Intelligent lighting control method based on wireless networking technology

Abstract

The invention relates to the field of intelligent control, and particularly discloses an intelligent lighting control method based on a wireless networking technology, which is not limited to single signal strength or LQI, but enables a network node to actively and continuously sense a bidirectional communication state with a neighbor node. A node obtains RSSI and LQI with timestamps from a neighbor data packet, actively sends a packet to monitor ACK, counts successful ACK and retransmission times, and calculates a recent packet delivery success rate and expected sending times. Based on the multi-dimensional dynamic parameter, the comprehensive link cost to each neighbor is calculated. When a routing decision is made, after a node receives a neighbor routing declaration, the total cost of a potential complete path is calculated in combination with the cost of a comprehensive link from the node to the neighbor and the cost of a gateway declarated by the neighbor, and the total cost is compared with the current optimal path. And if the new path is better, the next hop is updated, the suboptimal routing is avoided, and the control instruction is ensured to be reliable and timely.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to an intelligent lighting control method based on wireless networking technology. Background Art

[0002] With the rapid development of the Internet of Things (IoT) and the advancement of smart city construction, smart lighting, as a key component, is gaining increasing attention. Traditional lighting systems typically utilize wired control, which presents challenges such as complex wiring, high costs, and difficulty in expansion and retrofitting. These drawbacks are particularly prominent in large buildings, urban roads, or when upgrading existing structures. To overcome these limitations, smart lighting control solutions based on wireless networking technologies have emerged. Leveraging wireless communication technologies such as ZigBee, Wi-Fi, and Bluetooth Mesh, these solutions connect a large number of lighting nodes (such as lamps) into a self-organizing wireless network, enabling remote, centralized, and intelligent management of lighting equipment. This not only significantly improves energy efficiency and ease of management, but also lays the foundation for a richer range of scene-based linkage and a more personalized lighting experience.

[0003] However, existing wireless lighting control solutions still face significant challenges in their core wireless networking technologies, particularly routing mechanisms. In wireless mesh networks comprised of numerous lighting nodes, data (such as control commands and status reports) must be transmitted between end nodes and the gateway via multi-hop routing. The stability and efficiency of routing directly determine the responsiveness and reliability of the entire lighting system. Currently, many routing protocols select next-hop nodes based on a single or a few static combinations of link quality metrics, such as the received signal strength indicator (RSSI) or link quality indicator (LQI). The drawback of this approach is that, while RSSI reflects signal strength, it does not fully represent the actual communication quality of the link and can even cause high packet loss rates due to interference even under strong signals. LQI, on the other hand, is only a partial assessment of packet reception quality at the physical layer. Some improved solutions attempt to calculate link cost by using a linearly weighted combination of multiple metrics (such as RSSI, LQI, and packet loss rate), but these weights are typically pre-set or fixed. This static weighting approach is unsuitable for the complex and changing electromagnetic environment of smart lighting systems. For example, at certain times, the network may be subject to sudden interference, in which case the packet delivery success rate is a better indicator of link availability than signal strength. At other times, changes in network load may make the expected number of transmissions a more critical consideration. Fixed weights cannot dynamically reflect the relative importance of different link quality parameters under varying network conditions, leading to inaccurate link cost assessments and the easy selection of "pseudo-optimal" links. This can cause frequent routing switches, increased data transmission delays, and even communication interruptions, seriously impacting the control effectiveness and user experience of smart lighting systems.

[0004] Therefore, an optimized lighting control solution based on wireless networking technology is desired. Summary of the Invention

[0005] In order to solve the above technical problems, the present application is proposed. The embodiment of the present application provides an intelligent lighting control method based on wireless networking technology, which is not limited to a single signal strength or LQI, but allows network nodes to actively and continuously perceive the two-way communication status with neighboring nodes. The node obtains the RSSI and LQI with timestamps from the neighbor data packet, and actively sends packets to monitor ACK, counts the number of successful ACKs and retransmissions, and calculates the recent packet delivery success rate and the expected number of transmissions. Based on this multi-dimensional dynamic parameter, the comprehensive link cost to each neighbor is calculated. When making routing decisions, after receiving the neighbor routing announcement, the node combines its own comprehensive link cost to the neighbor and the gateway cost announced by the neighbor to calculate the total cost of the potential complete path and compare it with the current optimal path. If the new path is better, the next hop is updated. This refined, multi-source real-time data fusion link cost evaluation and dynamic routing update can accurately grasp the true quality of the link, avoid suboptimal routing, and ensure that control instructions are reliable and timely.

[0006] According to one aspect of the present application, a smart lighting control method based on wireless networking technology is provided, which includes: The first node receives a neighbor data packet from each of its neighbor nodes, and obtains a time-stamped RSSI value and a time-stamped LQI value of each neighbor node from the neighbor data packet; In the process of the first node sending a data packet to each of its neighboring nodes and expecting an ACK, counting the number of successful ACK receptions and the number of retransmissions, and calculating the recent packet delivery success rate and the recent expected number of transmissions from the first node to each of the neighboring nodes based on the number of successful ACK receptions and the number of retransmissions; Calculate, based on the timestamp RSSI value and the timestamp LQI value of each neighboring node and the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node, a comprehensive link cost value from the first node to each neighboring node; When the first node receives a routing announcement from an i-th neighbor node, the first node calculates a total link cost of reaching a target gateway through the i-th neighbor node; In response to the current total link cost to reach the target gateway being greater than the total link cost to reach the target gateway through the i-th neighbor node, the first node sets the i-th neighbor node as the next-hop node ID.

[0007] Compared with the existing technology, the present application provides an intelligent lighting control method based on wireless networking technology, which is not limited to a single signal strength or LQI, but allows network nodes to actively and continuously perceive the two-way communication status with neighboring nodes. The node obtains the RSSI and LQI with timestamps from the neighbor's data packet, and actively sends packets to monitor ACK, counts the number of successful ACKs and retransmissions, and calculates the recent packet delivery success rate and the expected number of transmissions. Based on this multi-dimensional dynamic parameter, the comprehensive link cost to each neighbor is calculated. When making routing decisions, after receiving the neighbor's routing announcement, the node combines its own comprehensive link cost to the neighbor and the gateway cost announced by the neighbor to calculate the total cost of the potential complete path and compare it with the current optimal path. If the new path is better, the next hop is updated. This refined, multi-source real-time data fusion link cost assessment and dynamic routing update can accurately grasp the true quality of the link, avoid suboptimal routing, and ensure that control instructions are reliable and timely. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] The above and other purposes, features, and advantages of the present application will become more apparent through a more detailed description of the embodiments of the present application in conjunction with the accompanying drawings. The accompanying drawings are intended to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the drawings, the same reference numerals generally represent the same components or steps.

