WO2022100191A1 - 一种分布式传感器网络数据融合方法及系统 - Google Patents

一种分布式传感器网络数据融合方法及系统 Download PDF

Info

Publication number
WO2022100191A1
WO2022100191A1 PCT/CN2021/113733 CN2021113733W WO2022100191A1 WO 2022100191 A1 WO2022100191 A1 WO 2022100191A1 CN 2021113733 W CN2021113733 W CN 2021113733W WO 2022100191 A1 WO2022100191 A1 WO 2022100191A1
Authority
WO
WIPO (PCT)
Prior art keywords
cluster
nodes
data
fusion
node
Prior art date
Application number
PCT/CN2021/113733
Other languages
English (en)
French (fr)
Inventor
王凯琢
于洁
Original Assignee
北京市天元网络技术股份有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京市天元网络技术股份有限公司 filed Critical 北京市天元网络技术股份有限公司
Publication of WO2022100191A1 publication Critical patent/WO2022100191A1/zh

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W40/00—Communication routing or communication path finding
    • H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
    • H04W40/04—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources
    • H04W40/10—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources based on available power or energy
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W84/00—Network topologies
    • H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Definitions

  • the present application relates to the technical field of communication engineering analysis and verification, and in particular, to a distributed sensor network data fusion method and system.
  • WSN Wireless Sensor Network
  • the present application provides a distributed sensor network data fusion method and system, which are used to solve the defect in the prior art that the energy efficiency of sensor nodes is not systematically optimized for wireless sensor networks.
  • the present application provides a distributed sensor network data fusion method, including:
  • data fusion is performed on the several in-cluster nodes and the several cluster head nodes respectively to obtain cluster head fusion data and cluster head fusion data;
  • the plurality of intra-cluster nodes transmit the intra-cluster fusion data to the corresponding cluster head nodes, and the plurality of cluster head nodes transmit the cluster head fusion data to the sink node.
  • cluster head nodes and several intra-cluster nodes corresponding to each cluster head node are obtained, specifically including:
  • a data fusion algorithm based on matrix analysis is used to preprocess the cluster head nodes.
  • time series-based data fusion algorithm preprocesses the nodes in the several clusters, specifically including:
  • each segment contains a single data or two adjacent data
  • Two segments that satisfy the preset merging condition and the least merging cost are selected for merging until all segments that satisfy the preset merging condition are merged.
  • the data fusion algorithm based on matrix analysis performs preprocessing on the several cluster head nodes, specifically including:
  • data fusion is performed on the nodes in the clusters and the cluster head nodes respectively to obtain the cluster head fusion data and the cluster head fusion data, which specifically include: :
  • the several intra-cluster nodes and the several cluster head nodes are respectively initialized;
  • each node saves the current remaining energy, performs the next round of cluster selection and cluster formation, and compares the current remaining energy with the next remaining energy. If the remaining energy is less than the product of the current remaining energy and the preset ratio, the cluster selection and clustering are restarted, otherwise the current stable state is maintained.
  • the several intra-cluster nodes and the several cluster head nodes are respectively initialized, specifically including:
  • the plurality of intra-cluster nodes and the plurality of cluster head nodes respectively obtain their corresponding data fusion algorithms
  • Each cluster head node sends a query command to the corresponding intra-cluster node to confirm the status of the corresponding intra-cluster node;
  • the corresponding intra-cluster node returns an acknowledgment message, enters a listening state, and starts a timer;
  • each cluster head node After receiving the confirmation message, each cluster head node performs a low energy consumption mode and waits for the corresponding intra-cluster nodes to perform data transmission.
  • the plurality of intra-cluster nodes transmit the intra-cluster fusion data to the corresponding cluster head node
  • the plurality of cluster head nodes transmit the cluster head fusion data to the sink node, specifically include:
  • Obtain the node closest to the cluster head node calculate the distance to the closest node and the transmission time of the closest node, assign a time sequence to the closest node, and obtain the fusion time of the closest node based on the time sequence;
  • each node in the cluster collects data according to the corresponding time series and fuses the data, and transmits the fused data to the cluster head node;
  • the time interval from the next round of data transmission is calculated according to the current node in the cluster, the fusion time of the node in the current cluster, and the transmission time from the node in the current cluster to the cluster head node.
  • the present application also provides a distributed sensor network data fusion system, including:
  • an acquisition module used for acquiring several sensor nodes and sink nodes in the distributed sensor network node set
  • a first processing module configured to obtain several cluster head nodes and several intra-cluster nodes corresponding to each cluster head node based on a preset node data fusion algorithm
  • the second processing module is configured to perform data fusion on the several in-cluster nodes and the several cluster head nodes respectively based on the distributed cluster fusion algorithm with improved residual energy and distance, to obtain cluster head fusion data and cluster head fusion data ;
  • a third processing module configured to transmit the intra-cluster fusion data to the corresponding cluster head nodes based on a preset fusion mechanism, and the cluster head nodes transmit the cluster head fusion data to the aggregation node.
  • the present application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, the processor implementing the program as described above when the processor executes the program. Describe the steps of the distributed sensor network data fusion method.
  • the present application further provides a non-transitory computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the data fusion method of any of the above-mentioned distributed sensor network data fusion methods is implemented. step.
  • the distributed sensor network data fusion method and system provided by the present application, through a low-energy clustering algorithm based on improved residual energy and distance, the cluster head randomly selects cooperative nodes in each round of data fusion process, and the intermediate nodes cooperate with the cluster head to carry out
  • the protection and fusion of data can effectively reduce the calculation amount and communication amount of nodes, and finally greatly improve the calculation amount, communication amount and fusion accuracy.
  • FIG. 1 is a schematic flowchart of a distributed sensor network data fusion method provided by the present application
  • Fig. 2 is the wireless sensor network structure diagram provided by this application.
  • Fig. 3 is the low-energy consumption clustering method implementation frame diagram that improves residual energy and distance provided by the present application
  • FIG. 4 is a schematic diagram of a low-energy distributed specific flow diagram for improving residual energy and distance provided by the present application
  • FIG. 5 is a schematic diagram of a data transmission process of a member node in a cluster provided by the present application.
  • Fig. 6 is the relational comparison diagram of the number of nodes and FND provided by this application.
  • FIG. 7 is a comparison diagram of the relationship between the number of nodes and LND provided by this application.
  • Fig. 8 is a network life cycle comparison diagram provided by this application.
  • FIG. 9 is a comparison of the relationship between the number of nodes and energy consumption provided by this application.
  • FIG. 10 is a schematic structural diagram of a distributed sensor network data fusion system provided by the present application.
  • FIG. 11 is a schematic structural diagram of an electronic device provided by the present application.
  • This application proposes a low-energy distributed algorithm strategy based on improved residual energy and distance, which advances the data security fusion operation to the data collection node, accumulates data for a certain period of time, performs data security fusion locally, and then transmits it to the data collection node.
  • Cluster head node the cluster head node once again performs distributed and secure fusion of data from different member nodes.
