Intelligent Internet of Things communication method

Abstract

The invention relates to the technical field of Internet of Things communication, and particularly discloses an intelligent Internet of Things communication method, which comprises the following steps of: calculating a network state score value of each interactive network by monitoring state parameters such as received signal strength, a signal-to-noise ratio and a packet loss rate of various available interactive networks in real time so as to quantify real-time performance of different networks; and generating a candidate network list based on the dynamic ranking of the score values. And when it is detected that the state score of the current interactive network is lower than that of the optimal candidate network, a standby communication channel is established on the target optimal network in advance, and seamless transition is realized through parallel transmission of the main and standby channels. Through the mode, when the current network quality is reduced or a better network option appears, the communication strategy can be adjusted in time and intelligently, and the standby channel is established in advance to synchronously transmit the data packet to realize smooth transition, so that the smoothness and the stability of man-machine interaction are ensured, and user experience reduction or interaction interruption caused by network problems is avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of Internet of Things communication, and more particularly, to an intelligent Internet of Things communication method. BACKGROUND

[0002] With the rapid development and wide application of Internet of Things (IoT) technology, various intelligent devices are increasingly integrated into people's daily life and work, and the convenience and reliability of human-computer interaction have become the key to improving user experience. At present, in the field of Internet of Things (IoT), Internet of Things devices often rely on only one network technology (such as Ethernet, Wi-Fi, Bluetooth, 4G / 5G, ZigBee, etc.) to realize data transmission and human-computer interaction when communicating. Due to the simplicity, low cost and easy deployment of single network solutions, they are widely used in many scenarios. However, due to the stability and quality of different types of communication networks being affected by various factors such as signal interference, device movement, network congestion, etc., it is difficult for a single fixed communication method to ensure that it can provide a continuous and smooth human-computer interaction experience in all situations. Network fluctuations or interruptions often lead to interaction delays, data loss or even service unavailability, seriously affecting user experience.

[0003] At present, although some devices may provide a simple manual switching function or support automatic attempts to connect to a backup network when the current network is unavailable, to a certain extent, it solves the problem of single network failure, but in the face of complex network environments, there are significant limitations. For example, when the preset network signal weakens but has not completely interrupted, the device may still adhere to the low-quality network, resulting in low communication efficiency; or when switching networks, there is a lack of comprehensive assessment of the real-time quality of each available network according to the fixed priority order, which may switch to a network that is not the best, or even cause data interruption or interaction lag during the switching process. That is, the existing technology cannot dynamically select the optimal link according to real-time network conditions, making it difficult to meet users' demand for high-quality and uninterrupted human-computer interaction, especially in scenarios where multiple network quality fluctuates dynamically due to device movement or environmental changes.

[0004] Therefore, an optimized intelligent Internet of Things communication method is expected. SUMMARY

[0005] To solve the above technical problems, the present application is proposed. Embodiments of the present application provide an intelligent Internet of Things communication method, which calculates network state score values of various interactive networks by real-time monitoring of state parameters such as received signal strength, signal-to-noise ratio and packet loss rate of various available interactive networks, quantifies real-time performance of different networks, and generates a candidate network list based on dynamic ranking of the score values. When the state score of the current interactive network is detected to be lower than the optimal candidate network, a backup communication channel is established on the target optimal network in advance, and seamless transition is achieved through parallel transmission of the primary and backup channels. In this way, the communication strategy can be adjusted in a timely and intelligent manner when the current network quality decreases or a better network option appears, and smooth transition is achieved through synchronous transmission of data packets through the pre-established backup channel, thereby ensuring the fluency and stability of human-computer interaction and avoiding user experience degradation or interaction interruption caused by network problems.

[0006] According to one aspect of the present application, an intelligent Internet of Things communication method is provided, which comprises: Step 1: Real-time monitoring of network state data of various interactive networks; Step 2: Calculating network state score values of various interactive networks based on network state data; Step 3: Ranking the various interactive networks based on the network state score values to obtain a candidate network list; Step 4: Selecting the optimal interactive network from the candidate network list for human-computer interaction; Step 5: When the current interactive network is not optimal, actively switching to the optimal interactive network in the candidate network list.

[0007] Compared with the prior art, the intelligent Internet of Things communication method provided by the present application calculates network state score values of various interactive networks by real-time monitoring of state parameters such as received signal strength, signal-to-noise ratio and packet loss rate of various available interactive networks, quantifies real-time performance of different networks, and generates a candidate network list based on dynamic ranking of the score values. When the state score of the current interactive network is detected to be lower than the optimal candidate network, a backup communication channel is established on the target optimal network in advance, and seamless transition is achieved through parallel transmission of the primary and backup channels. In this way, the communication strategy can be adjusted in a timely and intelligent manner when the current network quality decreases or a better network option appears, and smooth transition is achieved through synchronous transmission of data packets through the pre-established backup channel, thereby ensuring the fluency and stability of human-computer interaction and avoiding user experience degradation or interaction interruption caused by network problems. BRIEF DESCRIPTION OF DRAWINGS

[0008] The above and other objects, features and advantages of the present application will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings in which:

[0009] Figure 1 Flowchart of the smart Internet of things communication method according to an embodiment of the present application.

[0010] Figure 2 Data flow diagram of the smart Internet of things communication method according to an embodiment of the present application.

[0011] Figure 3 Flowchart of sub-step S5 of the smart Internet of things communication method according to an embodiment of the present application.

[0012] Figure 4 Flowchart of sub-step S2 of the smart Internet of things communication method according to an embodiment of the present application.

[0013] Figure 5 Flowchart of sub-step S22 of the smart Internet of things communication method according to an embodiment of the present application.

[0014] Figure 6 Flowchart of sub-step S222 of the smart Internet of things communication method according to an embodiment of the present application.