[0009] Figure 1 Flowchart of an intelligent lighting control method based on wireless networking technology according to an embodiment of the present application; Figure 2 Schematic diagram of data flow of an intelligent lighting control method based on wireless networking technology according to an embodiment of the present application; Figure 3 Flowchart for determining the weight vector of the intelligent lighting control method based on wireless networking technology according to an embodiment of the present application; Figure 4 This is a flowchart of an intelligent lighting control method based on wireless networking technology according to an embodiment of the present application, which performs low-dimensional spatial information aggregation on RSSI low-dimensional embedded coding vectors, LQI low-dimensional embedded coding vectors, recent packet delivery success rate low-dimensional embedded coding vectors, and recent expected number of transmission times to obtain a low-dimensional aggregated coding vector of the link status of neighboring active nodes. DETAILED DESCRIPTION

[0010] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0011] As used in this application and the claims, unless the context clearly indicates otherwise, the words "a," "an," "an," and / or "the" are not intended to refer to the singular but may include the plural. Generally speaking, the terms "comprises" and "include" only indicate the inclusion of the steps and elements specifically identified, and these steps and elements do not constitute an exclusive list. A method or apparatus may also include other steps or elements.

[0012] 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.

[0013] Flowcharts are used in this application to illustrate the operations performed by the systems according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in exact order. Instead, the various steps may be processed in reverse order or simultaneously, as needed. Furthermore, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0014] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described herein.

[0015] In existing wireless intelligent lighting systems, routing selection mechanisms often rely on one-sided or fixed link metrics, making them difficult to adapt to complex and changing wireless environments. This can lead to technical issues such as inaccurate link quality assessment, frequent route switching, and control command transmission delays. The present invention proposes an intelligent lighting control method based on wireless networking technology, which features an intelligent routing decision-making mechanism that adaptively learns and dynamically assesses link quality. This solution goes beyond the single signal strength or link quality indicator (LQI) and instead enables each node in the network (e.g., the first node) to proactively and continuously perceive the bidirectional communication status with its neighboring nodes. Specifically, the node not only captures the timestamped RSSI and LQI values ​​from received neighboring data packets in real time, but more importantly, it actively sends data to neighboring nodes and monitors the acknowledgment (ACK) responses, accurately counting the number of successful ACKs and necessary retransmissions. This allows the method to calculate the packet delivery success rate and expected number of transmissions, which accurately reflect recent data transmission efficiency. Based on these multi-dimensional, dynamically updated link parameters, the present invention further calculates the comprehensive link cost to each neighboring node. When making routing decisions, a node receives a routing announcement from a neighbor and combines its own comprehensive link cost to that neighbor with the cost to the target gateway announced by the neighbor to calculate the total cost of a potential complete path. This cost is then compared with the currently known optimal path cost. Only when the new path demonstrates a better total link cost will the node update its next-hop selection. This refined link cost assessment and dynamic routing update mechanism, based on multi-source real-time data fusion, more accurately assesses the true quality of wireless links, effectively avoiding suboptimal routing choices caused by incomplete assessments and ensuring reliable and timely response to control commands from the intelligent lighting system.

[0016] In the technical solution of the present application, an intelligent lighting control method based on wireless networking technology is proposed. Figure 1 Flowchart of an intelligent lighting control method based on wireless networking technology according to an embodiment of the present application. Figure 2Schematic diagram of data flow of the intelligent lighting control method based on wireless networking technology according to an embodiment of the present application. Figure 1 and Figure 2 As shown, according to the embodiment of the present application, the intelligent lighting control method based on wireless networking technology includes the following steps: S100, a first node receives neighbor data packets from each of its neighbor nodes, and obtains the RSSI value and LQI value with a time stamp of each neighbor node from the neighbor data packet; S200, in the process of the first node sending data packets to each of its neighbor nodes and expecting ACK, counting the number of successful ACK receptions and the number of retransmissions, and based on the number of successful ACK receptions and the number of retransmissions, respectively calculating the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighbor node; S300, based on the number of successful ACK receptions and the number of retransmissions, calculating the RSSI value and the LQI value with a time stamp of each neighbor node; The timestamp RSSI value and timestamp LQI value of the resident node and the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighbor node are used to calculate the comprehensive link cost value from the first node to each neighbor node; S400, when the first node receives the routing announcement from the i-th neighbor node, the first node calculates the total link cost of reaching the target gateway through the i-th neighbor node; S500, in response to the current total link cost to reach the target gateway being greater than the total link cost to reach the target gateway through the i-th neighbor node, the first node sets the i-th neighbor node as the next hop node ID.

[0017] Specifically, in step S100, the first node receives neighbor data packets from each of its neighbor nodes and obtains the timestamped RSSI and LQI values ​​of each neighbor node from the neighbor data packets. It should be understood that the RSSI value primarily reflects the attenuation of the signal during propagation and is a direct indicator of signal strength; the LQI value assesses the quality of the data packet's correct reception at the physical layer from a more comprehensive perspective (e.g., signal-to-noise ratio, bit error rate, etc.). The key to introducing timestamps is to capture the dynamic nature of these physical layer parameters. Because wireless channel quality is susceptible to real-time fluctuations due to various factors such as environmental interference, node mobility, and multipath effects, outdated link information cannot accurately reflect the current link status and may lead to erroneous routing decisions. Using timestamps, the system can assign higher weight to recent data or use it to analyze link quality trends, making link assessment more timely and predictive. It is particularly worth noting that the neighbor data packets include hello packets, normal data packets, and ACK packets.

