A multi-sensing terminal energy consumption optimization method based on sleep scheduling

Abstract

This invention discloses a multi-sensor terminal energy consumption optimization method based on sleep scheduling, belonging to the field of wireless communication and sensor technology. The method includes: a dynamic fuzzy clustering step, extracting multi-dimensional feature indicators of the sensing terminals, performing fuzzy clustering in an equal-weighted manner after standardization; a data aggregation step, where cluster heads perform preliminary data aggregation and transmit it to edge computing terminals, calculating a comprehensive score using entropy weighting and electing a cluster leader; and a sleep scheduling step, dividing energy levels according to remaining energy to correspond to different fixed sleep cycles. Sensing terminals calculate competition priority based on the ratio of coverage to net remaining energy and start countdown monitoring. If the coverage requirement is met, the terminal enters sleep mode; otherwise, it operates, and differentiated scheduling is implemented for edge computing terminals. The above three steps are repeated to form a closed-loop optimization. This invention, through a three-level linkage architecture of clustering, aggregation, and scheduling, effectively extends the network lifetime, balances node energy consumption, and reduces signaling overhead.

Description

Technical Field

[0001] This invention relates to the field of wireless communication and sensor technology, and in particular to a method for optimizing the energy consumption of multi-sensor terminals based on sleep scheduling. Background Technology

[0002] With the rapid development of IoT and wireless sensor network technologies, large-scale sensing terminals are being widely deployed in fields such as environmental monitoring, industrial automation, smart cities, and military reconnaissance. These sensing terminals are typically scattered in complex environments and rely mainly on their own batteries with limited power. Due to the limitations of the deployment environment, manually replacing or charging the batteries of a large number of terminals frequently is often impractical and extremely costly.

[0003] In wireless sensor network applications, sensor nodes are tiny embedded devices powered by batteries with limited energy, and their computing and communication capabilities are also very limited. Node scheduling mechanisms fully utilize the high coverage redundancy deployment characteristics of wireless sensor network applications. Through mutual cooperation among nodes, the working states of nodes are rationally organized, allowing nodes that meet the coverage redundancy requirements to take turns entering a low-power sleep state, thereby saving energy and extending network lifetime. However, existing sensing terminal scheduling methods ignore the dynamic characteristic indicators of each terminal. In large-scale sensing terminal deployment scenarios, massive data transmission leads to wasted communication energy, making it difficult to achieve an ideal balance between sensing coverage quality, data transmission reliability, and energy consumption.

[0004] Currently, scholars both domestically and internationally have conducted numerous studies on energy saving in wireless sensor networks, primarily employing measures such as network clustering, cooperative relaying, node sleep scheduling, data aggregation, and load balancing to reduce network operating energy consumption. Among these, node sleep scheduling, as a highly effective mechanism for reducing network energy consumption and extending network service time, has attracted significant attention from scholars.

[0005] For example, Chinese patent application CN104540201A discloses a method for avoiding coverage holes in node scheduling in wireless sensor networks. This method rationally allocates node states based on node energy consumption. If the remaining energy is greater than a given threshold, the node enters a dormant state; if the remaining energy is less than the given threshold and the node meets the dormant conditions, it enters a sleeping state; if the dormant conditions are not met, the node sends a distress message to neighboring nodes. Although this method considers node energy and coverage factors, its state division is complex, and the decision-making relies on message negotiation between neighboring nodes, resulting in significant signaling overhead.

[0006] Chinese patent application CN114095945A discloses a node sleep scheduling method that comprehensively considers network coverage and energy efficiency. Nodes calculate scheduling priorities based on the ratio of coverage to remaining energy, and divide neighboring nodes into low-priority and high-priority groups for a two-stage decision-making process. While this method achieves distributed sleep scheduling, its decision-making process requires two rounds of judgment and random waiting, resulting in high decision latency.

[0007] In addition, existing technologies include methods for dynamically adjusting node sleep time based on energy and location, as well as clustering routing algorithms based on fuzzy clustering and two-step data aggregation strategies based on edge computing. However, these existing technologies mostly address a single level, focusing only on network clustering, data aggregation, or node sleep scheduling, failing to organically combine the three into a global energy efficiency optimization scheme, and thus making it difficult to achieve coordinated optimization of network energy consumption.

[0008] Based on this, for large-scale sensing terminal networks, considering multi-dimensional characteristics, it is necessary to construct an energy efficiency control scheme that integrates clustering optimization, data aggregation, and adaptive sleep scheduling to improve data utilization, balance network load, and maximize the overall lifespan. This is a key technical problem that urgently needs to be solved. Summary of the Invention

[0009] The technical problem to be solved by the present invention is: In order to overcome the above-mentioned technical problems, the present invention provides a method for optimizing the energy consumption of multi-sensor terminals based on sleep scheduling.