[0015] Figure 7 Flowchart of sub-step S2222 of the smart Internet of things communication method according to an embodiment of the present application. DETAILED DESCRIPTION

[0016] As used in the present application and claims, the articles "a", "an", and "the" are not limited to refer to only one of a possible plurality of elements, but can be used in reference to one or more of a possible plurality of elements. As used in the present application and claims, the term "plurality" refers to two or more. As used in the present application and claims, the term "comprising" is not intended to be limiting, and is used in the sense that it is open-ended, and includes the possibility of additional elements, steps, and / or components.

[0017] Although the present application makes various references to certain modules in the system according to embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are illustrative only, and different aspects of the system and method can use different modules.

[0018] Flowcharts are used in this application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the foregoing or following operations are not necessarily performed in sequence. Instead, various steps can be processed in reverse order or simultaneously, as needed. Meanwhile, other operations can be added to these processes, or one or more steps of operations can be removed from these processes.

[0019] In the following, the example embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and not all embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein.

[0020] Embodiment 1 To solve the technical problems described in the above background art, the present application proposes an intelligent Internet of Things communication method, which calculates the network state score value of each interactive network by real-time monitoring of state parameters such as received signal strength, signal-to-noise ratio and packet loss rate of multiple available interactive networks, quantifies the real-time performance of different networks, and generates a candidate network list based on the dynamic ranking of the score value. When the state score of the current interactive network is lower than the optimal candidate network, a backup communication channel is established on the target optimal network in advance, and seamless transition is achieved through parallel transmission of the primary and backup channels. In this way, the communication strategy can be adjusted in time and intelligently when the current network quality decreases or a better network option appears, and the smoothness and stability of human-computer interaction are ensured through the pre-established backup channel for synchronous transmission of data packets to achieve smooth transition, thereby avoiding the decline of user experience or interruption of interaction caused by network problems.

[0021] Figure 1 A flowchart of the intelligent Internet of Things communication method according to the embodiments of the present application. Figure 2 A data flowchart of the intelligent Internet of Things communication method according to the embodiments of the present application. As shown in Figure 1 and Figure 2 The intelligent Internet of Things communication method comprises the steps of: S1, real-time monitoring of network state data of multiple interactive networks; S2, calculating the network state score value of each interactive network based on the network state data; S3, ranking the multiple interactive networks based on the network state score value to obtain a candidate network list; S4, selecting the optimal interactive network from the candidate network list for human-computer interaction; S5, actively switching to the optimal interactive network in the candidate network list when the current interactive network is not optimal.

[0022] In the above intelligent Internet of Things communication method, the step S1, the network state data of a plurality of interactive networks is monitored in real time. It should be understood that, considering the complex and changeable network environment of Internet of Things devices, a single network cannot guarantee continuous and stable communication quality, and the prior art often relies on fixed priorities when performing network switching, lacks real-time perception ability of a plurality of available network states, and cannot meet the dynamic switching demand. Therefore, in order to comprehensively grasp the real-time performance of each available network and provide a data basis for subsequent decision-making, based on the multi-dimensional network state perception principle, the dynamic parameters of a plurality of interactive networks are collected in parallel to realize real-time monitoring of network quality. In one specific example of the present application, the network state data includes received signal strength value, signal-to-noise ratio and packet loss rate. Among them, the received signal strength value is an important indicator to measure the strength of network signal, which directly reflects the attenuation in the signal transmission process. The signal-to-noise ratio reflects the relative strength of the signal and the noise, which is crucial for evaluating the communication quality. The packet loss rate reveals the loss proportion of data packets in the network transmission process, which is a key parameter to measure network stability and reliability. By activating all network interfaces (such as Ethernet, Wi-Fi, 4G / 5G, Bluetooth, etc.) in parallel when the device starts, the received signal strength (RSSI), signal-to-noise ratio (SNR) and packet loss rate of each interactive network are measured periodically, which can continuously perceive the instantaneous state change of each network and provide real-time data support for subsequent network switching decision.

[0023] Specifically, at the physical level, each network technology has its unique signal characteristics and transmission method, so the way to obtain the network state data of different network types is also different. For example, in the Wi-Fi network, the received signal strength (RSSI), signal-to-noise ratio (SNR) and link quality information can be read through the underlying driver interface provided by the wireless network card chip; while in the cellular network (such as 4G / 5G), the current base station connection strength, bit error rate and data throughput parameters can be obtained through the state register or AT instruction set built-in mobile communication module. In addition, the Bluetooth network can obtain connection quality feedback through the link manager in the Bluetooth protocol stack, and the wired network (such as Ethernet) can obtain bandwidth utilization, error frame number and other key indicators through switch port statistics information. Although the data sources and acquisition methods of various networks are not the same, the ultimate goal is to extract the core parameters that can reflect the current network operating conditions, and then send them to the subsequent scoring calculation module in a unified format.

[0024] In the implementation process, in order to ensure the timeliness and representativeness of the collected data, the system needs to set a reasonable sampling period. This period should be differentiated according to the stability characteristics of various networks. For example, for Wi-Fi or cellular networks with frequent signal fluctuations, a shorter sampling interval (such as 100 milliseconds) can be set to capture network quality trends in a timely manner; for relatively stable wired networks, the sampling time can be appropriately extended (such as 1 second) to reduce resource consumption. At the same time, to avoid misjudgment due to transient interference, the system also needs to introduce a sliding window mechanism to perform weighted average processing on historical data collected within a certain time range, thereby obtaining a more accurate network state representation. This mechanism not only helps to filter out sudden noise interference, but also reflects the trend of network quality changes.

[0025] In addition, in the process of collecting network status data, in addition to focusing on signal strength and error code conditions, the bandwidth occupation and actual data transmission rate of each network need to be recorded synchronously. These parameters reflect the current network load state and available resource capacity, and are important basis for determining whether it is suitable for carrying high-priority interactive tasks. For example, a Wi-Fi network with good signal strength may have a much lower actual transmission efficiency than expected if it is in a high-concurrency access state. At this time, even if its physical layer indicators perform well, it is not suitable for use as the main communication link. Therefore, the system needs to combine the statistical interface provided by the traffic monitoring tool or network protocol stack to regularly collect data such as data transmission rate, queue length, and packet loss retransmission times of each network interface, and use these data as important input dimensions for network status scoring.