[0018] More specifically, in one specific example of this application, the standardized operational procedures of a wireless communication protocol stack are generally followed: First, at the overall network level, each node (including the first node and its neighboring nodes) is equipped with a wireless transceiver module and runs the corresponding wireless networking protocol (such as Zigbee, Bluetooth Mesh, etc.). Neighboring nodes periodically broadcast Hello packets or, when conducting data communications, send normal data packets and ACK packets. Second, at the receiving level of the first node, its wireless transceiver module continuously monitors the designated channel. When a signal arrives from any neighboring node, the physical layer (PHY) hardware automatically measures the received signal power and converts it into an RSSI value. Simultaneously, the physical layer or MAC (Medium Access Control) layer calculates the LQI value based on the demodulation quality of the received data packet, preamble / sync header detection, CRC check results, or a specific signal quality assessment algorithm (such as one based on bit error rate or signal-to-noise ratio estimation). Third, at the data processing level, once a data packet is successfully received and decoded, the MAC or network layer software immediately reads the current node's system clock and timestamps the extracted RSSI and LQI values. This information (source neighbor ID, RSSI value, LQI value, timestamp) is then stored in the first node's neighbor table or a similar dynamic data structure for subsequent use by the link cost calculation module. This mechanism ensures that the first node can continuously and passively collect real-time physical layer information about the link status between it and each neighbor, laying a solid data foundation for building a dynamic and accurate routing foundation.

[0019] Specifically, in step S200, as the first node sends a data packet to each of its neighboring nodes and expects an ACK, it counts the number of successful ACK receptions and retransmissions. Based on these counts, the node calculates the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node. It should be understood that, unlike the RSSI and LQI at the physical layer, the PDR and ETX more directly reflect the ability of a data packet to successfully reach and be acknowledged by a neighboring node, given the combined influence of current channel conditions, interference levels, and MAC layer protocol interactions (such as collision avoidance and backoff mechanisms). The recent packet delivery success rate (Recent PDR) intuitively represents the probability that a data packet will be successfully received and acknowledged by a neighboring node after one or more recent attempts, and is a key metric for measuring link reliability. The recent expected number of transmissions (Recent ETX), the average number of transmission attempts required to successfully deliver a data packet (including the initial transmission and all retransmissions), reflects the transmission efficiency and cost of the link. A lower ETX indicates higher link efficiency, resulting in less radio resource usage and energy consumption. The introduction of these indicators based on actual transceiver interactions can make up for the one-sidedness that may exist when relying solely on physical layer parameters such as RSSI / LQI (for example, a strong signal but severe interference leads to packet loss, or a general signal but a stable link), thereby providing a more comprehensive basis for subsequent comprehensive link cost calculations and routing decisions that is closer to actual communication performance. Therefore, in the technical solution of the present application, the number of successful ACK receptions and the number of retransmissions are further counted, and based on the number of successful ACK receptions and the number of retransmissions, the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node are calculated respectively. This can quantitatively evaluate the reliability and efficiency of the wireless link based on the actual transmission effect of the data link layer and even the network layer.

[0020] More specifically, in an embodiment of the present application, calculating the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node includes: calculating the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node using the following formula, which is: , , in, is the number of times ACK is successfully received, is the number of retransmissions, is the recent package delivery success rate, The expected number of sends in the near future.

[0021] Specifically, in step S300, a comprehensive link cost value from the first node to each neighbor node is calculated based on the timestamped RSSI value and timestamped LQI value of each neighbor node, as well as the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighbor node. It should be understood that RSSI and LQI reflect the signal quality at the physical layer but may not fully reveal the actual packet transmission success rate; whereas the recent packet delivery success rate and the recent expected number of transmissions directly quantify the reliability and efficiency of the data link layer. Therefore, to overcome the limitations of evaluating link quality in complex wireless environments using a single or a few indicators, a comprehensive link cost value from the first node to each neighbor node is further calculated based on the timestamped RSSI value and timestamped LQI value of each neighbor node, as well as the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighbor node. By comprehensively considering these indicators at different levels and with different characteristics and assigning appropriate weights, the true "health" and communication cost of a wireless link can be more comprehensively and accurately reflected. This comprehensive cost value can effectively distinguish "pseudo-optimal" links that appear to have good signals but actually have high packet loss rates or low transmission efficiency, thereby guiding the routing protocol to select a truly stable and efficient next-hop node, significantly improving the performance and reliability of the entire wireless network. This is especially important for application scenarios such as smart lighting systems that have high requirements for real-time performance and stability.

[0022] More specifically, in an embodiment of the present application, based on the timestamp RSSI value and the timestamp LQI value of each neighboring node and the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node, the comprehensive link cost value from the first node to each neighboring node is calculated respectively, including: calculating the comprehensive link cost value from the first node to each neighboring node using the following link cost formula, wherein the link cost formula is: , in, is the weight vector, RSSI value with timestamp, is the LQI value with timestamp, The recent package delivery success rate, is the expected number of sends in the near future, is the normalization function, is the comprehensive link cost value.

[0023] In particular, the weight vector is the core of this method, determining the relative importance of each metric in assessing link quality. Link cost calculation cannot rely on fixed weights; instead, the weight vectors for each link quality parameter must be dynamically determined based on the changing link status. Dynamic weight adjustment allows the link cost calculation model to more accurately reflect the actual network environment, enabling more optimal routing decisions. Furthermore, intelligently and accurately generating this dynamic weight vector is key to implementing adaptive routing. Simply analyzing the independent changes in each parameter is insufficient, as complex nonlinear relationships exist between multiple parameters, such as RSSI, LQI, packet delivery success rate, and expected number of transmissions.

[0024] Figure 3 FIG. 1 is a flow chart of determining the weight vector of the intelligent lighting control method based on wireless networking technology according to an embodiment of the present application. Figure 3 As shown, according to the intelligent lighting control method based on wireless networking technology of an embodiment of the present application, the determination of the weight vector includes the steps of: S310, performing low-dimensional embedded coding on the timestamp RSSI value and timestamp LQI value of each neighbor node and the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighbor node to obtain an RSSI low-dimensional embedded coding vector, an LQI low-dimensional embedded coding vector, a recent packet delivery success rate low-dimensional embedded coding vector and a recent expected number of transmissions low-dimensional embedded coding vector; S320, performing low-dimensional spatial information aggregation on the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector and the recent expected number of transmissions low-dimensional embedded coding vector to obtain a neighbor active node link state low-dimensional aggregate coding vector; S330, performing weighted regression decoding on the neighbor active node link state low-dimensional aggregate coding vector to obtain the weight vector.