  • the qualitative analysis shows that the distributed data security fusion strategy can reduce the redundancy of data and reduce the amount of data transmission under the premise of ensuring data security, thereby further reducing the computational dimension and communication volume of network nodes.
  • FIG. 1 is a schematic flowchart of a distributed sensor network data fusion method provided by the present application, as shown in FIG. 1 , including:
  • the plurality of intra-cluster nodes transmit the intra-cluster fusion data to the corresponding cluster head nodes, and the plurality of cluster head nodes transmit the cluster head fusion data to the sink node.
  • the distributed clustering fusion algorithm that improves the residual energy and distance adopted in this application takes into account the heavy task of the cluster head node, which is responsible for broadcasting messages to all nodes, and receives member nodes in the cluster after forming a cluster.
  • the cluster head node sends the fused useful information to the base station, so that the base station can monitor the long-distance environmental changes in real time. Triggered by sending a message, the node transforms into a cluster head node after receiving the information and continues to send information to nearby nodes, and the rest of the network nodes in the distributed network will be in a waiting state until they receive information from nearby nodes. Otherwise, join the cluster after receiving the information sent by any cluster head.
  • the key distribution of the nodes will be carried out.
  • the random key pre-distribution method is adopted, which can reduce the probability of the node key being cracked as much as possible. lower, the security protection of data is higher. Therefore, cluster head nodes often consume more energy than non-cluster head nodes. How to reduce the energy consumption of cluster head nodes and balance the energy of nodes in the network is very important to prolong the life cycle of the entire network.
  • the optimal number of clusters is obtained on the basis of analyzing the energy consumed by the network.
  • the calculation amount of the nodes in the cluster should be considered, and on the other hand, the calculation amount of the cluster head node should also be considered.
  • the calculation amount includes arithmetic operations, that is, noise interference.
  • the basic idea of processing, encryption and decryption operations and data fusion operations is that after the clustering ends, the nodes in the cluster begin to sense data and perform fusion operations, and the number of cluster heads is selected to minimize the total energy consumed by the network in each round.
  • All in-cluster nodes will perform the following steps in the process of their intra-cluster fusion: after each node collects data, use the public value of the in-cluster node and multiple private random values to perform interference operations, that is, convert it into a quadratic polynomial. In the whole calculation process, each node encrypts multiple noise interference values and sends them to the cluster head node and the intermediate nodes. At the same time, after receiving the interference processing results sent by other nodes, the cluster head node and the intermediate nodes use the shared key to pair After the numerical value is decrypted, the arithmetic operation is performed to combine the polynomial, and the calculation formula of the optimal number of cluster heads is:
  • the model is based on two assumptions:
  • the energy consumed by the wireless communication sensor node to send and receive lbit information d distance is:
  • E elec is the energy consumed by the transmitting circuit and the receiving circuit, which depends on the digital codec, modulation and filtering of the signal, etc. In this model, the transmit and receive signals are the same.
  • N nodes there are a total of N nodes that are approximately evenly distributed. Many factors need to be considered. Especially when wireless sensor nodes are generally deployed, it is impossible to ensure that each node is evenly distributed, so that the interval and density between nodes are the same. The number of nodes placed on the same area is too different, or some are too dense and some are too loose, which is not only conducive to collecting information, but also not conducive to clustering, routing, and transmission, and also wastes a lot of energy, but usually For example, we will consider nodes to be evenly distributed in an M ⁇ M area.
  • the number of nodes in each cluster is N/k, including a cluster head node and N/k-1 member nodes.
  • the number of bits is l.
  • each cluster head node includes the following parts: receiving the data packets of its member nodes, performing data fusion processing, and sending the data packets after fusion processing to the remote receiver. Then the energy consumed by a cluster head node can be expressed as:
  • the distance between non-cluster head nodes and their respective cluster head nodes is relatively short, so the energy consumption of sending data can use the free space model.
  • the energy consumed by each non-cluster head node includes two parts: receiving (collecting) data packets and sending data packets to the cluster head node, the energy consumed by a non-cluster head node can be expressed as:
  • E non-CH lE elec + l ⁇ fs d 2 toCH
  • d toBS is the distance from the cluster head node to the base station
  • d toCH is the distance from the non-cluster head node to the cluster head node
  • E DA is the energy consumed by data fusion
  • E elec is the energy consumed by the transmitting circuit and the receiving circuit
  • ⁇ fs , ⁇ am p both represent the magnification of signal source enhancement, and two models are applied respectively.
  • E DA 5nJ/bit/signal
  • E elec 50nJ/bit
  • ⁇ fs 10pJ/bit/m 2
  • ⁇ amp 0.0013pJ/bit/m 4
  • d toCH represents the distance from the non-cluster head node to the cluster head node.
  • the energy consumed by all nodes in each cluster when transmitting lbit data in each round consists of two parts: the energy consumed by the cluster head node E CH and the energy consumed by the member nodes E non-CH , therefore, each round of the cycle is obtained.
  • the total energy consumed by the entire network is:
  • E Total l[(2N-k)E elec +NE DA +k ⁇ amp d 4 toBS +(Nk) ⁇ fs d 2 toCH ]
  • the entire monitoring area R is approximately denoted as M 2 , and the distance between the base station and the sensor node is expected to be:
  • This distance expectation mainly depends on the position coordinates (x*, y*) of the base station, (x, y) represents the position coordinates of the sensor node, which is defined by the expectation, and the value of d toBS is equal to E[d toBS ].
  • each cluster is approximately denoted as M 2 /k, then according to the definition of mathematical expectation, the expectation of the square of the distance d toBS from the cluster head node to the member node should be:
  • ⁇ (x,y) is the distribution density of sensor nodes in each cluster. Since the ideal cluster structure should be a circle, it is assumed that this area is a radius of The circle is transformed by the corresponding coordinates to get:
  • the function Since the second derivative f s (2) (k) of the function to k is always positive, the function has a minimum value, which is the optimal number of cluster heads k opt (is a function related to N, M ):
  • the energy E Total consumed by the network is the minimum value.
  • the clustering method commonly used by most network protocols is that once the cluster head node is selected, they actively send the information that they become the cluster head to all nodes. According to the strength of the emission source, the node selects the cluster it wants to join and informs the corresponding cluster head node; or the node chooses which cluster to join according to the distance between itself and the cluster head.
  • the calculation method of the distance is as follows:
  • the problem of cluster head cycle and cluster regrouping is considered as follows: when the cluster head node continues to work for a period of time or its residual energy is lower than a certain value, it announces the disbanding of the cluster, and then re-selects clusters according to the clustering algorithm. Group clusters. We do not want to re-select a cluster after each data transmission (from the node to the cluster head to the base station, counting one round), and frequent cluster selection will increase energy consumption.
  • the cluster head randomly selects cooperative nodes in the process of each round of data fusion, and the intermediate nodes cooperate with the cluster head to perform data protection and fusion, so as to effectively reduce the calculation amount of nodes and communication volume, and finally the calculation volume, communication volume and fusion accuracy are greatly improved.