[0026] Considering that Internet of Things devices usually have low computing power and limited storage resources, how to efficiently process massive network status data is also a key problem that needs to be considered in the implementation process. Therefore, the system can use an edge computing architecture to complete preliminary cleaning, compression, and feature extraction of data on the local embedded platform, and only upload key indicators to the decision module, thereby reducing the resource overhead of the overall system. For example, a lightweight data preprocessing unit can be deployed at each network interface to perform filtering and denoising, outlier removal, and unit standardization on raw signal strength, signal-to-noise ratio, and other data, and then the processed results are aggregated to the central controller. This approach not only improves the real-time performance of data processing, but also reduces the burden on the main control unit, making the entire monitoring process more efficient and stable.

[0027] During the whole monitoring process, the system also needs to have certain fault tolerance. When a network interface fails or is temporarily unavailable, the system should be able to automatically identify and skip data collection for that network, while recording the corresponding abnormal event log for subsequent fault diagnosis and recovery processing. In addition, to deal with sudden network interruption, the system can also design a heartbeat detection mechanism to periodically send probe packets to each network to confirm whether it is still available. Once it is found that a network has been unresponsive for a long time, it can be temporarily removed from the candidate list and notify the user or the cloud platform for further processing.

[0028] In the above intelligent Internet of Things communication method, the step S2, the network state score value of each interactive network is calculated based on the network state data. In a specific example of the present application, the network state data is input into a weighted evaluation model to obtain the network state score value, wherein the weight value of the received signal strength value is 0.3, the weight value of the signal-to-noise ratio is 0.5, and the weight value of the packet loss rate is 0.2. That is, in order to convert the original multi-dimensional and different dimension network state data into a unified and quantifiable index for comparison, so as to objectively evaluate the comprehensive real-time communication quality of each interactive network, the present application is based on the principle of weighted evaluation model in multi-attribute decision theory, and it is considered that different network parameters have different contribution degrees to the overall communication performance. By assigning a preset weight to each key parameter, the comprehensive performance of the network can be more accurately reflected. Specifically, the present application first normalizes the received signal strength value, the signal-to-noise ratio and the packet loss rate to convert them into dimensionless values, and then uses a linear weighting formula: network score = 0.3 x RSSI + 0.5 x SNR + 0.2 x (1-packet loss rate). Wherein, the weight distribution (0.3, 0.5, 0.2) reflects the dominant role of signal-to-noise ratio on communication stability, while considering the balance of signal strength and data integrity.

[0029] In the above intelligent Internet of Things communication method, the step S3, the multiple interactive networks are sorted based on the network state score value to obtain a candidate network list. It can be understood that due to the dynamic change of network environment (such as signal attenuation caused by device movement or temporary interference source), a fixed priority strategy may select a non-optimal network. Therefore, in order to ensure the real-time adaptability of the candidate network, the present application is based on the principle of dynamic priority adjustment, and the network state score value of each interactive network is updated periodically to realize dynamic optimization of the interactive network. For example, the score value of each network is updated every 5 seconds, and the candidate network list is generated in descending order of score (such as Ethernet > 5G > Wi-Fi > Bluetooth).

[0030] In the above intelligent Internet of Things communication method, the step S4 is to select the optimal interactive network from the candidate network list for human-computer interaction. Specifically, since the user has a very high requirement for interaction continuity (such as remote control or real-time video transmission), in order to quickly lock the best communication path in a complex environment, the application is based on the optimal path selection principle, through real-time updating of the candidate list, selects the interactive network with the highest network state score value as the working network for human-computer interaction, to ensure that data transmission can be performed through the optimal link at any time.

[0031] In the above intelligent Internet of Things communication method, the step S5 is to actively switch to the optimal interactive network in the candidate network list when the current interactive network is not optimal. That is, when it is detected that the current interactive network is inconsistent with the candidate interactive network at the top of the candidate network list, it means that there is a better communication option, at this time, the network switching mechanism is actively triggered, and the communication link is smoothly migrated to the optimal candidate interactive network, to realize seamless network transition. Wherein, Figure 3 The flowchart of the sub-step S5 of the intelligent Internet of Things communication method according to the embodiment of the application. As shown in the figure, Figure 3 The step S5 includes the steps of: S51, setting the current interactive network as a first interactive network and defining the optimal interactive network in the candidate network list as a second interactive network; S52, before deciding to switch to the second interactive network, pre-establishing a backup communication channel through the second interactive network and keeping the main communication channel and the backup communication channel simultaneously accessing the server for a short time; S53, during the switching to the second interactive network, ensuring that the heartbeat mechanism of the backup communication channel is normal and synchronizing and processing the data packets of the main communication channel and the backup communication channel.

[0032] Specifically, the step S51 sets the current interactive network as a first interactive network and defines the optimal interactive network in the candidate network list as a second interactive network. In one preferred embodiment of the application, in order to avoid communication instability caused by frequent switching of the interactive network, a "switching hysteresis interval" is introduced to prevent frequent switching caused by small score fluctuations. For example, when the network state score value of the current network (i.e. the first interactive network) is lower than the network state score value of the optimal candidate network (i.e. the second interactive network) by more than a preset threshold, the switching action is triggered, thereby avoiding unnecessary network jitter and further improving user experience.