[0025] Specifically, in step S310, low-dimensional embedding encoding is performed on the timestamped RSSI values ​​and LQI values ​​of each neighboring node, as well as the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node, to obtain a low-dimensional embedded encoding vector for RSSI, a low-dimensional embedded encoding vector for LQI, a low-dimensional embedded encoding vector for recent packet delivery success rate, and a low-dimensional embedded encoding vector for recent expected number of transmissions. It should be understood that the core technical purpose of performing low-dimensional embedding encoding on the timestamped RSSI values ​​and LQI values ​​of each neighboring node, as well as the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node, is to convert these raw link quality parameters, which may have different scales and complex distributions, into a dense, low-dimensional vector representation that is more amenable to subsequent advanced information processing (such as feature aggregation and weight regression). This conversion has several key advantages: First, low-dimensional embedding can capture the deep semantic information and potential associations inherent in the raw parameters, rather than just their apparent numerical values. For example, similar RSSI values ​​may represent completely different link states in different LQI contexts. Embedded coding can learn this contextual dependency. Secondly, mapping high-dimensional, sparse, or unstructured raw data into a low-dimensional space helps reduce the computational complexity and overfitting risk of subsequent models, improving the model's generalization capabilities. Thirdly, embedded coding allows different types of parameters (such as signal strength, link quality, transmission success rate, and transmission cost) to be unified into the same vector space for representation and comparison, laying the foundation for subsequent feature fusion and dynamic weight generation. The ultimate goal is to generate a set of vectors that more precisely and robustly characterize the characteristics of each link quality dimension, enabling more intelligent assessment of the overall link status.

[0026] More specifically, in one specific example of this application, for each time-stamped link parameter (e.g., a specific RSSI value) for each neighboring node, this value is used as input to the corresponding embedding layer. This timestamp information can be fed into the embedding layer as an additional feature or used to perform a weighted average of historical data before embedding to reflect timeliness. If the parameter is discretized into several categories (e.g., RSSI values ​​are categorized as "strong," "medium," and "weak"), the embedding layer can be a lookup table that takes a category index as input and outputs a corresponding low-dimensional vector. If the parameter is continuous, the embedding layer is a small feedforward neural network that maps the input scalar value into a fixed-dimensional vector. For example, an RSSI value of -60dBm is converted into a d-dimensional vector by a dedicated RSSI embedding network. Similarly, LQI values, RecentPDR values, and RecentETX values ​​are converted into d-dimensional v_LQI, v_PDR, and v_ETX vectors, respectively, through their respective (or shared) embedding mechanisms. This dimension d is typically much smaller than the range of possible values ​​for the original parameters and is a pre-set hyperparameter. It's worth noting that the parameters in the embedding layer (the vector values ​​in the lookup table, or the weights and biases in the neural network) are learned and optimized during the training of the intelligent routing decision model using a backpropagation algorithm based on the ultimate optimization objective (e.g., minimizing routing prediction error or maximizing network throughput). This means that the generation of the embedding vector is data-driven, automatically learning how to most effectively represent these link parameters to serve the final routing decision task.

[0027] Specifically, in step S320, low-dimensional spatial information aggregation is performed on the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector to obtain a low-dimensional aggregated coding vector for the link status of the neighboring active node. It should be understood that although these independent link parameter low-dimensional embedded coding vectors each reflect the characteristics of a certain aspect of the link, there are complex nonlinear correlations and dynamic interactions between them. Simply considering them independently or linearly combining them is difficult to fully and accurately characterize the true comprehensive status of the link. For example, a high RSSI value in the case of low LQI and low packet delivery success rate may indicate a strong interference environment rather than a high-quality link. Therefore, a more advanced mechanism is needed to intelligently fuse these multi-source heterogeneous information, capture the synergistic effects and dynamic changes between them, and thus generate a single, high-information coding vector that can uniformly characterize the overall health of the neighboring node's current active link. Therefore, in the technical solution of the present application, low-dimensional spatial information aggregation is further performed on the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmission times low-dimensional embedded coding vector to obtain a low-dimensional aggregated coding vector of the link status of the neighbor active node.

[0028] It's worth noting that low-dimensional spatial information aggregation aims to learn a feature baseline regression encoding vector that represents the "general" or "stable expectation" of link status. Then, by analyzing the dynamic deviations of individual parameters (such as the current RSSI vector) from this baseline—the unique or sudden changes they carry—and utilizing a softmax mechanism to convert these deviations into adaptive weights, the contribution of each parameter to the final aggregated state representation is dynamically adjusted. This mechanism enables the network to automatically determine, based on real-time link conditions, whether RSSI changes or fluctuations in packet delivery success rates are more significant in reflecting the essence of link quality at the current moment, thereby assigning greater influence to key parameters. The ultimate goal is to generate a "low-dimensional aggregate encoding vector of neighboring active node link status" that not only comprehensively considers various link metrics but, more importantly, dynamically reflects the relative importance of these metrics in the current specific network environment and their interactions. In other words, through this aggregation approach based on dynamic compensation of feature baselines, the resulting low-dimensional aggregate encoding vector of neighboring active node link status more accurately captures the complex dynamics of wireless links than any single parameter or simple weighted combination. It effectively filters out noise, highlights key changes, and adaptively adjusts the emphasis on different link parameters based on the environment. This means that the subsequent weighted regression decoding based on this aggregated coding vector (used to calculate W1 through W4 in the comprehensive link cost) will be more reliable and intelligent. Ultimately, this will directly translate into more optimized routing decisions, reduce unnecessary route switches, lower communication latency, and increase data transmission success rates, thereby comprehensively improving the overall performance, stability, and user experience of smart lighting control systems based on wireless networking technology.