  • obtaining several cluster head nodes and several intra-cluster nodes corresponding to each cluster head node based on the preset node data fusion algorithm specifically includes:
  • a data fusion algorithm based on matrix analysis is used to preprocess the cluster head nodes.
  • the described time series-based data fusion algorithm preprocesses the nodes in the several clusters, specifically including:
  • each segment contains a single data or two adjacent data
  • Two segments that satisfy the preset merging condition and the least merging cost are selected for merging until all segments that satisfy the preset merging condition are merged.
  • the data fusion algorithm based on matrix analysis performs preprocessing on the several cluster head nodes, specifically including:
  • the network structure is shown in Figure 2, including: the wireless sensor network is composed of a large number of sensor nodes and a sink node (base station node), and the specifications of the sensor nodes are the same; the network is a clustered type structure, the network forms multiple clusters, and the cluster structure will not change once it is determined; each cluster has only one cluster head node, and the cluster contains a large number of member nodes; there is single-hop communication between the member nodes in the cluster and the cluster head node; the base station It only communicates with the cluster head node of each cluster, and does not receive data from member nodes; the cluster head node can obtain the location information of all member nodes in the cluster.
  • each node plays different roles and have different assignments, which will affect the choice of the data fusion algorithm in the proposed hierarchical data fusion strategy.
  • the working status of each node is as follows:
  • the merging cost is mainly determined by the following two factors: one is the error brought about by the merging of the two segments; the other is the sub-time series corresponding to the merged segments the number of data contained;
  • the purpose is to allow nodes to control the size of the time series by themselves.
  • the second is the cluster head node: it is also initialized first, and then enters the low energy consumption mode.
  • the cluster head node is not responsible for collecting data, but only receives the data transmitted from the member nodes in its cluster. When the data arrives, the cluster head node enters the receiving state, accepts and transmits data. Store data. After receiving the data of all nodes in the cluster, perform data fusion. The data fusion here will be used in general energy-based network design. After data fusion processing, it is sent to the base station. The data fusion of the cluster head node is very different from that of the member nodes in the cluster. The member nodes in the cluster only fuse the data collected within a certain period of time, which can be determined according to the storage capacity of the node.
  • the cluster head node is to fuse the data transmitted by its member nodes in the cluster.
  • the cluster generally includes a large number of nodes. Relatively speaking, the workload of the cluster head node is much larger than that of the member nodes in the cluster. There is also much more energy, so the complexity of the data fusion itself and the effect of the fusion must be considered.
  • the data fusion algorithm based on matrix analysis is chosen. On the one hand, its algorithm is relatively simple, with less computation and less time consumption; on the other hand, its fusion accuracy is comparable to the data fusion results of D-S evidence combination.
  • Fusion data including:
  • the several intra-cluster nodes and the several cluster head nodes are respectively initialized;
  • each node saves the current remaining energy, performs the next round of cluster selection and cluster formation, and compares the current remaining energy with the next remaining energy. If the remaining energy is less than the product of the current remaining energy and the preset ratio, the cluster selection and clustering are restarted, otherwise the current stable state is maintained.
  • the several in-cluster nodes and the several cluster head nodes are respectively initialized, specifically including:
  • the plurality of intra-cluster nodes and the plurality of cluster head nodes respectively obtain their corresponding data fusion algorithms
  • Each cluster head node sends a query command to the corresponding intra-cluster node to confirm the status of the corresponding intra-cluster node;
  • the corresponding intra-cluster node returns an acknowledgment message, enters a listening state, and starts a timer;
  • each cluster head node After receiving the confirmation message, each cluster head node performs a low energy consumption mode and waits for the corresponding intra-cluster nodes to perform data transmission.
  • both the cluster head node and the member nodes in the cluster need to complete the initialization work on their nodes:
  • the cluster head node and the member nodes in the cluster use a node flag bit to distinguish their different positions.
  • the flag Si of the cluster head node is set to 1, and the flag bit Si of the member nodes in the cluster is set to 0;
  • the cluster head node sends a query command to the member nodes in the cluster to see if the member nodes are ready;
  • the member nodes in the cluster return a confirmation message, enter the monitoring state, and turn on the timer at the same time;
  • the cluster head node After receiving the confirmation message, the cluster head node enters the low energy consumption mode and waits for the data transmission of the member nodes in the cluster.
  • the member nodes in the cluster they each start monitoring and collecting data, start timing, and store the collected data.
  • the timer reaches the preset time, a command is issued, and the member nodes in the cluster perform data processing on the collected data. Fusion, after the fusion processing is completed, the data results are sent to the cluster head node, and a new round of data collection and timing is started at the same time.
  • the cluster head node In order to save energy, it enters the low energy consumption mode after initialization, until its member nodes in the cluster transmit data, and then starts data fusion. After the fusion processing is completed, the data results are sent to the base station, while continuing to wait and receive clusters Incoming data from member nodes.
  • the base station After receiving the data from the cluster head node, it can be directly transmitted to the required users, or it can be transmitted to the users after decision-making data fusion. Since the base station does not have energy restrictions, it does not need to consider its energy saving. .
  • the low-energy clustering fusion method of improving the remaining energy and distance can not only save the energy consumption of the member nodes in the cluster to send data, but also greatly save the energy consumption of the cluster head node to receive data. Because through the data fusion of the member nodes in the cluster, it is no longer the original data that is sent once every time it is collected. Through fusion, it only needs to be sent once in a period of time, so that the number of times the cluster head node receives data and data fusion is also reduced. , thereby saving the energy consumption of nodes and effectively prolonging the network life cycle.
  • each node saves its remaining energy En_current-1 at the end of the previous round, and then transmits data to the receiver through cluster selection, clustering, and data transmission .
  • En_current ⁇ 0.5E n_current-1 that is, more than half of the remaining energy of the node is consumed in this round, and the cluster needs to be re-selected to prevent the node energy from being too low to make the next round work normally. As long as a node En_current ⁇ 0.5E n_current-1 appears, the cluster is reselected.
  • the plurality of intra-cluster nodes transmit the intra-cluster fusion data to the corresponding cluster head nodes based on the preset fusion mechanism, and the plurality of cluster head nodes transmit the cluster head fusion data to the corresponding cluster head nodes.
  • the aggregation node specifically including:
  • Obtain the node closest to the cluster head node calculate the distance to the closest node and the transmission time of the closest node, assign a time sequence to the closest node, and obtain the fusion time of the closest node based on the time sequence;
  • each node in the cluster collects data according to the corresponding time series and fuses the data, and transmits the fused data to the cluster head node;
  • the time interval from the next round of data transmission is calculated according to the current node in the cluster, the fusion time of the node in the current cluster, and the transmission time from the node in the current cluster to the cluster head node.
  • the cluster head node will either wait for the data packets of the remaining nodes, which will increase the delay; or strictly control the time, if the data packets of the member nodes do not arrive within the specified time, it will be included in the future data fusion. , the data fusion effect will be poor.