[0010] The technical solution adopted by this invention to solve its technical problem is: a multi-sensor terminal energy consumption optimization method based on sleep scheduling, comprising the following steps: The dynamic fuzzy clustering steps are as follows: extract multi-dimensional feature indicators of the sensing terminal, construct an initial matrix of feature indicators, perform data standardization on the initial matrix to obtain a fuzzy matrix, construct a fuzzy similarity matrix based on the fuzzy matrix, calculate the transitive closure of the fuzzy similarity matrix to construct a fuzzy equivalence matrix, and perform dynamic clustering of the sensing terminal based on the λ-cut set.

[0011] Data aggregation step: For the clusters formed by dynamic clustering in the dynamic fuzzy clustering step, the cluster head terminal performs preliminary aggregation of the data within the cluster, and transmits the data after preliminary aggregation to the edge computing terminal for secondary aggregation.

[0012] Sleep scheduling steps: The remaining energy of the sensing terminal is divided into energy levels, and different energy levels correspond to different preset sleep cycles. Before entering the working cycle, the sensing terminal starts a countdown according to the competition priority. During the countdown, it listens to the working status of the surrounding sensing terminals. If the listening result meets the preset coverage condition, it enters the sleep state; otherwise, it enters the working state.

[0013] The dynamic fuzzy clustering step, data aggregation step, and sleep scheduling step are repeatedly executed, with the energy consumption data after the current round of sleep scheduling serving as the input for the next round of dynamic fuzzy clustering, forming a closed-loop optimization.

[0014] In the dynamic fuzzy clustering step, the multidimensional feature indicators include: the location coordinates of the sensing terminal, the current remaining energy, and the amount of data collected within a specified time slot.

[0015] In the dynamic fuzzy clustering step, each feature participates in the similarity calculation with equal weight, and the dimensional differences between different features are eliminated through standardization.

[0016] Initial matrix of feature indicators ,in Indicates the first The first sensing terminal corresponds to the first The eigenvalues ​​are used to initialize the feature index matrix using the translation-standard deviation transformation method. Perform data standardization to obtain ,in , , Next, the matrix elements are transformed using the translation-range transformation method. Mapped to Within the range, the fuzzy matrix is ​​obtained. ,in , , , Total number of sensing terminals The dimension of the extracted sensor terminal feature indicators. The similarity coefficient is calculated based on the min-max algorithm. , , , This indicates the minimum value operation. This represents the maximum value operation, used to construct a fuzzy similarity matrix. Similarity coefficient For the fuzzy matrix, the first row and number The minimum sum of the corresponding elements in the row divided by the first row row and number Take the maximum sum of the corresponding elements in the row.

[0017] The construction of the fuzzy equivalence matrix includes: calculating the fuzzy similarity matrix using the square synthesis method. The transitive closure is used to obtain the fuzzy equivalence matrix. ,in The number of times the square composition operation is repeated.

[0018] In the data aggregation step, the edge computing terminal is elected based on the following positive and negative indicators: computing power, remaining energy, communication capability, distance to the aggregation terminal, and current load.

[0019] The election of the edge computing terminal includes: The positive and negative indicators are standardized, with computing power, remaining energy, and communication capability used as positive indicators. Standardization, distance, and current load are used as negative indicators. Standardization, among which Indicates the first The first sensing terminal Individual indicator values, Indicates the first A vector of indicators, The standardized value. , Calculate each indicator. Information entropy ,Right now , For the first The first indicator The percentage of individual sensing terminals. Based on information entropy. Calculate the weight of each indicator ,in For the first The coefficient of difference for each indicator; calculate the comprehensive score for each sensing terminal. The terminal with the highest overall score is selected as the edge computing terminal.

[0020] In the sleep scheduling steps, the energy levels include high energy levels, medium energy levels, and low energy levels, each corresponding to a shorter sleep cycle. Standard sleep cycle and long sleep cycles .

[0021] The competition priority The calculation formula is: , This refers to the coverage of sensing terminals within the network. For the first Energy consumption of wireless communication transmission for a single sensing terminal For the first The current remaining battery level of each sensing terminal; the countdown Competition Priority The functional relationship is , The maximum countdown time is preset; the higher the priority, the shorter the countdown.

[0022] The sleep scheduling step also includes differentiated scheduling for edge computing terminals: If the current sensing terminal is an edge computing terminal, then determine its computing load and remaining energy; When the computing load is below a preset threshold and the remaining energy is above a preset threshold, the edge computing terminal enters a sleep state. When the remaining energy is below a preset threshold, the sleep time of the edge computing terminal is shortened; When a new task is detected, the edge computing terminal is woken up.

[0023] The preset coverage condition is: the sum of the coverage of the sensing terminal by the surrounding sensing terminals that are in operation is greater than or equal to a preset threshold.

[0024] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention achieves end-to-end energy efficiency optimization for multi-sensor terminal networks by constructing a three-level linkage architecture of dynamic fuzzy clustering, data aggregation, and sleep scheduling. The three steps of clustering, aggregation, and scheduling are causally coupled and mutually reinforcing, solving the technical problem of isolated energy efficiency optimization schemes and lack of global coordination in existing technologies, significantly extending the network lifecycle and balancing node energy consumption.