[0033] Specifically, the step S52, before deciding to switch to the second interactive network, a standby communication channel is pre-established through the second interactive network and the main communication channel and the standby communication channel are kept accessing the server at the same time for a short time. It should be understood that, since the conventional switching relies on the single-channel interruption and then reconstruction (such as disconnecting Wi-Fi and then connecting 5G), it inevitably leads to a short communication interruption. Therefore, in order to eliminate the service suspension phenomenon in the switching process, the application pre-establishes a standby communication channel on the optimal candidate network (i.e. the second interactive network) before switching and maintains the coexistence of the double channels, based on the parallel transmission and state synchronization of the main and standby channels, to realize seamless transition. Specifically, after determining to switch to the second interactive network, a standby TCP session (such as the Socket connection of the 5G network) is immediately established through the protocol stack, and the session state (including identity authentication information, heartbeat packet timestamp, data cache, etc.) of the main channel is synchronized. For example, the main channel (Wi-Fi) continuously transmits real-time video stream, the standby channel (5G) synchronously receives the copy of the same data packet, and the server records the data sequence number of the double channels. In this way, the standby channel already has complete communication capability, ensuring that there is no need to re-hands or identity authentication when switching, and significantly reducing the switching delay.

[0034] Specifically, the step S53 ensures that the heartbeat mechanism of the standby communication channel is normal and the data packets of the main communication channel and the standby communication channel are synchronized and processed during the switching to the second interactive network. In a specific example of the present application, the synchronization and processing of the data packets of the main communication channel and the standby communication channel include: marking the data packets of the main communication channel and the standby communication channel with a sequence number to obtain the priority of the data packets; and processing the data packets in the order of the priority of the data packets. It should be understood that the stability of the standby channel directly affects the success rate of switching (for example, network jitter may cause the standby channel to fail), and parallel double channels may cause data packets to be out of order or repeated. Therefore, in order to guarantee the continuity of communication and the integrity of data after switching, the present application realizes zero-aware switching by verifying the availability of the standby channel in real time and strictly managing the order of data flow based on the heartbeat monitoring and data sequence control principle. Specifically, during the switching preparation phase, periodic heartbeat detection (such as sending an ICMP request once per second) is performed on the standby channel, and if the heartbeat is timed out for three consecutive times, the switching is abandoned and a candidate network is reselected. During the switching process, in order to realize the synchronization and processing of the data packets of the main communication channel and the standby communication channel, a "sequence marking" technology is used to add a unique sequence number and a timestamp (such as using an incremental UUID + nanosecond timestamp) to each data packet, and the server processes the data according to the marking order to ensure that the transmission order of the data packets in the double channels is consistent, avoiding the risk of data interruption and out-of-order, and when the data packets are reassembled at the receiving end, the data can be restored in the correct order. At the same time, by comparing the timestamps of the data packets, repeated data packets can be detected and discarded, ensuring the uniqueness and integrity of the data. After the standby channel is stably established, the communication traffic is smoothly switched to the second interactive network, and the original main communication channel is disconnected. In this way, intelligent, active and smooth network switching can be realized, ensuring that the device communicates in the optimal network condition, effectively avoiding interactive lag, delay or interruption caused by network quality fluctuations or decline, thereby significantly improving the stability and smoothness of human-computer interaction, and bringing users a nearly "zero-awareness" seamless switching experience.

[0035] Embodiment 2 In particular, considering that in the above-mentioned intelligent Internet of Things communication method, the network status score value of each interactive network is calculated by weighted scoring of the network status data, but since the network status data is essentially dynamic and changes over time, isolated instantaneous data points are often difficult to fully reflect the true performance and potential trends of the network, and may be interfered with by short-term fluctuations or noise. Therefore, in order to more accurately capture the actual network status characteristics of each interactive network, in an embodiment of the present application, a network status evaluation method based on time series data association analysis is proposed. It uses time series analysis technology to perform time series modeling analysis on the network status data of the interactive network within a predetermined time window, revealing the continuous changes and evolution patterns of network performance in the time dimension, thereby more accurately evaluating the current network status and providing a more reliable basis for subsequent interactive network selection and switching.

[0036] Figure 4 FIG is a flowchart of sub-step S2 of the smart Internet of Things communication method according to an embodiment of the present application. Figure 4 As shown, the step S2 includes the steps of: S21, sorting the network status data of the interactive network according to the timestamp to obtain the short time series of received signal strength, the short time series of signal-to-noise ratio and the short time series of packet loss rate; S22, time series encoding the short time series of received signal strength, the short time series of signal-to-noise ratio and the short time series of packet loss rate to obtain the time series correlation feature of received signal strength, the time series correlation feature of signal-to-noise ratio and the time series correlation feature of packet loss rate; S23, fusing the time series correlation feature of received signal strength, the time series correlation feature of signal-to-noise ratio and the time series correlation feature of packet loss rate to obtain the multimodal time series fusion feature of network status; S24, feature decoding the multimodal time series fusion feature of network status to obtain the network status score value.

[0037] Specifically, the step S21, the network state data of the interaction network is data arranged according to the time stamp to obtain a short time sequence of received signal strength, a short time sequence of signal-to-noise ratio and a short time sequence of packet loss rate. It should be understood that, since the network parameters (such as signal strength, signal-to-noise ratio, packet loss rate) fluctuate dynamically with time and have short-term correlation (for example, the signal strength shows a continuous attenuation trend due to the movement of the device). Therefore, in order to capture the time sequence change rule of the network state parameter and provide structured input for subsequent time sequence modeling analysis, based on the time sequence data segmentation and alignment principle, the collected received signal strength (RSSI), signal-to-noise ratio (SNR) and packet loss rate are respectively divided into equal-length short time sequences (such as sampling once per second, each sequence contains 10 data points) in units of fixed time window (such as 10 seconds) according to the time stamp, and the missing values are filled by interpolation method. For example, when the Bluetooth signal loses a second of data due to interference, a linear interpolation of adjacent time points is used to generate a replacement value. In this way, the original discrete data is converted into a parameter sequence with time continuity, which helps to maintain the integrity and continuity of the time sequence data and lays a foundation for subsequent time sequence modeling analysis.