[0029] Figure 4 The flowchart of the intelligent lighting control method based on wireless networking technology according to the embodiment of the present application is to perform fine-grained enhancement of the temperature time series pattern feature on the temperature time series feature coding vector to obtain the temperature time series enhanced feature coding vector. Figure 4As shown, according to the intelligent lighting control method based on wireless networking technology of the embodiment of the present application, step S320 includes: S321, performing feature baseline learning on the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector and the recent expected number of transmissions low-dimensional embedded coding vector to obtain a neighbor data packet coding feature baseline regression coding vector; S322, respectively calculating the dynamic compensation information of the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector and the recent expected number of transmissions low-dimensional embedded coding vector relative to the neighbor data packet coding feature baseline regression coding vector to obtain a set of neighbor data packet coding feature dynamic compensation weight factors; S323, based on the set of neighbor data packet coding feature dynamic compensation weight factors and the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector and the recent expected number of transmissions low-dimensional embedded coding vector, dynamically compensate the neighbor data packet coding feature baseline regression coding vector to obtain the neighbor active node link state low-dimensional aggregation coding vector.

[0030] More specifically, in step S321, feature baseline learning is performed on the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector to obtain a neighbor data packet coding feature baseline regression coding vector, which is expressed by the formula: , , in, The set of neighbor data packet encoding feature vectors consisting of RSSI low-dimensional embedded coding vector, LQI low-dimensional embedded coding vector, recent packet delivery success rate low-dimensional embedded coding vector and recent expected number of transmission times low-dimensional embedded coding vector, are the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector and the recent expected number of transmission times low-dimensional embedded coding vector, and the neighbor data packet coding feature vector set. vectors, and are the learnable parameter matrix and the learnable bias vector, respectively. for activation function, Baseline regression encoding vector for neighbor packet encoding features.

[0031] It's understandable that wireless link states are highly dynamic and complex. Directly processing these real-time, potentially noisy parameter embedding vectors makes it difficult to capture their inherent stable structure and long-term trends. Feature baseline learning aims to extract a relatively stable and representative encoding representation of the "average state" or "expected pattern" from these variable input features. This baseline can be considered an abstract summary of the network's comprehensive performance of these four link parameters under "normal" or "general" conditions. In complex wireless environments, the instantaneous values ​​of various parameters may fluctuate significantly, but a robust routing decision system must be able to distinguish between temporary perturbations and signals of fundamental changes in link quality. Therefore, establishing such a feature baseline is a prerequisite for subsequent dynamic compensation and accurate evaluation. This neighbor packet encoding feature baseline regression code vector is not simply an average of the input parameters. Instead, it is learned from historical data or a current batch of data via a specialized feature baseline learning network. This network aims to capture the long-standing, relatively stable statistical relationships and core features between these four low-dimensional embedding vectors. For example, in most cases, higher RSSI and LQI values ​​are associated with higher packet delivery success rates and lower expected transmission times. This general pattern can be encoded into the feature baseline. By learning such a baseline, the system can establish an "expected model" of link quality, providing a basis for subsequent judgment of whether real-time changes in specific parameters are significant or deviate from the norm.

[0032] Accordingly, according to an embodiment of the present application, step S322 calculates the dynamic compensation information of the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector relative to the neighbor data packet coding feature baseline regression coding vector to obtain a set of neighbor data packet coding feature dynamic compensation weight factors, including: calculating the characteristic dynamic compensation factors of the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector relative to the neighbor data packet coding feature baseline regression coding vector to obtain a set of neighbor data packet coding feature dynamic compensation factors; and performing regularization processing based on the Softmax activation function on the set of neighbor data packet coding feature dynamic compensation factors to obtain a set of neighbor data packet coding feature dynamic compensation weight factors.

[0033] More specifically, the characteristic dynamic compensation factors of the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector relative to the neighbor data packet coding feature baseline regression coding vector are calculated respectively to obtain a set of neighbor data packet coding feature dynamic compensation factors, which are expressed as follows: , in, is the two-norm of the vector, is the point product by position, for function, The first in the set of dynamic compensation factors for neighbor data packet encoding characteristics Dynamic compensation factor for the encoding characteristics of neighboring data packets.

[0034] It should be understood that the baseline regression encoding vector of neighboring packet encoding features only provides a macroscopic, average description of the link state. Actual link quality is dynamic, and the real-time performance of individual parameters may deviate significantly from this baseline. To capture these critical, real-time, and personalized changes in link characteristics, it is necessary to quantify the "difference" or "degree of deviation" between each parameter embedding vector and the baseline vector. These differences contain unique and important information about the current link state that cannot be fully captured by the baseline itself. Therefore, a dynamic compensation factor is calculated for each specific parameter relative to the baseline level. This factor does not simply represent the original parameter value, but rather characterizes the "novelty," "importance," or "difference in contribution" of the parameter's current state relative to the "conventional expectation" (i.e., the baseline). For example, if the current RSSI embedding vector indicates signal strength far below the average level represented by the baseline, the corresponding compensation factor will reflect the degree of this negative deviation. Using these compensation factors, the system can identify which link parameter or parameters are most significantly different from the "normal" at the current moment, or, in other words, which parameter carries the most critical information about the current link state change. For example, if the compensation factor for the recent packet delivery success rate shows a sharp decline, far exceeding baseline expectations, the system can identify this as a critical negative signal. This refined, dynamic deviation assessment for each parameter makes the subsequent feature aggregation process more intelligent and adaptive, because it no longer simply averages or fixes the weights of each parameter, but can adjust their influence based on their dynamic compensation needs relative to the baseline, thus laying a solid foundation for generating a highly accurate and robust low-dimensional aggregate encoding vector of the link status of neighboring active nodes.

[0035] More specifically, the set of dynamic compensation factors of the neighbor data packet coding features is regularized based on the Softmax activation function to obtain a set of dynamic compensation weight factors of the neighbor data packet coding features, which is expressed as follows: , in, is the preset parameter value, is the exponential function value with the natural constant e as the base, The first in the set of dynamic compensation weight factors for neighbor data packet encoding characteristics The weight factor of each neighbor's packet encoding feature is dynamically compensated.