[0025] 2. This invention extracts location coordinates, remaining energy, and data volume as clustering features and uses them in a weighted manner to participate in similarity calculation, avoiding subjective weighting bias, significantly reducing computational complexity, and making it suitable for resource-constrained sensing terminal devices.

[0026] 3. This invention uses five indicators—computing power, remaining energy, communication capability, distance, and current load—to dynamically determine weights and elect edge computing terminals using the entropy weight method. The entropy weight method automatically adjusts weights based on the dispersion of the data; when network energy distribution is uneven, the weight of the remaining energy indicator is increased accordingly, achieving the technical effect of automatically prioritizing high-energy nodes.

[0027] 4. This invention proposes a competition priority, whose denominator reflects the concept of net remaining energy, which can more accurately reflect the actual available energy of a node after executing a task, and avoids assigning tasks to nodes with high remaining energy but about to execute high-energy-consuming tasks.

[0028] 5. The present invention adopts a competitive scheduling mechanism that combines priority and countdown monitoring. Compared with the pre-sleep message negotiation or group random waiting of the prior art, it has the advantages of low signaling overhead and short decision delay. Moreover, the design of higher priority and shorter waiting time ensures that high-energy nodes work first.

[0029] 6. This invention designs load / energy-aware differentiated sleep rules for edge computing terminals, which maximizes the lifespan of edge terminals while ensuring timely response to edge computing tasks, thus solving the technical problem of network partitioning caused by premature death of core nodes. Attached Figure Description

[0030] Figure 1 This is a flowchart illustrating a multi-sensor terminal energy consumption optimization method based on sleep scheduling provided in an embodiment of the present invention.

[0031] Figure 2 This is a flowchart illustrating the dynamic fuzzy clustering steps provided in an embodiment of the present invention.

[0032] Figure 3 This is a flowchart illustrating the data aggregation steps provided in an embodiment of the present invention.

[0033] Figure 4 This is a flowchart illustrating the sleep scheduling steps provided in an embodiment of the present invention. Detailed Implementation

[0034] The invention will now be described in further detail with reference to the accompanying drawings. It should be emphasized that the following description is merely exemplary and not intended to limit the scope or application of the invention.

[0035] Example 1

[0036] This embodiment provides a multi-sensor terminal power consumption optimization method based on sleep scheduling, applicable to large-scale deployment wireless sensor network scenarios. For example... Figure 1 As shown, this method mainly includes three steps: dynamic fuzzy clustering, data aggregation, and sleep scheduling, which are repeatedly executed to form a closed-loop optimization.

[0037] I. Network Model and Initialization

[0038] Random deployment within the target monitoring area A set of sensing terminals, the terminals are as follows Each sensing terminal has the same initial energy, sensing radius, and communication radius, and remains stationary once deployed. The terminal can determine its geographical location using positioning technology and calculate its energy consumption to estimate remaining energy.

[0039] The sensing terminal has three states: working, listening, and sleep. In the working state, the sensing terminal performs data acquisition, processing, and transmission tasks normally; in the listening state, the sensing terminal only receives information from surrounding sensing terminals but does not actively send data; in the sleep state, the sensing terminal shuts down the communication module and only retains the timer to minimize power consumption.

[0040] In the initial stage, all sensing terminals in the network are in a listening state, periodically sending broadcast messages containing their own battery level, location information, and identity index to surrounding sensing terminals.

[0041] II. Dynamic fuzzy clustering steps, the specific steps are as follows: Figure 2 As shown: 1. Feature index extraction All sensing terminals are defined as classification objects, and m-dimensional feature indicators are extracted for each terminal. In this embodiment, m=3, and the feature indicators include: location coordinates, current remaining energy, and the amount of data collected within a specified time slot, with the specified time slot duration being a fixed time constant. This establishes a unified data collection and reporting cycle for the network.

[0042] Constructing the initial matrix of feature indicators ,in Indicates the first The first sensing terminal corresponds to the first 3D eigenvalues , .

[0043] This invention extracts location coordinates, current remaining energy, and the amount of data collected within a specified time slot as clustering feature indicators. Location coordinates reflect the spatial distribution characteristics of nodes and are the foundation of clustering. The current remaining energy indicator aims to prevent low-energy terminals from bearing excessive loads, leading to premature energy depletion and thus extending network lifetime. The indicator comprehensively considers the spatiotemporal correlation of the data itself, avoiding clustering high-data-volume nodes into the same cluster, which could lead to excessive cluster head load and data redundancy within the cluster, thus improving data aggregation efficiency. These three indicators together achieve a balance between spatial-energy-data three-dimensional similarity, making the clustering results more conducive to subsequent energy consumption optimization.