[0038] Specifically, the step S22, the received signal strength short time sequence, the signal-to-noise ratio short time sequence and the packet loss rate short time sequence are time sequence coded to obtain received signal strength time sequence correlation features, signal-to-noise ratio time sequence correlation features and packet loss rate time sequence correlation features. Wherein, Figure 5 The flow chart of the sub-step S22 of the intelligent Internet of Things communication method according to the embodiment of the application. As shown in Figure 5 The step S22 includes the steps of: S221, time sequence coding of the received signal strength short time sequence, the signal-to-noise ratio short time sequence and the packet loss rate short time sequence based on the forward LSTM model to obtain initial received signal strength time sequence correlation features, initial signal-to-noise ratio time sequence correlation features and initial packet loss rate time sequence correlation features; S222, time sequence fine-grained reinforcement of the initial received signal strength time sequence correlation features, the initial signal-to-noise ratio time sequence correlation features and the initial packet loss rate time sequence correlation features to obtain the received signal strength time sequence correlation features, the signal-to-noise ratio time sequence correlation features and the packet loss rate time sequence correlation features.

[0039] More specifically, the step S221 performs time series coding on the short-time sequence of received signal strength, the short-time sequence of signal-to-noise ratio and the short-time sequence of packet loss rate based on a forward LSTM model to obtain initial time series correlation features of received signal strength, initial time series correlation features of signal-to-noise ratio and initial time series correlation features of packet loss rate. It can be understood that the present application takes into account that the simple weighted scoring method ignores the time sequence dependence of parameters, for example, a sustained decline in signal strength may indicate further deterioration of network quality, even if the signal strength at the current time point is still within the good range, when the network state is evaluated, such trend changes should also be considered. Therefore, in order to fully tap and utilize the time sequence dependence and correlation in the time series data of each network state parameter, the present application introduces a forward LSTM model to perform time series coding on the short-time sequence of received signal strength, the short-time sequence of signal-to-noise ratio and the short-time sequence of packet loss rate. As a special recurrent neural network (RNN), LSTM is particularly suitable for processing and predicting time sequence dependence in time series data. In the present application, for each short-time sequence (such as 10-second RSSI data), a three-layer LSTM network is constructed, in which the input layer receives single-variable time series data (such as the short-time sequence of received signal strength), the hidden layer performs forward transfer coding on the parameter information of each time step through the gating mechanism (forget gate, input gate, output gate) to capture the long-term dependence and internal pattern of time series data, and the output layer generates corresponding time series correlation features to abstract the time dynamics of single-dimensional parameters into high-dimensional semantic features. For example, after the forward LSTM coding of the short-time sequence of received signal strength, the output features (initial time series correlation features of received signal strength) can represent the change trend and potential fluctuation pattern of signal strength.

[0040] More specifically, the step S222, the initial received signal strength time sequence correlation feature, the initial signal to noise ratio time sequence correlation feature and the initial packet loss rate time sequence correlation feature are time sequence fine-grained enhanced to obtain the received signal strength time sequence correlation feature, the signal to noise ratio time sequence correlation feature and the packet loss rate time sequence correlation feature. It can be understood that the time sequence correlation feature obtained by the initial encoding of the forward LSTM model can represent the time dynamics of the network state parameters, but may be affected by the instantaneous noise or abnormal value in the original data, and contains redundant noise or minor fluctuations. Therefore, in order to improve the feature expression ability of the core law of the network state, the initial received signal strength time sequence correlation feature, the initial signal to noise ratio time sequence correlation feature and the initial packet loss rate time sequence correlation feature are respectively processed by time sequence expression enhancement, and the original feature is locally decomposed, distilled and reconstructed to suppress noise and minor fluctuations, and to enhance the main time sequence feature (for example, the slow decay trend caused by device movement in the initial signal to noise ratio time sequence correlation feature is retained, and the instantaneous fluctuation caused by burst interference is suppressed), to obtain more refined received signal strength time sequence correlation feature, signal to noise ratio time sequence correlation feature and packet loss rate time sequence correlation feature. Next, the enhancement coding of the received signal strength short time sequence is taken as an example for detailed description.

[0041] Figure 6 The flow chart of the sub-step S222 of the intelligent Internet of Things communication method according to the embodiment of the application. As shown in the figure, Figure 6 The step S222 includes the steps of: S2221, performing feature decomposition based on one-dimensional convolution coding on the initial received signal strength time sequence correlation feature to obtain a set of initial received signal strength local time sequence correlation hidden features; S2222, based on the global correlation topology of the set of initial received signal strength local time sequence correlation hidden features, performing feature distillation coding on each initial received signal strength local time sequence correlation hidden feature in the set of initial received signal strength local time sequence correlation hidden features to obtain a set of distilled initial received signal strength local time sequence correlation hidden features; S2223, performing feature reconstruction based on a self-attention mechanism on the set of distilled initial received signal strength local time sequence correlation hidden features to obtain the received signal strength time sequence correlation feature.

[0042] In one specific example of the application, the step S2221 is expressed by the formula: wherein, represents the initial received signal strength time sequence correlation feature, represents the one-dimensional convolution coding operation based on the convolution kernel, is the scale of the one-dimensional convolution kernel, represents a set of implicit features of the local temporal correlation of the initial received signal strength, 、 、 and They represent the first, second, and third in the set of the initial received signal strength local temporal correlation implicit features. and The implicit characteristics of the local temporal correlation of the initial received signal strength, The number of implicit features associated with the local temporal sequence of the initial received signal strength.

[0043] That is, by sliding the one-dimensional convolution kernel to scan the initial received signal strength temporal correlation features, the weight sharing mechanism and the local receptive field characteristics are utilized to decouple and make explicit the local temporal correlation patterns in different local time domains from the initial received signal strength temporal correlation features, and convert the overall temporal features into a set of distributed initial received signal strength local temporal correlation implicit features, providing a finer-grained feature input for subsequent feature enhancement and reconstruction.