[0036] It should be understood that although the previous step has quantified the unique deviations of each link parameter (RSSI, LQI, PDR, and ETX) from the feature baseline, these raw dynamic compensation factors have different numerical ranges and signs (positive or negative), making them incapable of being directly used as weights to measure their relative importance. To effectively integrate this deviation information in the subsequent aggregation process and ensure that the network adaptively focuses on the key changes that most impact the current link status, these raw factors need to be converted into a set of normalized importance weights with clear physical meaning. Therefore, a set of dynamic compensation weight factors for neighboring packet encoding features is obtained through regularization. Each weight factor in this set is non-negative, and the sum of all weight factors is unity, giving it the properties of a probability distribution. This means that each weight factor directly reflects the relative contribution or importance of the deviation of its corresponding link parameter among all parameter deviations. This conversion introduces an inherent competition mechanism in the network, whereby features with greater deviations and richer information (for example, a sudden deterioration of a link parameter, resulting in a significant increase in its compensation factor) receive higher weights, thereby exerting greater influence in subsequent feature aggregation.

[0037] More specifically, step S323 dynamically compensates the baseline regression coding vector of the coding feature of the neighbor data packet based on the set of dynamic compensation weight factors of the coding feature of the neighbor data packet and the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector to obtain the low-dimensional aggregate coding vector of the link state of the neighbor active node, which is expressed as follows: , in, A low-dimensional aggregate coding vector for the link state of the neighboring active node.

[0038] It should be understood that although a "commonality" or "stable expectation" representation of link status has been obtained through feature baseline learning, and the "unique deviations" and "relative importance" of each parameter from the baseline have been quantified through dynamic compensation factors and softmax weights, this information is still fragmented. To form a single, comprehensive representation that fully and adaptively reflects the true current link status, these different levels of information must be skillfully integrated. Simple superposition or averaging cannot capture the complex nonlinear relationships between parameters and their dynamic contributions in different scenarios. Therefore, the baseline regression code vector of the neighbor packet coding feature is dynamically compensated using the set of dynamic compensation weight factors of the neighbor packet coding feature and the RSSI low-dimensional embedded coding vector, LQI low-dimensional embedded coding vector, recent packet delivery success rate low-dimensional embedded coding vector, and recent expected transmission count low-dimensional embedded coding vector. This generates a "neighbor active node link status low-dimensional aggregate coding vector" that not only preserves the stable baseline information of the link status, but also keenly captures and incorporates the specificity and dynamics of each original feature (RSSI, LQI, PDR, ETX). This process is not a simple linear summation, but rather a more sophisticated fusion mechanism. It allows the baseline to maintain its stability as a reference while being "corrected" or "enhanced" based on the real-time performance of each parameter and its assigned dynamic weights. In this way, feature deviations that are determined to be more important and indicative by the Softmax weighting factor will have a greater impact on the final aggregated coding vector, ensuring that the aggregated result accurately reflects the true quality of the current link rather than being disturbed by unimportant fluctuations. This comprehensive and dynamic aggregate coding can more accurately assess the overall cost of the link and effectively avoid the suboptimal routing selection caused by partial information or fixed weights in traditional methods. Ultimately, this will directly translate into more efficient and stable routing decisions in the smart lighting system, improving the reliability and responsiveness of control commands, and ensuring that the system can continue to provide high-quality service in complex and changing wireless environments.

[0039] Preferably, here, when evaluating each neighbor data packet encoding feature encoding vector Relative to established feature baseline When the deviation or contribution difference of the neighbor data packet is greater than the expected value, the encoding feature vector of the neighbor data packet is To feature baseline With smooth mapping, each neighbor data packet encoding feature encoding vector The calculated dynamic compensation weight factor of neighbor data packet encoding characteristics is generalizable.

[0040] Therefore, we first encode the feature encoding vector for the neighbor data packets To feature baseline The deviation mapping is used to obtain the displacement gradient, that is: , Then, the displacement gradient vector Constructing a metric correlation change matrix , where the displacement gradient vector is a column vector.

[0041] That is, if the above encoding vector based on the neighbor data packet encoding feature is To feature baseline The differential term of the distance-related mapping is used as a parameterized mapping, then it will guide the mapping state through the displacement gradient, thereby, by solving the displacement gradient vector , and construct the metric correlation change matrix under symmetry To ensure no deformation, the metric can be used to associate the change matrix To determine the boot stability.

[0042] Therefore, the metric correlation change matrix is ​​calculated The Frobenius norm of , that is, the low-rank stability characterization, is used as a stability factor to calculate the characteristic dynamic compensation factor, namely: , Thus, the neighbor data packet encoding feature encoding vector is realized To feature baseline Guided by the smooth mapping, small noise or abnormal disturbances are amplified, affecting the generalization accuracy of the dynamic compensation weight factor of the neighbor data packet encoding characteristics.

[0043] Specifically, in step S330, the low-dimensional aggregated coding vector of the neighboring active node link status is subjected to weighted regression decoding to obtain the weight vector. It should be understood that while the aforementioned steps have successfully refined and compressed multi-dimensional link quality information (RSSI, LQI, PDR, ETX) into a highly abstract and information-rich low-dimensional aggregated coding vector of the neighboring active node link status, this vector itself is not directly usable as a weight for link cost calculation. Therefore, a mechanism is required to decode this highly comprehensive low-dimensional aggregated coding vector of the neighboring active node link status and regress these specific weights, so that these weights can dynamically and adaptively reflect the true state of the current link and the importance of each component. Therefore, in this embodiment of the present application, the low-dimensional aggregated coding vector of the neighboring active node link status is passed through a decoder-based weight regressor to obtain the weight vector. Based on the comprehensive link status of the current neighboring active node, the relative importance of each link quality indicator (RSSI, LQI, Recent PDR, Recent ETX) in calculating the comprehensive link cost is intelligently determined. By inputting a low-dimensional aggregated code vector of neighboring active node link states into a decoder-based weight regressor, the system learns and outputs a dynamic weight vector (W1, W2, W3, W4). This regressor, essentially a trained neural network model, captures the complex patterns inherent in the aggregated code vector and maps them to a weight combination that most accurately reflects the current link characteristics. For example, in environments with severe signal interference, the system may assign higher weights to RecentPDR and RecentETX to emphasize data reliability and efficiency; whereas in environments with clear signals and minimal interference, the weights of RSSI and LQI may be increased accordingly. By dynamically generating the weight vector, the system overcomes the limitations of fixed or empirical weights in traditional routing protocols, enabling link cost calculation to respond in real time to complex changes in the wireless environment. This means that the resulting calculated comprehensive link cost more accurately reflects the true quality of a link, guiding the routing protocol in selecting the optimal next-hop node. This dynamic and intelligent weight distribution mechanism significantly improves the reliability, real-time nature of data transmission in smart lighting control systems and the overall robustness of the network. It ensures that control commands can be delivered efficiently and stably in a variety of complex and changing deployment environments, thereby providing users with a smoother and more reliable smart lighting experience.

