevent weight failure or excessive amplification. Each normalized value in the multidimensional feature vector is multiplied by the corresponding weight value, and all the product results are summed up to obtain the score value of the candidate connection strategy under the current network state.

4. The network connection optimization method for a cellular communication module according to claim 1, characterized in that: The step of dynamically adjusting the temperature parameter of the probability distribution calculated in the normalization process according to the performance evaluation result of the current connection state to dynamically balance exploration and utilization in connection strategy selection includes: The connection status indicators also include the signal quality comprehensive score change rate and the connection strategy switching success rate; When it is detected within the evaluation period that the rate of change of the comprehensive signal quality score decreases by more than a preset percentage threshold compared to the previous evaluation period, the temperature parameter is reduced by a preset first adjustment factor to make the sampling probability distribution after the score normalization tend to be balanced, thereby increasing the probability of selecting a non-optimal connection strategy and enhancing the exploratory nature of the connection strategy; When it is detected within the evaluation period that the connection strategy switching success rate exceeds a preset success rate threshold, the temperature parameter is amplified according to a preset second adjustment factor so that the sampling probability distribution after score normalization is more concentrated, and candidate connection strategies with high scores are sampled first to enhance the utilization of the connection strategy.

5. The network connection optimization method for a cellular communication module according to claim 1 or 3, characterized in that: Also includes: When the offline event is detected as a continuous offline event, in addition to dynamically adjusting the temperature parameter, the weight value used in the scoring calculation process is also adaptively adjusted. The determination conditions for the continuous offline event include any of the following conditions being met: The number of disconnections within a set time period exceeds a preset threshold, or the number of failures of the cellular communication module to receive base station heartbeat messages exceeds a preset threshold within a set time window, or the received signal power is continuously lower than a preset threshold and the duration exceeds a set time. In the case of determining that it is the continuous disconnection event, the system performs an adaptive adjustment operation of the weight value, specifically including: increasing the weight value related to the operator network service quality indicator and / or reducing the weight value related to the working frequency band type; In the case where the event is not determined to be the continuous disconnection event, the system enters a conventional incremental learning process to maintain the continuous learning capability of the scoring model of the candidate connection strategy.

6. The network connection optimization method for a cellular communication module according to claim 1, characterized in that: Before the step of periodically performing probability sampling from the plurality of candidate connection strategies according to the probability distribution, the method further includes: According to the normalized results of the score values, the candidate connection strategies with higher score rankings are constructed into a candidate priority queue, and the probability sampling operation is limited to be performed only in the candidate priority queue, wherein the candidate connection strategies included in the candidate priority queue are the candidate connection strategies with the top k score rankings, where k is an adjustable parameter.

7. An electronic device, characterized in that: The electronic device comprises: One or more processors; and a memory storing computer program instructions, wherein the computer program instructions, when executed, cause the processor to execute the network connection optimization method for a cellular communication module according to any one of claims 1 to 6.

8. A computer-readable storage medium having a computer program and / or instructions stored thereon, characterized in that: When the computer program and / or instructions are executed by a processor, the network connection optimization method for a cellular communication module according to any one of claims 1 to 6 is implemented.

9. A computer program product comprising a computer program and / or instructions, characterized in that When the computer program and / or instruction is executed by a processor, the network connection optimization method for a cellular communication module according to any one of claims 1 to 6 is implemented.

Images