[0044] Figure 7 Flowchart of sub-step S2222 of the intelligent Internet of Things communication method according to an embodiment of the present application. Figure 7 As shown, the step S2222 includes the steps of: S22221, calculating the feature structure correlation coefficient between any two initial received signal strength local timing correlation implicit features in the set of the initial received signal strength local timing correlation implicit features to obtain an initial received signal strength local timing feature structure correlation topology matrix composed of multiple feature structure correlation coefficients; S22222, inputting the initial received signal strength local timing feature structure correlation topology matrix into a gated mask function to obtain an initial received signal strength local timing feature structure fine-grained correlation mask topology matrix; S22223, based on the initial received signal strength local timing feature structure fine-grained correlation mask topology matrix, performing feature structure feedback modulation on each initial received signal strength local timing correlation implicit feature in the set of the initial received signal strength local timing correlation implicit features to obtain the set of the distilled initial received signal strength local timing correlation implicit features.

[0045] In a specific example of the present application, step S22221 is expressed as follows: in, The first in the set of implicit features of the local temporal correlation of the initial received signal strength The implicit characteristics of the local temporal correlation of the initial received signal strength, represents transpose, denotes a computation of a norm, denotes a bandwidth parameter, denotes an exponential function with base e, denotes and denotes a feature structure correlation coefficient between the initial received signal strength local timing feature structure correlation topology matrix and the initial received signal strength local timing feature structure correlation topology matrix, that is, the element value at the position of the initial received signal strength local timing feature structure correlation topology matrix.

[0046] That is, by quantifying the proximity and correlation of the implicit features of any two initial received signal strength local timing correlation features on the manifold structure, the feature structure correlation coefficients are arranged to form the initial received signal strength local timing feature structure correlation topology matrix, thereby explicitly modeling the internal structural relationship and geometric correlation between the features, providing deep representations containing local correlation structure information of the feature space for subsequent feature distillation coding, thereby improving the targeting and efficiency of feature distillation.

[0047] In one specific example of the present application, the step S22222 is represented by the formula: wherein, denotes a gating mask weight matrix, denotes an initial received signal strength local timing feature structure correlation topology matrix, denotes a gating mask bias matrix, denotes a sigmoid activation function, denotes an initial received signal strength local timing feature structure fine-grained correlation mask topology matrix.

[0048] That is, by using the gating mask function as a nonlinear transformation mechanism, the mask is dynamically generated according to the context information in the initial received signal strength local timing feature structure correlation topology matrix, and the correlation strength between the features is adaptively filtered and adjusted, highlighting the key local time domain correlation structure and suppressing noise or irrelevant structure correlation, achieving attention focusing on feature correlation, thereby obtaining an initial received signal strength local timing feature structure fine-grained correlation mask topology matrix that is more consistent with the actual network state.

[0049] In particular, in a preferred example of the present application, the step S22223 includes: first, performing local structural equalization optimization on the initial received signal strength local timing feature structure fine-grained correlation mask topology matrix to obtain an optimized received signal strength local timing feature structure fine-grained correlation mask topology matrix. Here, considering that the geometric coupling configuration distribution of the initial received signal strength local timing feature structure correlation topology matrix on the implicit low-dimensional differential cluster structure will be nonlinear and unsaturated, the global correlation topology configuration of the initial received signal strength local timing feature structure correlation topology matrix is ​​compressed due to the non-affine correlation constraint, and this will become more significant due to the correlation polarization gain of the gated mask function, thereby affecting the intrinsic geometric microstructure coupling expression effect of the initial received signal strength local timing feature structure fine-grained correlation mask topology matrix. Therefore, the present application further performs local structural equalization optimization processing on the initial received signal strength local timing feature structure fine-grained correlation mask topology matrix.

[0050] Specifically, for each eigenvalue of the fine-grained correlation mask topology matrix of the initial received signal strength local temporal feature structure First, the gradient vector field term is introduced to compensate for the local nonlinear geometric distortion, thereby realizing the geometric microstructure regularization of the fine-grained correlation mask topology matrix of the local temporal feature structure of the initial received signal strength: in, express Middle Rank Elements of the column, In the fine-grained correlation mask topology matrix representing the local temporal feature structure of the initial received signal strength The corresponding gradient vector field term, express Middle Rank Elements of the column, It means partial derivative.

[0051] Then, the gradient vector field term is used as an exogenous driving factor to regulate the average field of each eigenvalue: in, represents the characteristic mean between all eigenvalues ​​of the fine-grained correlation mask topology matrix of the local temporal feature structure of the initial received signal strength, represents the gain modulation coefficient, Represents the first in the fine-grained correlation mask topology matrix of the optimized received signal strength local temporal feature structure Rank Elements of a column.

[0052] In this way, under the action of the external field driving term as a high-order differential characteristic, the nonlinear saturation of the geometric correlation distribution under the mean field (i.e., the gradient sensitivity attenuation) is reversely promoted, thereby compensating for the sub-geometric coupling grid decoupling caused by the correlation polarization enhancement through the mean field synergy response under the mean field, thereby improving the substantial geometric fine-grained correlation structure expression effect of the fine-grained correlation mask topology matrix of the local temporal characteristic structure of the initial received signal strength.

[0053] Next, the initial received signal strength local temporal correlation implicit feature and the optimized received signal strength local temporal feature structure fine-grained correlation mask topology matrix are input into the feature dense feedback distillation unit to obtain the distilled initial received signal strength local temporal correlation implicit feature, which is expressed as follows: in, Represents the fine-grained correlation mask topology matrix of the optimized received signal strength local temporal feature structure, represents dot product, represents the matrix multiplication operation, represents a nonlinear activation function, represents the distillation weight matrix, express The characteristic scale value of Indicates the first in the set of implicit features of the local temporal correlation of the initial received signal strength The implicit features of the local temporal correlation of the initial received signal strength are distilled.