  • the time series length of the data collected by the member nodes in the cluster be s i
  • the data fusion time is t i (related to the length of the time series, and can be represented by s i during calculation)
  • the time for data transmission to the cluster head node is T i (related to the transmission distance, that is, the distance d i from the member nodes in the cluster to the cluster head node)
  • the transmission time slot allocated by the cluster head node to each member node is w
  • this transmission time slot w indicates that the cluster head node receives a
  • the time required for data packets transmitted by non-cluster head nodes, only one node is allowed to communicate with one time slot, where i represents the number of member nodes in the cluster, i 1,2,...,k, k represents a cluster Contains the number of member nodes.
  • the time distribution of the member nodes in the cluster from data collection to data transmission to the cluster head node is shown in Figure 5.
  • the time series s x , s 1 , s 2 ,..., s k-1 corresponding to the member nodes in each cluster can be obtained by the following formula.
  • the member nodes in the cluster start to collect data according to their respective time series and perform data fusion, and then transmit to the cluster head node.
  • the above heartbeat mechanism can ensure that the data of the member nodes in the cluster are transmitted to the cluster head node in turn, and the cluster head node can perform data fusion immediately after receiving the data, which not only improves the speed, but also optimizes the effect of data fusion, greatly reducing the It also reduces the energy consumption of data transmission, and also avoids the serious consequences of channel contention and data loss caused by member nodes in the cluster transmitting data to the cluster head node at the same time.
  • each cluster contains n k nodes, that is, a cluster head node and n k -1 cluster member nodes.
  • the member nodes in the cluster and the cluster head are used to fuse the data in the network respectively.
  • Each member node in the cluster can collect multiple data packets within a period of time, and then fuse them according to the data fusion algorithm based on time series to synthesize a data packet. Therefore, there is only one output data packet; the cluster head node receives the data of the member nodes in the cluster and then performs data fusion based on matrix analysis. After adding the time heartbeat mechanism, the data of the member nodes can be sent to the cluster head node in turn.
  • the number of packets to be sent by the network in each round is Compared with the case where no data fusion is used in the network or only in the cluster head node, the energy consumption of the network is obviously reduced.
  • the second is to reduce network congestion.
  • the data fusion of time series is performed first, so that the original need to transmit data multiple times in a period is changed to only one transmission, which effectively disperses the contention and resource waste caused by the separate transmission of data. .
  • the application is simulated through a virtual environment
  • the NS2 tool namely Network Simulator Version 2
  • the NS2 tool is an object-oriented, discrete event-driven network environment simulator, which is mainly used to solve problems in network research.
  • NS2 provides the simulation of TCP, routing, multicast and other protocols on wireless or wired networks, and can completely simulate the entire network environment. It uses a complete set of C++ class libraries to implement most common network protocols and chains.
  • the model of the road layer, using the instances of these classes can build a model of the entire network, and includes detailed implementation details.
  • the hierarchical clustering routing algorithm protocol in order to carry out experiments and performance comparisons, the hierarchical clustering routing algorithm protocol is used, and the source code of the hierarchical clustering routing algorithm protocol is successfully added to NS2, and then the DCHS clustering Algorithm and EDCSA clustering algorithm make corresponding source code modification to LEACH, and finally compare and analyze the LEACH protocol and the modified low-energy clustering protocol with improved residual energy and distance.
  • each node has 4000 bits (500 bytes) of data to be transmitted in each round;
  • E elec 50nJ/bit, representing the power consumption of the receiver circuit and the transmitter circuit to process 1-bit data
  • ⁇ fs 10pJ/bit/m 2 , which represents the power consumption of the transmitter signal amplifying circuit to transmit 1-bit data to a single area;
  • ⁇ amp 0.0013pJ/bit/m 4 , the power consumption of the transmitter signal amplifying circuit to transmit 1-bit data to a single area under the dual-path model, which is suitable for the transmit power of the transmitter amplifier when the distance is long;
  • E DA 5nJ/bit/signal, which means that when the cluster head node performs data fusion, the energy consumption required to process 1 bit of data is to consider the tradeoff between communication overhead and processing overhead. In our model, we use it as a parameter for statistics.
  • the number of nodes is from 50 to 1000, and 50 nodes are added each time, which is easy to understand the relationship between the algorithm performance and the network scale, as shown in Table 1.
  • Sink node location (50, 200) initial energy 2J packet length 500Bytes number of nodes 50-1000 Node energy threshold 0.01J
  • E elec 50nJ/bit EDA 5nJ/bit/signal ⁇ fs 10pJ/bit/m 2 ⁇ amp 0.0013pJ/bit/m 4 d 0 82m
  • the next step is to verify the business in multiple dimensions:
  • Figure 6 is a comparison of the time of death of the first node of the network using the hierarchical clustering routing algorithm, DCHS and EDCSA under the environment of different number of nodes (the number of nodes is from 50 to 1000).
  • Figure 7 shows the comparison of the dead time of all nodes in the network using hierarchical clustering routing algorithm, DCHS and low-energy clustering with improved residual energy and distance under the environment of different number of nodes (the number of nodes is from 50 to 1000).
  • Figure 8 shows the relationship between the number of dead nodes and the network working time (number of rounds) with a scene of 200m ⁇ 200m and the number of nodes is 100.
  • Figure 9 shows the simulation results of the average energy consumption of the three algorithms of hierarchical clustering routing algorithm, DCHS and low-energy clustering method with improved residual energy and distance under different network scales (number of nodes). This algorithm is averaged 10 times for each different network size (the number of nodes ranges from 50 to 1000).
  • the average energy consumption of the low-energy clustering method with improved residual energy and distance is the smallest, because it not only considers the residual energy of nodes, but also considers The distance from the node to the base station can better balance the network load and reduce the energy difference between nodes, thereby prolonging the life cycle of the network, and basically achieving the purpose of reducing energy consumption and maximizing the network life cycle of the wireless sensor network protocol as much as possible. .
  • This application achieves the goal of balanced network energy consumption by controlling the number of clusters, electing as many nodes with remaining energy as possible as close as possible to the base station as cluster heads, and finally using the existing hierarchical clustering routing algorithm protocol to simulate and simulate Performance analysis, the experimental results show that the overall performance of the low-energy clustering method with improved residual energy and distance is better than the hierarchical clustering routing algorithm and DCHS, and achieves the design purpose of saving the energy consumption of wireless sensor networks;
  • the hierarchical data fusion strategy integrates data fusion The operation is advanced to the data collection node, accumulates the data for a certain period of time, and performs data fusion locally, and then transmits it to the cluster head node.
  • the cluster head node fuses the data from different member nodes again;
  • the data fusion strategy can reduce the redundancy of data and reduce the amount of data transmission, thus prolonging the network life cycle.
  • the distributed sensor network data fusion system provided by the present application will be described below, and the distributed sensor network data fusion system described below and the distributed sensor network data fusion method described above may refer to each other correspondingly.
  • FIG. 10 is a schematic structural diagram of a distributed sensor network data fusion system provided by the present application. As shown in FIG. 10 , it includes: an acquisition module 1001 , a first processing module 1002 , a second processing module 1003 and a third processing module 1004
  • the acquisition module 1001 is used to acquire several sensor nodes and sink nodes in the distributed sensor network node set; the first processing module 1002 is used to acquire several cluster head nodes and several cluster head nodes corresponding to each cluster head node based on a preset node data fusion algorithm.