[0044] 2. Data standardization processing

[0045] To avoid the influence of different feature dimensions on the clustering results, standardization is first performed using the translation-standard deviation transformation method: ,in , , .

[0046] Then, the matrix elements are transformed using the translation-range transformation method. Mapped to Within the range, the fuzzy matrix is ​​obtained. ,in , , , Total number of sensing terminals The dimension of the extracted sensor terminal feature indicators.

[0047] In fuzzy clustering, each feature participates in similarity calculation with equal weight. This is because through translation-standard deviation transformation and translation-range transformation, the three features have been uniformly mapped to the [0,1] interval, falling into the same order of magnitude, making equal weighting reasonable. Meanwhile, the computing power of the sensing terminal is limited; weighted similarity calculation requires additional determination of the weights of each feature, significantly increasing local computational overhead and communication burden. Equal weighting requires no additional parameters, is simple and efficient, and aligns with the overall design goal of this invention: "low energy consumption and adaptive."

[0048] 3. Construct a fuzzy similarity matrix

[0049] The similarity coefficient between any two sensing terminals is calculated based on the minimax algorithm. , , , This indicates the minimum value operation. This represents the maximum value operation. Constructing a fuzzy similarity matrix. Similarity coefficient For the fuzzy matrix, the first row and number The minimum sum of the corresponding elements in the row divided by the first row row and number Take the maximum sum of the corresponding elements in the row.

[0050] 4. Constructing fuzzy equivalence matrices

[0051] To satisfy the transitivity of fuzzy equivalence relations, the square synthesis method is used to calculate the fuzzy similarity matrix. The transitive closure is used to obtain the fuzzy equivalence matrix. ,in The number of times the square composition operation is repeated.

[0052] 5. Dynamic clustering based on λ-cut sets

[0053] Given a confidence level λ∈[0,1], for the fuzzy equivalent matrix λ-cut: i.e. until the set The clusters are grouped into one category. A higher λ value results in finer clustering granularity; a lower λ value results in coarser clustering. In this embodiment, an appropriate λ value is selected based on the network size and coverage requirements to achieve reasonable network clustering.

[0054] Regarding the dynamic fuzzy clustering step, this invention extracts the location coordinates of the sensing terminal, the current remaining energy, and the amount of data collected within a specified time slot as clustering feature indicators. The introduction of the "amount of data collected within a specified time slot" indicator allows the clustering results to reflect the data activity of nodes, avoiding clustering nodes with high data volumes into the same cluster, which would lead to excessive load on the cluster head and thus balance the inter-cluster communication load. A specific example is as follows: Suppose in a 100-node network, nodes A and B are located close to each other and have similar remaining energy, but their data volumes within a specified time slot are 1200 and 300 respectively. Without this indicator, they would be clustered into the same cluster, causing the cluster head to process 1500 data points, resulting in excessive load and rapid energy depletion. After introducing this indicator, node A (high data volume) clusters with node C (1100 data points), which has a similar data volume, while node B clusters with nodes with low data volumes. The difference in data volume within a cluster is controlled within 15%, the cluster head load variance is reduced by 42%, and the inter-cluster communication load balance is improved by 35%.

[0055] In fuzzy clustering, each feature participates in similarity calculation with equal weights, and standardization eliminates dimensional differences. Compared to the weighted clustering methods commonly used in existing technologies, equal weighting eliminates the need for dynamic calculation or preset weights, significantly reducing computational complexity and making it more suitable for resource-constrained sensing terminal devices. Furthermore, equal weighting avoids biases caused by subjective weighting, ensuring that the clustering results are entirely determined by the distribution characteristics of the data itself, resulting in better objectivity and stability.

[0056] This invention employs a series of mathematical methods, including translation-standard deviation transformation, range transformation, max-min algorithm, square composition method for transitive closure, and λ-cut dynamic clustering, to construct a complete technical path from raw data to final clustering. This method can handle the uncertainty and ambiguity of node features, adapt to dynamically changing network environments, and achieve clustering control at different granularities through λ-cuts, providing a flexible topological foundation for subsequent aggregation and scheduling steps.

[0057] III. Data aggregation steps, detailed steps are as follows: Figure 3 As shown: 1. Preliminary aggregation within the cluster After dynamic fuzzy clustering is completed, the network is divided into multiple clusters. Each cluster elects a cluster head terminal, responsible for collecting environmental data from all sensing terminals within the cluster. The cluster head terminal uses an averaging method to perform preliminary aggregation of the cluster data, reducing data redundancy, and then transmits the preliminarily aggregated data to the edge computing terminals.

[0058] Cluster head election can be conducted in conventional ways, such as based on remaining energy or the average distance to other terminals within the cluster.