[0054] That is, based on the fine-grained correlation mask topology matrix of the optimized received signal strength local temporal feature structure, the initial received signal strength local temporal correlation implicit features are subjected to feature mask modulation to achieve feature distillation, so that each initial received signal strength local temporal correlation implicit feature is refined based on the effective information of its associated neighbors and the expression consistency between features is coordinated, thereby generating a set of distilled initial received signal strength local temporal correlation implicit features. In this way, each initial received signal strength local temporal correlation implicit feature can be integrated into the comprehensive perspective of multi-source correlation features while retaining its own characteristics, forming a feature representation with more complete information, less ambiguity, and mutual coordination.

[0055] In a specific example of the present application, step S2223 is expressed as follows: in, denote a set of distilled initial received signal strength local temporal correlation implicit features, 、 and denote the first, second and third distilled initial received signal strength local temporal correlation implicit features in the set of distilled initial received signal strength local temporal correlation implicit features, 、 、 and denote a query matrix, a key matrix and a value matrix respectively, 、 and denote a query embedding matrix, a key embedding matrix and a value embedding matrix respectively, denote a softmax activation function, denote the received signal strength temporal correlation feature.

[0056] That is, by using the powerful global information integration capability of the self-attention mechanism, long-distance dependency relationships are dynamically captured among the distilled initial received signal strength local temporal correlation implicit features, key features are adaptively determined according to the context, and the refined local detailed information and its complex global dependency relationships are reconstructed into a single, unified and high-order received signal strength temporal correlation feature, so as to realize the fusion and sublimation of feature information from the local to the global.

[0057] Specifically, the step S23 fuses the received signal strength temporal correlation feature, the signal-to-noise ratio temporal correlation feature and the packet loss rate temporal correlation feature to obtain a network state multi-modal temporal fusion feature. It can be understood that, since different network parameters (RSSI, SNR and packet loss rate) have complementarity in representing the network state (for example, a high signal-to-noise ratio can make up for the deficiency of a low signal strength), in order to integrate multi-dimensional information and build a global evaluation basis, the present application fuses the received signal strength temporal correlation feature, the signal-to-noise ratio temporal correlation feature and the packet loss rate temporal correlation feature based on a multi-modal feature fusion strategy, so as to avoid the limitation of a single parameter and help to evaluate the network quality based on a global perspective. In the embodiments of the present application, the received signal strength temporal correlation feature, the signal-to-noise ratio temporal correlation feature and the packet loss rate temporal correlation feature are fused by using weighted summation, so as to differentially consider the importance of each parameter in the network state evaluation, and obtain the network state multi-modal temporal fusion feature.

[0058] Specifically, the step S24 is to decode the network state multi-modal time sequence fusion feature to obtain the network state score value. Specifically, in order to map the high-dimensional abstract network state multi-modal time sequence fusion feature to a quantifiable score value, the present application is based on the feature-score regression principle, and a feature decoding module is constructed by using a fully connected neural network. By performing multi-level nonlinear mapping and regression calculation on the network state multi-modal time sequence fusion feature, it is converted into a single-dimensional network state score value, so as to intuitively reflect the comprehensive state of the current network. The higher the value, the better the network quality, and vice versa. Thus, a reliable basis is provided for network selection and switching.

[0059] In summary, the intelligent Internet of Things communication method based on the embodiments of the present application is illustrated. It calculates the network state score value of each interactive network by real-time monitoring of state parameters such as the received signal strength, signal-to-noise ratio and packet loss rate of multiple available interactive networks, quantifies the real-time performance of different networks, and generates a candidate network list based on the dynamic ranking of the score value. When the state score of the current interactive network is lower than the optimal candidate network, a backup communication channel is established on the target optimal network in advance, and seamless transition is achieved through parallel transmission of the primary and backup channels. In this way, the communication strategy can be adjusted intelligently and timely when the current network quality decreases or a better network option appears, and smooth transition is achieved through synchronous transmission of data packets through the pre-established backup channel, thereby ensuring the smoothness and stability of human-computer interaction and avoiding the decline of user experience or interruption of interaction caused by network problems.

[0060] The basic principles of the present application are described above in combination with specific embodiments. However, it should be noted that the advantages, advantages, effects and the like mentioned in the present application are only examples and not limitations, and these advantages, advantages, effects and the like cannot be considered as the necessary possession of each embodiment of the present application. In addition, the specific details of the above embodiments are only for the purpose of example and understanding, and are not limited to the present application, and the above details do not limit the present application to the above specific details.

[0061] In the above embodiments, the description of each embodiment has its own focus, and the parts not described or recorded in a certain embodiment can be referred to the relevant description of other embodiments. In the several embodiments provided by the present application, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiments described above are merely schematic, for example, the unit division is only a logical function division, and there can be other division manners in actual implementation. The units described as separated components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., they can be located in one place or distributed on a plurality of network units. In actual implementation, some or all of the units can be selected according to actual needs to achieve the purposes of the embodiments.

[0062] It is apparent that for one skilled in the art, the present application is not limited to the details of the above exemplary embodiments, and can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all aspects as illustrative and non-restrictive, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended to cover all changes falling within the meaning and range of equivalents of the claims. Any reference signs in the claims should not be considered as limiting the claims involved.

[0063] Furthermore, it is obvious that the word "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. Multiple units recited in a system claim can also be implemented by one unit by means of software or hardware.

[0064] Finally, it should be noted that the above description is given for the purpose of illustration and description. Furthermore, the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the present application. Although the technical solutions are modified or replaced by equivalents in the preferred embodiments, they do not deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. A smart Internet of Things communication method, characterized in that: include: Step 1: Real-time monitoring of network status data of multiple interactive networks; Step 2: Calculate the network status score of each interactive network based on the network status data; Step 3: sorting the multiple interactive networks based on the network status score values ​​to obtain a candidate network list; Step 4: Selecting the optimal interaction network from the candidate network list for human-computer interaction; Step 5: When the current interactive network is not optimal, actively switch to the optimal interactive network in the candidate network list.