  • the second processing module 1003 is used to perform data fusion on the several in-cluster nodes and the several cluster head nodes based on the improved distributed clustering fusion algorithm of residual energy and distance, to obtain cluster head fusion data and Cluster head fusion data;
  • the third processing module 1004 is configured to transmit the intra-cluster fusion data to the corresponding cluster head node based on a preset fusion mechanism, and the plurality of cluster head nodes transmit the cluster head fusion data to the sink node.
  • the cluster head randomly selects cooperative nodes in each round of data fusion, and the intermediate nodes cooperate with the cluster head to perform data protection and fusion, so as to effectively reduce the computational complexity of nodes and communication volume, and finally the calculation volume, communication volume and fusion accuracy are greatly improved.
  • FIG. 11 illustrates a schematic diagram of the physical structure of an electronic device.
  • the electronic device may include: a processor (processor) 1110, a communication interface (Communications Interface) 1120, a memory (memory) 1130 and a communication bus 1140,
  • the processor 1110 , the communication interface 1120 , and the memory 1130 communicate with each other through the communication bus 1140 .
  • the processor 1110 can invoke the logic instructions in the memory 1130 to execute a distributed sensor network data fusion method, the method comprising: acquiring a number of sensor nodes and sink nodes in a distributed sensor network node set; based on a preset node data fusion algorithm, Obtain several cluster head nodes and several intra-cluster nodes corresponding to each cluster head node; based on the distributed cluster fusion algorithm with improved residual energy and distance, perform data fusion on the several intra-cluster nodes and the several cluster head nodes respectively to obtain cluster head fusion data and cluster head fusion data; based on a preset fusion mechanism, the plurality of intra-cluster nodes transmit the intra-cluster fusion data to the corresponding cluster head nodes, and the plurality of cluster head nodes transmit the cluster head fusion data data to the sink node.
  • the above-mentioned logic instructions in the memory 1130 can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product.
  • the technical solution of the present application can be embodied in the form of a software product in essence, or the part that contributes to the prior art or the part of the technical solution.
  • the computer software product is stored in a storage medium, including Several instructions are used to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present application.
  • the aforementioned storage medium includes: U disk, mobile hard disk, Read-Only Memory (ROM, Read-Only Memory), Random Access Memory (RAM, Random Access Memory), magnetic disk or optical disk and other media that can store program codes .
  • the present application also provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, when the program instructions are executed by a computer
  • the computer can execute the distributed sensor network data fusion method provided by the above methods, and the method includes: obtaining a number of sensor nodes and convergence nodes in a distributed sensor network node set; The cluster head node and several intra-cluster nodes corresponding to each cluster head node; based on the distributed clustering fusion algorithm with improved residual energy and distance, the data of the several intra-cluster nodes and the several cluster head nodes are respectively fused to obtain Cluster head fusion data and cluster head fusion data; based on a preset fusion mechanism, the plurality of intra-cluster nodes transmit the intra-cluster fusion data to the corresponding cluster head node, and the plurality of cluster head nodes transmit the cluster head fusion data to the sink node.