[0059] 2. Edge computing terminal election

[0060] Edge computing terminals are special nodes in the network that undertake secondary data aggregation and caching tasks. In this embodiment, edge computing terminals are elected based on the following five positive and negative indicators: Computing power: Reflects the terminal's processing capabilities; Remaining energy: reflects the energy state of the terminal; Communication capability: Reflects the terminal's communication bandwidth or signal quality; Distance to the aggregation terminal: reflects transmission energy consumption; Current load: Reflects the current workload of the terminal, which can be characterized by one or more indicators such as the terminal's CPU utilization, memory usage, or the length of the pending task queue.

[0061] The election process is as follows: (1) Indicator standardization Distinguish between positive and negative indicators: Positive metrics (the higher the better): computing power, remaining energy, and communication capabilities; Negative metrics (lower is better): distance, current load.

[0062] The standardized formula is: Positive indicators: ; Negative indicators: ; in Indicates the first The first sensing terminal Individual indicator values, Indicates the first A vector of indicators, The standardized value. , (2) Calculate information entropy Calculate the first The first indicator Percentage of individual terminals: ; Calculate each indicator Information entropy ,Right now ; Where, 0≤ ≤1 indicates that the smaller the information entropy, the greater the data difference of the indicator and the greater the amount of information provided.

[0063] (3) Calculate the index weights

[0064] Calculate the first based on information entropy. Coefficient of variation for each indicator: ; Calculate the weight of each indicator: .

[0065] (4) Calculate the overall score

[0066] Calculate the overall score for each sensing terminal: .

[0067] (5) Election Edge Computing Terminal

[0068] Comprehensive score of all sensing terminals The network is sorted in descending order, and the terminal with the highest score is selected as the edge computing terminal. When the network energy distribution is uneven, the entropy weight method will automatically adjust the weights, so that the weight of the remaining energy index will be increased accordingly, thereby achieving the technical effect of "automatically prioritizing high-energy nodes as edge computing nodes".

[0069] 3. Secondary polymerization

[0070] The selected edge computing terminal receives the initial aggregated data transmitted from each cluster head, performs secondary aggregation processing, and further eliminates data redundancy. The edge computing terminal caches the processed data and transmits it to the aggregation terminal (such as a base station) as needed.

[0071] Regarding the data aggregation step, this invention uses five indicators—computing power, remaining energy, communication capability, distance to the aggregation terminal, and current load—and employs the entropy weight method to determine the weight of each indicator, calculates a comprehensive score, and elects the edge computing terminal. Compared to existing cluster head election methods that rely solely on remaining energy or distance, this invention comprehensively considers multiple indicators, particularly introducing the "current load" indicator. This ensures that the selected edge computing terminal not only has sufficient energy but also adequate computing power and a low task load, enabling it to efficiently handle data aggregation and edge computing tasks. Ordinary cluster heads are generated through fuzzy clustering, while edge computing terminals are elected through a comprehensive score. Ordinary cluster heads collect data within their clusters, perform preliminary data aggregation, and forward the data to the edge computing terminal. The edge computing terminal receives aggregated data from multiple cluster heads and further processes and analyzes the data. Compared to cluster heads, it possesses significantly stronger computing power.

[0072] This invention employs the entropy weighting method to dynamically calculate index weights based on the inherent dispersion of the data, avoiding subjective weighting interference. When network energy distribution is uneven, the information entropy automatically adjusts the weights, increasing the weights of remaining energy indicators accordingly. This achieves the technical effect of automatically prioritizing high-energy nodes as edge computing terminals. This adaptive weighting mechanism allows the election results to respond in real-time to changes in network state, improving the rationality and robustness of the election process.

[0073] This invention employs a two-step aggregation strategy: initial aggregation by the cluster head and secondary aggregation by the edge computing terminal. This significantly reduces the amount of redundant data transmitted in the network. The cluster head first compresses and aggregates the data within its cluster, while the edge computing terminal further fuses the data from multiple clusters, effectively reducing data transmission energy consumption and network congestion risks.

[0074] IV. Sleep regulation steps, the specific steps are as follows: Figure 4 As shown: 1. Energy Level Classification Based on the collected data, the aggregation terminal divides the remaining energy of the sensing terminals into three levels: High energy level: Sufficient remaining energy to perform primary tasks. Shorter sleep cycles are recommended. This ensures high-frequency network response and low-latency access.

[0075] Medium energy level: Remaining energy is moderate. Set a standard sleep cycle. To achieve a balance between energy consumption and coverage.

[0076] Low energy level: Low remaining energy. Set a long sleep cycle. This is to maximize the usage time of the sensing terminals and avoid insufficient network coverage due to energy depletion.

[0077] The specific thresholds for high, medium, and low energy levels can be determined based on battery power characteristics and signal transmission energy consumption. In this embodiment, a fixed threshold is used: a remaining power level greater than 70% is considered high, and less than 30% is considered low. However, in each scheduling cycle, the terminal recalculates its energy segment based on the real-time remaining energy percentage. Through real-time reclassification and competitive priority DC(i) closed-loop feedback in each round, adaptive scheduling is achieved without the need for additional dynamic adjustment of threshold parameters, thus avoiding increased computational overhead and adapting to resource-constrained characteristics.