2. The intelligent Internet of Things communication method according to claim 1, characterized in that: In step 1, the network status data includes a received signal strength value, a signal-to-noise ratio, and a packet loss rate.

3. The intelligent Internet of Things communication method according to claim 2, characterized in that: The step 2 includes: inputting the network status data into a weighted evaluation model to obtain the network status score value, wherein the weight value of the received signal strength value is 0.3, the weight value of the signal-to-noise ratio is 0.5, and the weight value of the packet loss rate is 0.

2.

4. The intelligent Internet of Things communication method according to claim 1, characterized in that: The step 5 comprises: Setting the current interaction network as the first interaction network and defining the best interaction network in the candidate network list as the second interaction network; Before deciding to switch to the second interactive network, pre-establish a backup communication channel through the second interactive network and keep the primary communication channel and the backup communication channel accessing the server simultaneously for a short period of time; During the switching to the second interactive network, it is ensured that the heartbeat mechanism of the backup communication channel is normal and the data packets of the primary communication channel and the backup communication channel are synchronized and processed.

5. The intelligent Internet of Things communication method according to claim 4, characterized in that: Synchronizing and processing data packets of the primary communication channel and the backup communication channel, including: Sequencing and marking the data packets of the primary communication channel and the backup communication channel to obtain the priority of the data packets; Data packets are processed in order of their priority.

6. The intelligent Internet of Things communication method according to claim 5, characterized in that: The step 2 comprises: Arrange the network status data of the interactive network according to the timestamp to obtain the short-time series of received signal strength, signal-to-noise ratio and packet loss rate; Performing time series encoding on the received signal strength short-time series, the signal-to-noise ratio short-time series, and the packet loss rate short-time series to obtain received signal strength time series correlation features, signal-to-noise ratio time series correlation features, and packet loss rate time series correlation features; The received signal strength time series correlation features, signal-to-noise ratio time series correlation features, and packet loss rate time series correlation features are integrated to obtain the multimodal time series fusion features of the network status; Feature decoding is performed on the network state multimodal time series fusion feature to obtain the network state score value.

7. The intelligent Internet of Things communication method according to claim 6, characterized in that: Perform time series encoding on the received signal strength short-time series, the signal-to-noise ratio short-time series, and the packet loss rate short-time series to obtain received signal strength time series correlation features, signal-to-noise ratio time series correlation features, and packet loss rate time series correlation features, including: Perform time series encoding based on the forward LSTM model on the short time series of received signal strength, signal-to-noise ratio and packet loss rate to obtain the initial time series correlation features of received signal strength, initial time series correlation features of signal-to-noise ratio and initial time series correlation features of packet loss rate; The initial received signal strength timing correlation feature, the initial signal-to-noise ratio timing correlation feature, and the initial packet loss rate timing correlation feature are subjected to timing fine-grained enhancement to obtain the received signal strength timing correlation feature, the signal-to-noise ratio timing correlation feature, and the packet loss rate timing correlation feature.

8. The intelligent Internet of Things communication method according to claim 7, characterized in that: Performing time-series fine-grained enhancement on the initial received signal strength time-series correlation feature to obtain the received signal strength time-series correlation feature includes: Performing feature decomposition based on one-dimensional convolution coding on the initial received signal strength time series correlation feature to obtain a set of initial received signal strength local time series correlation implicit features; Based on the global correlation topology of the set of initial received signal strength local temporal correlation implicit features, perform feature distillation coding on each initial received signal strength local temporal correlation implicit feature in the set of initial received signal strength local temporal correlation implicit features to obtain a set of distilled initial received signal strength local temporal correlation implicit features; The set of local temporal correlation implicit features of the distilled initial received signal strength is subjected to feature reconstruction based on a self-attention mechanism to obtain the received signal strength temporal correlation features.

9. The intelligent Internet of Things communication method according to claim 8, characterized in that: Based on the global correlation topology of the set of initial received signal strength local temporal correlation implicit features, feature distillation coding is performed on each initial received signal strength local temporal correlation implicit feature in the set of initial received signal strength local temporal correlation implicit features to obtain a set of distilled initial received signal strength local temporal correlation implicit features, including: Calculating a feature structure correlation coefficient between any two initial received signal strength local time series correlation implicit features in the set of initial received signal strength local time series correlation implicit features to obtain an initial received signal strength local time series feature structure correlation topology matrix composed of a plurality of feature structure correlation coefficients; Inputting the initial received signal strength local temporal feature structure correlation topology matrix into a gated mask function to obtain an initial received signal strength local temporal feature structure fine-grained correlation mask topology matrix; Based on the fine-grained correlation mask topology matrix of the initial received signal strength local timing feature structure, feature structure feedback modulation is performed on each initial received signal strength local timing correlation implicit feature in the set of initial received signal strength local timing correlation implicit features to obtain the set of distilled initial received signal strength local timing correlation implicit features.

10. The intelligent Internet of Things communication method according to claim 9, characterized in that: Based on the fine-grained correlation mask topology matrix of the initial received signal strength local temporal sequence feature structure, performing feature structure feedback modulation on each initial received signal strength local temporal sequence correlation implicit feature in the set of the initial received signal strength local temporal sequence correlation implicit features to obtain the set of the distilled initial received signal strength local temporal sequence correlation implicit features, including: Performing local structure equalization optimization on the initial received signal strength local timing feature structure fine-grained correlation mask topology matrix to obtain an optimized received signal strength local timing feature structure fine-grained correlation mask topology matrix; The initial received signal strength local temporal correlation implicit feature and the optimized received signal strength local temporal feature structure fine-grained correlation mask topology matrix are input into a feature dense feedback distillation unit to obtain the distilled initial received signal strength local temporal correlation implicit feature.

Images