  • the present application also provides a non-transitory computer-readable storage medium on which a computer program is stored, the computer program is implemented by a processor to execute the distributed sensor network data fusion methods provided above, and the computer program is executed by a processor.
  • the method includes: acquiring several sensor nodes and sink nodes in a distributed sensor network node set; based on a preset node data fusion algorithm, acquiring several cluster head nodes and several intra-cluster nodes corresponding to each cluster head node; based on the improved residual energy and The distributed clustering fusion algorithm of distance, respectively fuses the data of the several nodes in the cluster and the several cluster head nodes to obtain the cluster head fusion data and the cluster head fusion data; based on the preset fusion mechanism, the several clusters
  • the node transmits the intra-cluster fusion data to the corresponding cluster head node, and the plurality of cluster head nodes transmit the cluster head fusion data to the sink node.
  • the device embodiments described above are only illustrative, wherein the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in One place, or it can be distributed over multiple network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution in this embodiment. Those of ordinary skill in the art can understand and implement it without creative effort.

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Signal Processing (AREA)
  • Mobile Radio Communication Systems (AREA)

Abstract

本申请提供一种分布式传感器网络数据融合方法及系统，包括：获取若干传感器节点和汇聚节点；基于预设节点数据融合算法，获得若干簇头节点以及若干簇内节点；基于改进剩余能量和距离的分布式分簇融合算法，将若干簇内节点和若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；基于预设融合机制，若干簇内节点传递簇内融合数据至对应的簇头节点，若干簇头节点传递簇头融合数据至汇聚节点。本申请通过基于改进剩余能量和距离的低能耗分簇算法，在每轮数据融合过程中由簇头随机选取协作节点，通过中间节点配合簇头进行数据的保护融合，以有效降低节点的计算量和通信量，最终在计算量、通信量和融合精度上都有较大的提升。

Description

一种分布式传感器网络数据融合方法及系统
相关申请的交叉引用
本申请要求于2020年11月10日提交的申请号为202011248029.3，发明名称为“一种分布式传感器网络数据融合方法及系统”的中国专利申请的优先权，其通过引用方式全部并入本文。
技术领域
本申请涉及通信工程分析验证技术领域，尤其涉及一种分布式传感器网络数据融合方法及系统。
背景技术
现代化通信网络、高质量无线通信网络为现代化社会带来了极大的便捷性，无线网络建设质量也一直都是大家重点关注。其中，无线传感器网络(Wireless Sensor Network，WSN)为多跳自组的网络系统，不仅易于为攻击者提供窃听和篡改数据的机会，也会造成不必要的带宽浪费，因此，WSN数据安全与数据融合的交叉研究已逐渐成为当前的研究热点之一。
随着研究的不断深入，研究人员提出了多种无线传感器网络的隐私保护数据融合方法，这些算法虽然也是将无线传感器采集到的信息进行融合处理然后传送给用户，但是它们融合的目的是为了得到能更精确地反映观测对象的数据信息，以实现对观测数据对象更好的理解，而并没有将传感器节点的能量效率作为首要的优化目标。
发明内容
本申请提供一种分布式传感器网络数据融合方法及系统，用以解决现有技术中针对无线传感器网络没有系统地对传感器节点的能量效率进行优化的缺陷。
第一方面，本申请提供一种分布式传感器网络数据融合方法，包括：
在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；
基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对 应的若干簇内节点；
基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；
基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
进一步，所述基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点，具体包括：
基于时间序列的数据融合算法，对所述若干簇内节点进行预处理；
基于矩阵分析的数据融合算法，对所述若干簇头节点进行预处理。
进一步，所述基于时间序列的数据融合算法，对所述若干簇内节点进行预处理，具体包括：
获取预设时间序列，将所述预设时间序列划分为预设分段数，其中每个分段包含单个数据或相邻两个数据；
提取所述预设分段数中任一两个相邻分段，分别计算所述任一两个相邻分段的合并代价；
选择满足预设合并条件且所述合并代价最小的两个分段进行合并，直到将所有满足所述预设合并条件的分段全部进行合并。
进一步，所述基于矩阵分析的数据融合算法，对所述若干簇头节点进行预处理，具体包括：
确定置信距离测度和置信距离矩阵，以及分别确定关系矩阵与最佳融合数；
基于所述置信距离测度、所述置信距离矩阵、所述关系矩阵和所述最佳融合数，实现数据最优融合。
进一步，所述基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据，具体包括：
所述若干簇内节点和所述若干簇头节点分别进行初始化；
执行第一轮选簇和组簇后，每个节点保存当前剩余能量，执行下一轮选簇和组簇，比较所述当前剩余能量和下一剩余能量，若存在任一个节点 的所述下一剩余能量小于所述当前剩余能量和预设比例之积，则重新开始选簇和组簇，否则维持当前稳定状态。
进一步，所述若干簇内节点和所述若干簇头节点分别进行初始化，具体包括：
分别采用预设簇内标记对所述若干簇内节点进行标记，采用预设簇头标记对所述若干簇头节点进行标记；
通过查询节点标记位，所述若干簇内节点和所述若干簇头节点分别获取各自对应的数据融合算法；
每个簇头节点向对应的簇内节点发出查询命令，确认所述对应的簇内节点的状态；
所述对应的簇内节点返回确认消息，进入监听状态，并打开计时器；
每个簇头节点收到所述确认消息后，进行低能耗模式，等待所述对应的簇内节点进行数据传输。
进一步，所述基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点，具体包括：
基于时间心跳机制，获得所述若干簇内节点采集数据的时间序列长度、数据融合时间、数据传输到簇头节点时间和簇头节点分配给每个簇内节点的传输时隙；
获取距离簇头节点最近的节点，计算最近节点距离和最近节点传输时间，给所述最近的节点分配时间序列，基于所述时间序列得到所述最近的节点的融合时间；
分别计算获得每个簇内节点对应的时间序列，每个簇内节点根据对应的时间序列采集数据并进行数据融合，将融合的数据传输给簇头节点；
待所有簇内节点完成当前一轮数据传输后，根据当前簇内节点、当前簇内节点的融合时间和当前簇内节点到簇头节点传输时间计算得到距离下一轮数据传输的时间间隔。
第二方面，本申请还提供一种分布式传感器网络数据融合系统，包括：
获取模块，用于在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；
第一处理模块，用于基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点；
第二处理模块，用于基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；