[0044] Specifically, in step S400, when the first node receives a route announcement from the i-th neighbor node, the first node calculates the total link cost to reach the target gateway via the i-th neighbor node. It should be understood that in a wireless networking environment, data packets typically need to traverse multiple intermediate hops to travel from the source node to the target gateway. Simply knowing the link quality (i.e., the comprehensive link cost) from the first node to its direct neighbors is insufficient, as a seemingly high-quality direct link may connect to a neighbor that is far away from the gateway or has a poor path quality to the gateway. Wireless network environments are complex and volatile, and link quality fluctuates in real time due to interference, node mobility, or load changes. Therefore, nodes cannot rely on static or outdated routing information. Route announcements are a mechanism by which neighbor nodes proactively share their path cost information to the target gateway. Therefore, when the first node receives a route announcement from the i-th neighbor node, the first node calculates the total link cost to reach the target gateway via the i-th neighbor node. By combining the direct link cost from the first node to the declaring neighbor with the cost to the gateway announced by the neighbor, the first node can comprehensively assess the overall cost of forwarding data to the gateway via this neighbor. This is a key step in implementing adaptive, distributed routing decisions, aiming to continuously optimize data transmission paths and ensure that control instructions reach the target gateway with the lowest overall cost (usually translating to the highest reliability and efficiency). By calculating the total link cost from the i-th neighbor node to the target gateway, end-to-end path optimization is achieved, ensuring efficient and reliable data transmission. This allows the first node to select the optimal next hop from a global perspective, rather than based solely on local optima.

[0045] More specifically, in one specific example of the present application, the first node calculates the total link cost from its first node to the target gateway via the i-th neighbor node, including calculating the total link cost from the first node to the target gateway via the i-th neighbor node based on the combined link cost from the first node to each neighbor node. Specifically, first, at the level of overall network communication and information exchange, each node in the network (including the first node and its neighbor nodes) runs a protocol that supports the exchange of routing information. The i-th neighbor node periodically or when its routing information changes significantly, broadcasts or multicasts a route announcement message to other nodes within one hop (including the first node). This route announcement message crucially contains the i-th neighbor node's currently known cumulative link cost to the target gateway. This value is calculated by the i-th neighbor node based on its own link cost to the next hop and the cost to the gateway announced by the next hop. Second, at the level of the first node receiving and processing the route announcement, after successfully receiving and parsing the route announcement from the i-th neighbor node, the first node extracts the link cost from the i-th neighbor node to the target gateway. At the same time, the first node itself calculates and maintains the direct link cost between itself and its i-th neighbor node based on the combined link costs from the first node to each neighbor node, using the aforementioned complex link quality assessment mechanism (involving RSSI, LQI, recent packet delivery success rate, recent expected number of transmissions, and a dynamic weight vector). Furthermore, to calculate the total link cost, the first node sums these two cost values. This sum represents the expected total cost of this path if the first node selects the i-th neighbor node as the next hop to forward the packet to the destination gateway. Finally, to make routing decisions, this calculated total link cost value is used by the first node to update its routing table or candidate path list. It is compared with the first node's currently known best total link cost to the destination gateway, as well as with the path costs provided by other potential neighbor nodes, providing a key quantitative basis for subsequent route selection (i.e., deciding whether to select the i-th neighbor node as the new next hop). This cost calculation method, based on a combination of real-time announcements and local evaluation, is a core component of the distributed dynamic routing algorithm, ensuring that the network can autonomously adapt to changes in topology and link quality.

[0046] Specifically, in step S500, in response to the current total link cost to reach the target gateway being greater than the total link cost to reach the target gateway through the i-th neighbor node, the first node sets the i-th neighbor node as the next-hop node ID. It should be understood that in the wireless networking environment of intelligent lighting control, the dynamics and complexity of wireless links require that routing decisions must be highly adaptive. When the first node receives a routing announcement from the i-th neighbor node and calculates the potential total link cost to reach the target gateway through the neighbor, it compares it with the currently known total link cost to reach the target gateway. The core reason for this is that this is a key decision point for achieving path optimization and network self-healing. This comparison mechanism enables nodes to continuously evaluate and identify whether there are alternative paths that are better than the current path, thereby avoiding suboptimal routing selection due to link quality degradation or the emergence of new paths.

[0047] In particular, in the technical solution of the present application, in response to the total link cost currently reaching the target gateway being less than or equal to the total link cost reaching the target gateway through the i-th neighbor node, the i-th neighbor node is discarded as the next-hop node ID. In other words, if the new path cost through the i-th neighbor node is lower than the current path cost, the next hop is immediately switched to the i-th neighbor node, aiming to utilize better link resources and improve the efficiency and reliability of data transmission. Conversely, if the new path cost is not better than or equal to the current path cost, the existing next hop is maintained unchanged, which avoids unnecessary routing oscillations and resource consumption, ensures routing stability, and prevents the selection of inferior paths.

[0048] In summary, according to the embodiment of the present application, the intelligent lighting control method based on wireless networking technology is explained, which is not limited to a single signal strength or LQI, but allows network nodes to actively and continuously perceive the two-way communication status with neighboring nodes. The node obtains the RSSI and LQI with timestamps from the neighbor data packet, and actively sends packets to monitor ACK, counts the number of successful ACKs and retransmissions, and calculates the recent packet delivery success rate and the expected number of transmissions. Based on this multi-dimensional dynamic parameter, the comprehensive link cost to each neighbor is calculated. When making routing decisions, after receiving the neighbor route announcement, the node combines its own comprehensive link cost to the neighbor and the gateway cost announced by the neighbor to calculate the total cost of the potential complete path and compare it with the current optimal path. If the new path is better, the next hop is updated. This refined, multi-source real-time data fusion link cost evaluation and dynamic route update can accurately grasp the true quality of the link, avoid suboptimal routing, and ensure that control instructions are reliable and timely.