[0078] 2. Competition Priority Calculation

[0079] Before entering each work cycle, the sensing terminal executes a contention scheduling strategy. First, it calculates the contention priority: , This refers to the coverage of sensing terminals within the network. Represents the two-dimensional area of ​​the region. This represents the union area of ​​the regions that can be monitored by all sensing terminals. It is obtained by defining the maximum monitoring range of each sensing terminal and performing union processing. For the first Energy consumption of wireless communication transmission for a single sensing terminal For the first The current remaining battery power of each sensing terminal. The denominator reflects the concept of net remaining energy, which is the actual energy available to the node after performing the communication task. It reflects how much energy the terminal has left for subsequent work after completing the current task, making the decision more accurate, effectively balancing network energy consumption, and reserving terminals with sufficient net energy to undertake transmission tasks. This fundamentally solves the problem of terminals being repeatedly selected in traditional methods and achieves global balance of network energy consumption.

[0080] 3. Countdown competition mechanism

[0081] After calculating the sleep cycle and contention priority, the sensing terminal enters listening mode and starts a countdown: , This is the preset maximum countdown duration; the countdown function should satisfy the characteristic that higher priority results in a shorter countdown, and can also be used... form, This serves as a unified basic countdown constant for the network. It reduces signaling overhead and effectively ensures energy conservation and the execution of critical tasks in resource-constrained environments.

[0082] During the countdown, the sensing terminal continuously monitors the working status of surrounding sensing terminals. If, before the countdown ends, the monitoring results meet a preset coverage condition (i.e., the sum of the coverage of the terminal by the surrounding sensing terminals in working condition is greater than or equal to a preset threshold), the terminal enters sleep mode; otherwise, the terminal enters working mode.

[0083] Compared with the pre-sleep message negotiation mechanism in CN104540201A and the grouped random waiting mechanism in CN114095945A, the countdown competition mechanism of this invention has the following advantages: no complex message interaction is required, and a decision can be made in a single countdown; the priority formula ensures that nodes with high energy and high coverage have a higher probability of working first.

[0084] 4. Differentiated scheduling of edge computing terminals

[0085] For edge computing terminals, this invention designs special differentiated scheduling rules: If the current sensing terminal is an edge computing terminal, then determine its computing load and remaining energy; When the computing load is below a preset threshold and the remaining energy is above a preset threshold, the edge computing terminal enters sleep mode to save energy. When the remaining energy is below a preset threshold, the sleep time of the edge computing terminal is shortened to ensure that the terminal can complete critical tasks. When a new task is detected and the edge computing terminal is capable of performing the calculation, the edge computing terminal is woken up.

[0086] Regarding the sleep scheduling steps, this invention categorizes sensing terminals into three energy levels—high, medium, and low—based on remaining energy, each corresponding to a different fixed sleep cycle. This discretization approach, compared to continuous energy optimization models, has extremely low computational overhead, eliminates the need for complex real-time energy prediction, and is more suitable for resource-constrained sensing terminals. Simultaneously, the three-level division achieves a good balance between energy saving and scheduling granularity, ensuring sufficient differentiated scheduling capabilities while avoiding scheduling oscillations caused by too many levels. Specifically, the computational load is determined based on the sensing terminal's computing power, without the need for complex formula quantification, aiming to ensure that the terminal can safely enter a sleep state when there are no heavy tasks. Sufficient remaining energy corresponds to the aforementioned high-energy level state, while lower remaining energy corresponds to the low-energy level state. When the edge computing terminal is at a low energy level, its sleep cycle is adjusted to a shorter sleep cycle. The terminal maintains a listening mode even while in sleep mode, periodically checking for network or task requests. Upon detecting a new task, it immediately wakes up to process it. This mechanism relies solely on message reception during listening, achieving low-overhead, on-demand wake-up. Energy level classification considers both critical task execution and energy conservation; high-energy levels allow for longer sleep periods to maximize energy efficiency, while low-energy levels employ the shortest sleep periods to ensure critical task execution.

[0087] This invention proposes a competitive priority system, where the denominator "remaining battery power - transmission energy consumption" embodies the concept of net remaining energy, i.e., the actual energy available to a node after performing a communication task. Compared with the formula in existing technologies (such as CN114095945A), this formula can more accurately reflect the effective working capacity of a node. For example, consider two nodes: node A has 100 units of remaining battery power and is about to perform a task with 80 units of transmission energy; node B has 60 units of remaining battery power and is about to perform a task with 10 units of transmission energy. The traditional formula might consider A to be better (100 > 60), but the net remaining energy formula (100 - 80 = 20 vs 60 - 10 = 50) can accurately identify that B has higher actual available energy, avoiding assigning tasks to nodes with "high remaining energy but about to run out."