第三处理模块，用于基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
第三方面，本申请还提供一种电子设备，包括存储器、处理器及存储在存储器上并可在处理器上运行的计算机程序，所述处理器执行所述程序时实现如上述任一种所述分布式传感器网络数据融合方法的步骤。
第四方面，本申请还提供一种非暂态计算机可读存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现如上述任一种所述分布式传感器网络数据融合方法的步骤。
本申请提供的分布式传感器网络数据融合方法及系统，通过基于改进剩余能量和距离的低能耗分簇算法，在每轮数据融合过程中由簇头随机选取协作节点，通过中间节点配合簇头进行数据的保护融合，以有效降低节点的计算量和通信量，最终在计算量、通信量和融合精度上都有较大的提升。
附图说明
为了更清楚地说明本申请或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作一简单地介绍，显而易见地，下面描述中的附图是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。
图1是本申请提供的一种分布式传感器网络数据融合方法的流程示意图；
图2是本申请提供的无线传感器网络结构图；
图3是本申请提供的改进剩余能量和距离的低能耗分簇方法实现框架图；
图4是本申请提供的改进剩余能量和距离的低能耗分布式具体流程示意图；
图5是本申请提供的簇内成员节点的数据传输过程示意图；
图6是本申请提供的节点数量与FND的关系对比图；
图7是本申请提供的节点数量与LND的关系对比图；
图8是本申请提供的网络生存周期对比图；
图9是本申请提供的节点数量与能量消耗关系对比；
图10是本申请提供的一种分布式传感器网络数据融合系统的结构示意图；
图11是本申请提供的电子设备的结构示意图。
具体实施方式
为使本申请的目的、技术方案和优点更加清楚，下面将结合本申请中的附图，对本申请中的技术方案进行清楚、完整地描述，显然，所描述的实施例是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有作出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请提出了一种基于改进剩余能量和距离的低能耗分布式算法策略，将数据安全融合操作提前到了数据的采集节点上，累积一定时间的数据，并在本地进行数据安全融合，然后传输给簇头节点，簇头节点对来自不同成员节点的数据再一次进行分布式安全融合。最后通过定性分析说明该分布式数据安全融合策略在保证数据安全的前提下，可以降低数据的冗余性，减少数据传输量，从而进一步降低网络节点的计算维度和通信量。
图1是本申请提供的一种分布式传感器网络数据融合方法的流程示意图，如图1所示，包括：
S1，在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；
S2，基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点；
S3，基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇 内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；
S4，基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
具体地，本申请所采用的改进剩余能量和距离的分布式分簇融合算法，考虑了簇头部节点的任务很重，它负责广播消息给所有节点，成簇之后要接收本簇内成员节点传送来的数据，并且融合这些相关数据，最后簇头节点要发送融合后的有用信息给基站，使得基站可以实时监测到远距离的环境变化等。通过发送消息触发，节点在接收到该信息后转换成簇头节点并继续向附近的节点发送信息，分布式网络内的其余网络节点会一直处于等待状态直至接收到从附近节点发送过来的信息，否则就在接收到由任一簇头发送过来的信息后加入该簇。在拓扑结构形成后，会进行节点的密钥分发，根据安全多方计算原理均采用随机密钥预分布的方法，这样可以尽量降低节点密钥被破解的概率，在基站与节点都不可信任的情况下，对数据的安全保护较高。因此，簇头节点往往比非簇头节点消耗更多的能量，如何降低簇头节点的能量消耗和均衡网内节点的能量，对延长整个网络的生存周期至关重要。
最佳的簇内数目是在分析网络消耗的能量的基础上得到的，一方面要考虑簇内节点的计算量,另一方面还要考虑簇头节点计算量，计算量包含算术运算即噪声干扰处理、加解密运算和数据融合运算，基本思想就是在成簇结束后，簇内节点开始感应数据，并进行融合操作，并且选择的簇头数目使得每轮网络消耗的总能量最小。
所有簇内节点都会在其簇内融合过程中执行以下步骤：在每个节点采集完数据后，利用簇内节点公开的数值和多个私有的随机值进行干扰操作，即转成2次多项式。在整个计算过程中，每个节点将多个噪声干扰值进行加密后发送给簇头节点和中间节点，同时簇头节点和中间节点接收到其他节点发送的干扰处理结果后，用共享密钥对数值解密后，,再进行算术运算组合多项式，最优簇头数目的计算公式：
Figure PCTCN2021113733-appb-000001
其中，N表示节点总数，节点均匀分布在M×M的区域内，d to BS是从簇头节点到信号源的距离，ε fs＝10pJ/bit/m 2，ε amp＝0.0013pJ/bit/m 4，是常数。
由于假设过于理想化，得到的最佳簇头数目也是比较理想状态下的个数，而在应用中需要考虑很多其他的因素。通过综合考虑观测区域面积、节点总数、信号源位置等多个因素，并采用典型能量消耗模型，该模型基于两点假设：
(1)无线通信网络中所有节点完全相同；
(2)网络信息在各个方向的能量消耗相同。
无线通信传感器节点发送和接收lbit信息d距离，所消耗的能量分别为：
Figure PCTCN2021113733-appb-000002
E R(l)＝E R-elec(l)＝lE elec
如果接收、发送器之间的距离小于临界值d 0，则使用向量模型；如果大于临界值d 0，则使用多选择模型。E elec是发射电路和接收电路所消耗的能量，它取决于信号的数字编解码、调制和滤波等等因素，在这个模型中发射和接收信号两者相同。
首先假定共有N个节点近似均匀地分布，需要考虑多种因素下，尤其一般部署无线传感器节点的时候，是不可能保证每个节点均匀分布，使节点之间间隔一样，密度一样，同样也不能在同等面积上放置的节点数相差太悬殊，或者有的太密集，有的太疏松，那样不仅不利于收集信息，也不利于分簇、选路、传输，也会浪费大量的能量，不过通常来说，我们会认为节点均匀地分布在M×M的区域内。
如果无线传感器通信网络包含有k个簇，则每个簇里面的节点数为N/k，其中包括一个簇头节点和N/k-1个成员节点，无线通信过程中每次数据传输包含的比特数为l。
采用无线信号源模型的能量公式，一般情况下信号源和簇头之间的距离较远，所以适合采用多路径模型来计算能量消耗。每个簇头节点所消耗的能量包括以下几个部分：接收其成员节点的数据包，进行数据融合处理，将经过融合处理后的数据包发送给远程接收端。则一个簇头节点消耗的能量可表示为：
Figure PCTCN2021113733-appb-000003
一般来说，非簇头节点到各自的簇头节点距离较近，故发送数据的能耗可使用自由空间模型，每个非簇头节点所消耗的能量包括两部分：接收(采集)数据包和发送数据包给簇头节点，则一个非簇头节点消耗的能量可表示为：
E non-CH＝lE elec+lε fsd 2 toCH
其中，d toBS是从簇头节点到基站的距离，d toCH表示从非簇头节点到簇头节点的距离，E DA表示数据融合消耗的能量，E elec表示发送电路和接收电路消耗的能量，ε fs，ε amp都表示信号源增强的放大倍数，分别应用两种模型，这几个参数是常数，通常分别取为E DA＝5nJ/bit/signal，E elec＝50nJ/bit，ε fs＝10pJ/bit/m 2，ε amp＝0.0013pJ/bit/m 4，d toCH表示从非簇头节点到簇头节点的距离。
每一轮循环传送lbit数据时每个簇中所有节点消耗的能量由两部分组成：簇头节点所消耗的能量E CH和成员节点所消耗的能量E non-CH，因此，得到每一轮循环整个网络所消耗的总能量为：
Figure PCTCN2021113733-appb-000004
Figure PCTCN2021113733-appb-000005
整理后，得：
E Total＝l[(2N-k)E elec+NE DA+kε ampd 4 toBS+(N-k)ε fsd 2 toCH]
由于E DA，E elec，ε fs，ε amp这几个参数都是常数，因此，要想求解E Totle，关键是得把d toBS和d toCH求出来。下面通过分布密度和数学期望来求解距离。
整个监测区域R近似记作M 2，基站与传感器节点的距离期望为：
Figure PCTCN2021113733-appb-000006
这个距离期望主要取决于基站的位置坐标(x*,y*)，(x,y)表示传感器节点的位置坐标，由期望定义，d toBS的值等于E[d toBS]。
每个簇所占的区域近似记作M 2/k，那么根据数学期望的定义，簇头节点到成员节点的距离d toBS的平方的期望应为：
E[d toCH 2]＝∫∫(x 2+y 2)ρ(x,y)dxdy
其中ρ(x,y)是传感器节点在每个簇中的分布密度，由于最理想的簇结 构应该是圆，所以假设这个区域是一个半径为