[0049] While various embodiments of the present disclosure have been described above, the above descriptions are illustrative, non-exhaustive, and not intended to be limiting of the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is selected to best explain the principles of the embodiments, their practical applications, or improvements to existing technologies, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. An intelligent lighting control method based on wireless networking technology, characterized in that: include: The first node receives a neighbor data packet from each of its neighbor nodes, and obtains a time-stamped RSSI value and a time-stamped LQI value of each neighbor node from the neighbor data packet; In the process of the first node sending a data packet to each of its neighboring nodes and expecting an ACK, counting the number of successful ACK receptions and the number of retransmissions, and calculating the recent packet delivery success rate and the recent expected number of transmissions from the first node to each of the neighboring nodes based on the number of successful ACK receptions and the number of retransmissions; Calculate, based on the timestamp RSSI value and the timestamp LQI value of each neighboring node and the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node, a comprehensive link cost value from the first node to each neighboring node; When the first node receives a routing announcement from an i-th neighbor node, the first node calculates a total link cost of reaching a target gateway through the i-th neighbor node; In response to the current total link cost to reach the target gateway being greater than the total link cost to reach the target gateway through the i-th neighbor node, the first node sets the i-th neighbor node as the next-hop node ID.

2. The intelligent lighting control method based on wireless networking technology according to claim 1, characterized in that: The neighbor data packets include hello packets, common data packets and ACK packets.

3. The intelligent lighting control method based on wireless networking technology according to claim 2, characterized in that: Calculating a recent packet delivery success rate and a recent expected number of transmissions from the first node to each neighboring node, including: calculating a recent packet delivery success rate and a recent expected number of transmissions from the first node to each neighboring node using the following formula, wherein the formula is: in, is the number of times ACK is successfully received, is the number of retransmissions, is the recent package delivery success rate, The expected number of sends in the near future.

4. The intelligent lighting control method based on wireless networking technology according to claim 3, characterized in that: Calculating, based on the timestamped RSSI value and the timestamped LQI value of each neighboring node and the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighboring node, a comprehensive link cost value from the first node to each neighboring node, including: calculating the comprehensive link cost value from the first node to each neighboring node using the following link cost formula, wherein the link cost formula is: in, is the weight vector, RSSI value with timestamp, is the LQI value with timestamp, The recent package delivery success rate, is the expected number of sends in the near future, is the normalization function, is the comprehensive link cost value.

5. The intelligent lighting control method based on wireless networking technology according to claim 4, characterized in that: Determining the weight vector includes the following steps: Performing low-dimensional embedded coding on the timestamp RSSI value and the timestamp LQI value of each neighbor node, as well as the recent packet delivery success rate and the recent expected number of transmissions from the first node to each neighbor node to obtain an RSSI low-dimensional embedded coding vector, an LQI low-dimensional embedded coding vector, a recent packet delivery success rate low-dimensional embedded coding vector, and a recent expected number of transmissions low-dimensional embedded coding vector; Perform low-dimensional spatial information aggregation on the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmission times low-dimensional embedded coding vector to obtain a low-dimensional aggregate coding vector of the neighboring active node link status; The low-dimensional aggregate coding vector of the neighbor active node link state is subjected to weight regression decoding to obtain the weight vector.

6. The intelligent lighting control method based on wireless networking technology according to claim 5, characterized in that: The RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector are aggregated in low-dimensional space to obtain a low-dimensional aggregate coding vector of the link status of the neighboring active node, including: Performing feature baseline learning on the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmission times low-dimensional embedded coding vector to obtain a neighbor data packet coding feature baseline regression coding vector; Calculating dynamic compensation information of the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector relative to the neighbor data packet coding feature baseline regression coding vector to obtain a set of neighbor data packet coding feature dynamic compensation weight factors; Based on the set of dynamic compensation weight factors of the neighbor data packet coding features and the RSSI low-dimensional embedded coding vector, LQI low-dimensional embedded coding vector, recent packet delivery success rate low-dimensional embedded coding vector and recent expected number of transmissions low-dimensional embedded coding vector, the neighbor data packet coding feature baseline regression coding vector is dynamically compensated to obtain the neighbor active node link status low-dimensional aggregate coding vector.

7. The intelligent lighting control method based on wireless networking technology according to claim 6, characterized in that: Calculating dynamic compensation information of the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector relative to the neighbor data packet coding feature baseline regression coding vector to obtain a set of neighbor data packet coding feature dynamic compensation weight factors, including: Calculating the characteristic dynamic compensation factors of the RSSI low-dimensional embedded coding vector, the LQI low-dimensional embedded coding vector, the recent packet delivery success rate low-dimensional embedded coding vector, and the recent expected number of transmissions low-dimensional embedded coding vector relative to the neighbor data packet coding feature baseline regression coding vector to obtain a set of neighbor data packet coding feature dynamic compensation factors; The set of neighbor data packet coding feature dynamic compensation factors is subjected to regularization processing based on a Softmax activation function to obtain a set of neighbor data packet coding feature dynamic compensation weight factors.

8. The intelligent lighting control method based on wireless networking technology according to claim 7, characterized in that: The low-dimensional aggregate coding vector of the link state of the neighboring active node is subjected to weight regression decoding to obtain the weight vector, comprising: passing the low-dimensional aggregate coding vector of the link state of the neighboring active node through a decoder-based weight regressor to obtain the weight vector.

9. The intelligent lighting control method based on wireless networking technology according to claim 8, characterized in that: The first node calculates the total link cost from the first node to the target gateway through the i-th neighbor node, including: calculating the total link cost from the first node to the target gateway through the i-th neighbor node based on the comprehensive link cost value from the first node to each neighbor node.

10. The intelligent lighting control method based on wireless networking technology according to claim 8, characterized in that: In response to the current total link cost to the target gateway being less than or equal to the total link cost to the target gateway through the i-th neighbor node, the i-th neighbor node is discarded as the next-hop node ID.

Images