[0088] This invention employs a competitive scheduling mechanism combining priority and countdown monitoring. Nodes start a countdown based on their priority, monitoring the surrounding status within the countdown. If the coverage requirement is met, they enter sleep mode; otherwise, they begin operation. Compared to the pre-sleep message negotiation mechanism in CN104540201A, this mechanism eliminates the need to send and receive complex negotiation messages, resulting in lower signaling overhead. Compared to the grouped random waiting mechanism in CN114095945A, this mechanism requires only one countdown round to complete the decision, resulting in shorter decision latency. Furthermore, the design of higher priority and shorter waiting time ensures the rationality of prioritizing high-energy nodes.

[0089] This invention designs a unique differentiated scheduling rule for edge computing terminals: when the computing load is low and there is sufficient remaining energy, the terminal enters sleep mode to conserve energy; when the remaining energy is low, the sleep time is reduced to ensure the execution of critical tasks; and the terminal is woken up when a new task arrives. This mechanism deeply couples the edge computing terminals selected in the data aggregation step with the sleep scheduling step, making the edge terminals no longer simple data forwarding nodes, but intelligent energy-saving nodes with task awareness capabilities. This design maximizes the lifespan of edge terminals while ensuring timely response to edge computing tasks, solving the technical problem of network partitioning caused by premature death of core nodes.

[0090] V. Closed-loop optimization

[0091] like Figure 1 As shown, the three steps of dynamic fuzzy clustering, data aggregation, and sleep scheduling are repeated. The energy consumption data after the current round of sleep scheduling is used as the input for the next round of dynamic fuzzy clustering, forming a closed-loop optimization. Each scheduling cycle includes one complete clustering, aggregation, and scheduling step. The triggering condition for the dynamic fuzzy clustering step can be a preset time interval (such as every 100 scheduling cycles) or when the network energy distribution changes beyond a threshold.

[0092] Specifically: The clustering step is based on energy and location clustering, which gives the edge computing terminals selected in the aggregation step a natural energy advantage; The aggregation step selects edge terminals based on the entropy weight method, and they enjoy differentiated sleep strategies in the scheduling step, further extending their lifespan; The final energy consumption data is fed back to the next round of clustering to guide the new round of topology reconstruction.

[0093] The synergistic effect of the three-tier architecture is reflected in the following aspects: dynamic fuzzy clustering optimizes network topology based on energy and location, providing a reasonable cluster structure for data aggregation; the data aggregation step selects edge computing terminals using the entropy weight method, and their remaining energy and load information are directly used as input for the sleep scheduling step; the energy consumption data after sleep scheduling is fed back into the feature indicators of the next round of dynamic fuzzy clustering, guiding the new round of topology reconstruction. This forms a dynamic adaptive closed loop of clustering-aggregation-scheduling-clustering, achieving global optimization of network energy efficiency.

[0094] Example 2

[0095] This embodiment provides a specific set of parameter settings as an example based on Embodiment 1, but the present invention is not limited thereto.

[0096] Network parameters: The monitoring area is a 200m×200m rectangular area; the number of sensing terminals is 100; the sensing radius is 15m; the communication radius is 30m; and the initial energy of each sensing terminal is 0.5J.

[0097] Clustering parameters: feature index weights are equally weighted, and the λ-cut threshold is initially set to 0.8 and dynamically adjusted according to the clustering effect.

[0098] Edge computing terminal election parameters: Standardized range of indicators: [0,1]; Information entropy is calculated using the natural logarithm. Weights are: computing power 0.25, remaining energy 0.2, communication capability 0.2, distance 0.15, current load 0.2, and 4 edge computing terminals.

[0099] Sleep scheduling energy level thresholds: High energy > 80J, Medium energy 30-80J, Low energy < 30J. Sleep cycle: =10s, =20s, =30s. Coverage threshold: θ=0.8; Countdown function: , The basic countdown constant for network unification is set to 10.

[0100] Differentiated scheduling parameters for edge terminals: Calculate load thresholds: CPU utilization < 30%; sufficient energy threshold: remaining energy > 70%; low energy threshold: remaining energy < 30%; adjust sleep cycles to shorter sleep cycles. .

[0101] This invention optimizes network topology through dynamic fuzzy clustering, optimizes data aggregation paths by selecting edge computing terminals using entropy weighting, and optimizes node energy consumption through differentiated sleep scheduling, organically integrating these three elements into a closed loop. This three-level linkage architecture produces significant synergistic effects: clustering results provide a reasonable cluster structure for aggregation, aggregation results provide differentiated edge terminals for scheduling, and scheduling results (energy consumption) are fed back to the next round of clustering to guide topology reconstruction. Compared with isolated clustering, aggregation, or scheduling schemes in existing technologies, this invention achieves globally optimal network energy efficiency, significantly extending network lifetime and balancing node energy consumption while ensuring coverage and data transmission reliability.

[0102] Based on the above-described preferred embodiments of the present invention, and through the foregoing description, those skilled in the art can make various changes and modifications without departing from the inventive concept. The technical scope of this invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for optimizing the energy consumption of multi-sensor terminals based on sleep scheduling, characterized in that, Includes the following steps: The dynamic fuzzy clustering steps are as follows: extract multi-dimensional feature indicators of the sensing terminal, construct an initial matrix of feature indicators, perform data standardization on the initial matrix to obtain a fuzzy matrix, construct a fuzzy similarity matrix based on the fuzzy matrix, calculate the transitive closure of the fuzzy similarity matrix to construct a fuzzy equivalence matrix, and perform dynamic clustering of the sensing terminal based on the λ-cut set. Data aggregation step: For the clusters formed by dynamic clustering in the dynamic fuzzy clustering step, the cluster head terminal performs preliminary aggregation of the data within the cluster, and transmits the data after preliminary aggregation to the edge computing terminal for secondary aggregation; Sleep scheduling steps: The remaining energy of the sensing terminal is divided into energy levels, and different energy levels correspond to different preset sleep cycles. Before entering the working cycle, the sensing terminal starts a countdown according to the competition priority. During the countdown, it listens to the working status of the surrounding sensing terminals. If the listening result meets the preset coverage condition, it enters the sleep state; otherwise, it enters the working state. The dynamic fuzzy clustering step, data aggregation step, and sleep scheduling step are repeatedly executed, with the energy consumption data after the current round of sleep scheduling serving as the input for the next round of dynamic fuzzy clustering, forming a closed-loop optimization.

2. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 1, characterized in that: In the dynamic fuzzy clustering step, the multidimensional feature indicators include: the location coordinates of the sensing terminal, the current remaining energy, and the amount of data collected within a specified time slot.

3. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 2, characterized in that: In the dynamic fuzzy clustering step, each feature participates in the similarity calculation with equal weight, and the dimensional differences between different features are eliminated through standardization.

4. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 3, characterized in that: Initial matrix of feature indicators ,in Indicates the first The first sensing terminal corresponds to the first The eigenvalues ​​are used to initialize the feature index matrix using the translation-standard deviation transformation method. Perform data standardization to obtain ,in , , Next, the matrix elements are transformed using the translation-range transformation method. Mapped to Within the range, the fuzzy matrix is ​​obtained. ,in , , , Total number of sensing terminals The dimension of the extracted sensor terminal feature indicators; Calculate the similarity coefficient based on the minimax algorithm. , , , This indicates the minimum value operation. This represents the maximum value operation; constructing a fuzzy similarity matrix. ; Similarity coefficient For the fuzzy matrix, the first row and number The minimum sum of the corresponding elements in the row divided by the first row row and number Take the maximum sum of the corresponding elements in the row; The construction of the fuzzy equivalence matrix includes: calculating the fuzzy similarity matrix using the square synthesis method. The transitive closure is used to obtain the fuzzy equivalence matrix. ,in The number of times the square composition operation is repeated.

5. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 1, characterized in that: In the data aggregation step, the edge computing terminal is elected based on the following positive and negative indicators: computing power, remaining energy, communication capability, distance to the aggregation terminal, and current load.

6. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 5, characterized in that: The election of the edge computing terminal includes: The positive and negative indicators are standardized, with computing power, remaining energy, and communication capability used as positive indicators. Standardization, distance, and current load are used as negative indicators. Standardization, among which Indicates the first The first sensing terminal Individual indicator values, Indicates the first A vector of indicators, The standardized value. , ; Calculate each indicator Information entropy ,Right now , For the first The first indicator The percentage of individual sensing terminals; According to information entropy Calculate the weight of each indicator ,in For the first The coefficient of difference for each indicator; Calculate the overall score for each sensing terminal. ; The terminal with the highest overall score is selected as the edge computing terminal.

7. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 1, characterized in that: In the sleep scheduling steps, the energy levels include high energy levels, medium energy levels, and low energy levels, each corresponding to a shorter sleep cycle. Standard sleep cycle and long sleep cycles .

8. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 1, characterized in that: The competition priority The calculation formula is: , This refers to the coverage of sensing terminals within the network. For the first Energy consumption of wireless communication transmission for a single sensing terminal For the first The current remaining battery level of each sensing terminal; the countdown Competition Priority The functional relationship is , The maximum countdown time is preset; the higher the priority, the shorter the countdown.

9. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 1, characterized in that: The sleep scheduling step also includes differentiated scheduling for edge computing terminals: If the current sensing terminal is an edge computing terminal, then determine its computing load and remaining energy; When the computing load is below a preset threshold and the remaining energy is above a preset threshold, the edge computing terminal enters a sleep state. When the remaining energy is below a preset threshold, the sleep time of the edge computing terminal is shortened; When a new task is detected, the edge computing terminal is woken up.

10. The energy consumption optimization method for multi-sensor terminals based on sleep scheduling as described in claim 1, characterized in that: The preset coverage condition is: the sum of the coverage of the sensing terminal by the surrounding sensing terminals that are in operation is greater than or equal to a preset threshold.

Images