Figure PCTCN2021113733-appb-000007
的圆，并通过相应的坐标变换得：
Figure PCTCN2021113733-appb-000008
其中，ρ(r,θ)对于r，θ是常量，且ρ(r,θ)＝1/(M 2/k)，因此得：
Figure PCTCN2021113733-appb-000009
Figure PCTCN2021113733-appb-000010
令E Totle＝f s(k)，对f s(k)求k的一阶导数，并令其等于零，即：
Figure PCTCN2021113733-appb-000011
由于该函数对k的二阶导数f s (2)(k)恒为正数，因此，函数有最小值，这个最即为最佳簇头个数k opt(是一个有关N，M的函数)：
Figure PCTCN2021113733-appb-000012
当k为上面的值时，网络消耗的能量E Total为最小值。
大多数网络协议常用的成簇方法是，一旦簇头节点被选定，它们便主动向所有节点发送自己成为簇头的信息。依据发射源的强弱，节点选择它所要加入的簇，并告知相应的簇头节点；或者节点根据自己与簇头之间的距离来选择加入哪个簇，距离的计算方法为
Figure PCTCN2021113733-appb-000013
通过对簇头循环和重新组簇的问题进行了如下的考虑：当簇头节点持续工作一段时间后或者其剩余能量低于一定值时，宣布解散簇，然后重新根据成簇算法进行选簇和组簇。我们不希望每进行完一次数据传输(从节点到簇头再到基站，算一轮)就重新选一次簇，频繁的选簇，会增加能量的消耗。
本申请通过基于改进剩余能量和距离的低能耗分簇算法，在每轮数据融合过程中由簇头随机选取协作节点，通过中间节点配合簇头进行数据的保护融合，以有效降低节点的计算量和通信量，最终在计算量、通信量和融合精度上都有较大的提升。
基于上述实施例，所述基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点，具体包括：
基于时间序列的数据融合算法，对所述若干簇内节点进行预处理；
基于矩阵分析的数据融合算法，对所述若干簇头节点进行预处理。
其中，所述基于时间序列的数据融合算法，对所述若干簇内节点进行 预处理，具体包括：
获取预设时间序列，将所述预设时间序列划分为预设分段数，其中每个分段包含单个数据或相邻两个数据；
提取所述预设分段数中任一两个相邻分段，分别计算所述任一两个相邻分段的合并代价；
选择满足预设合并条件且所述合并代价最小的两个分段进行合并，直到将所有满足所述预设合并条件的分段全部进行合并。
其中，所述基于矩阵分析的数据融合算法，对所述若干簇头节点进行预处理，具体包括：
确定置信距离测度和置信距离矩阵，以及分别确定关系矩阵与最佳融合数；
基于所述置信距离测度、所述置信距离矩阵、所述关系矩阵和所述最佳融合数，实现数据最优融合。
具体地，对于无线传感器网络，网络结构如图2所示，包括：该无线传感器网络由大量的传感器节点和一个汇聚节点(基站节点)构成，传感器节点的规格是一样的；网络为分簇式结构，网络形成多个簇，簇结构一旦确定就不再变化；每个簇只有一个簇头节点，簇内含有大量的成员节点；簇内成员节点与簇头节点之间是单跳通信；基站只与各分簇的簇头节点进行通信，不会接收来自成员节点的数据；簇头节点可以得到簇内所有成员节点的位置信息。
进一步地，由于分簇式网络结构中，簇内成员节点和簇头节点担任着不同的角色，分配的工作也不同，这将影响到提出的分级数据融合策略中对于数据融合算法的选择。各节点的工作状态分别如下：
一是簇内成员节点：首先进行初始化，然后自动进入监听状态，开始采集数据，再将数据信息通过无线方式传送给簇头节点，同时进入监听，并开始下一轮的数据采集；从时间因素上考虑在簇内成员节点上实现数据融合，采用基于时间序列的数据融合算法作为簇内成员节点上的融合算法。
此处，基于时间序列的数据融合算法如下：
1)将整个时间序列划分为m(m≥n/2)个分段，每个分段包含一个数据 或相邻的两个数据；
2)分别计算相邻两个分段的合并代价，合并代价主要由以下两方面因素决定：一是两个分段合并后带来的误差；二是合并后分段的所对应的子时间序列所包含的数据的个数；
3)选择满足合并条件要求且合并代价最小的两个分段进行合并；
4)重复上述过程，将所有满足合并条件的分段都进行合并。
通过加入一个计时器，目的是使节点可以自己控制时间序列的大小。
二是簇头节点：首先也是初始化，然后进入低能耗模式，簇头节点不负责采集数据，只接收来自其簇内成员节点传送的数据，当数据到达时，簇头节点进入接收状态，接受并存储数据，当接收完其所有簇内节点的数据后，进行数据融合，这里的数据融合在一般的基于能耗的网络设计中都会使用。数据融合处理之后，发送给基站。簇头节点的数据融合和簇内成员节点有很大的不同，簇内成员节点只是将一定时间内采集的数据进行融合，这个时间可以根据节点的存储空能力而定。而簇头节点是要把其簇内成员节点传送过来的数据进行融合，簇内一般包括了大量的节点，相对而言，簇头节点的工作量要比簇内成员节点大得多，消耗的能量也多得多，因此必须考虑数据融合本身的复杂度和融合的效果。选择采用基于矩阵分析的数据融合算法，一方面它的算法较为简单，计算量少，时间消耗也较少；另一方面，其融合精度与D-S证据组合的数据融合结果不相上下。
此处，基于矩阵分析的数据融合算法如下：
1)置信距离测度和置信距离矩阵的确定；
2)关系矩阵与最佳融合数的确定；
3)对数据的最优融合。
基于上述任一实施例，所述基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据，具体包括：
所述若干簇内节点和所述若干簇头节点分别进行初始化；
执行第一轮选簇和组簇后，每个节点保存当前剩余能量，执行下一轮选簇和组簇，比较所述当前剩余能量和下一剩余能量，若存在任一个节点的所述下一剩余能量小于所述当前剩余能量和预设比例之积，则重新开始 选簇和组簇，否则维持当前稳定状态。
其中，所述若干簇内节点和所述若干簇头节点分别进行初始化，具体包括：
分别采用预设簇内标记对所述若干簇内节点进行标记，采用预设簇头标记对所述若干簇头节点进行标记；
通过查询节点标记位，所述若干簇内节点和所述若干簇头节点分别获取各自对应的数据融合算法；
每个簇头节点向对应的簇内节点发出查询命令，确认所述对应的簇内节点的状态；
所述对应的簇内节点返回确认消息，进入监听状态，并打开计时器；
每个簇头节点收到所述确认消息后，进行低能耗模式，等待所述对应的簇内节点进行数据传输。
具体地，当散播完传感器节点，并确定了网络簇结构之后，簇头节点和簇内成员节点都需要完成其节点上的初始化工作：
1)簇头节点和簇内成员节点用一个节点标志位来区分它们不同的地位，簇头节点的标志为Si设为1，而簇内成员节点的标志位Si设为0；
2)通过查询节点标志位载入各自对应的数据融合算法；
3)簇头节点向簇内成员节点发出一个查询命令，看成员节点是否准备就绪；
4)簇内成员节点返回一个确认消息，并进入监听状态，同时打开计时器；
5)簇头节点接到确认消息后，进入低能耗模式，等待簇内成员节点的数据传输。
接下来是整体的算法框架，包括了簇内成员节点和簇头节点的两级数据融合，初始化之后，无线传感器网络开始工作。
对于簇内成员节点：它们各自开始监听和采集数据，并开始计时，对采集的数据进行存储，当计时器时间达到预设的时间，则发出命令，簇内成员节点对已采集的数据进行数据融合，融合处理完成后将数据结果发送给簇头节点，同时开始新一轮的数据采集并进行计时。
对于簇头节点，为了节省能量，初始化之后进入低能耗模式，直到它 的簇内成员节点传送数据过来，然后开始进行数据融合，融合处理完成后将数据结果发送给基站，同时继续等待和接收簇内成员节点的数据的到来。
对于基站，接收到来自簇头节点的数据后，可以直接传送给所需用户，也可以进行决策性的数据融合后传给用户，由于基站并不存在能量上的限制，因此无需考虑它的节能。
如图3所示，改进剩余能量和距离的低能耗分簇融合方法，不仅能节省簇内成员节点发送数据的能量消耗，也大大节省了簇头节点接收数据的能量消耗。因为通过簇内成员节点的数据融合，不再是原来的每采集一次数据就发送一次，通过融合，只需在一个时间段内发送一次，这样簇头节点接收数据和数据融合的次数也降低了，从而节省了节点的能量消耗，有效延长了网络生命周期。
可以理解的是，通过设置具体的数值比例实现策略，如图4所示。每个节点保存一下自己前一轮结束时的剩余能量E n_current-1，然后通过选簇、成簇、数据传送到接收端，一轮之后，再保存一下自己此时的剩余能量E n_current，当E n_current≤0.5E n_current-1，即这一轮中消耗了该节点剩余能量的一半以上，则需要重新选簇，以防止节点能量过低，不能使下一轮工作正常进行。只要出现一个节点E n_current≤0.5E n_current-1，就重新选簇。
基于上述任一实施例，所述基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点，具体包括：
基于时间心跳机制，获得所述若干簇内节点采集数据的时间序列长度、数据融合时间、数据传输到簇头节点时间和簇头节点分配给每个簇内节点的传输时隙；
获取距离簇头节点最近的节点，计算最近节点距离和最近节点传输时间，给所述最近的节点分配时间序列，基于所述时间序列得到所述最近的节点的融合时间；
分别计算获得每个簇内节点对应的时间序列，每个簇内节点根据对应的时间序列采集数据并进行数据融合，将融合的数据传输给簇头节点；
待所有簇内节点完成当前一轮数据传输后，根据当前簇内节点、当前 簇内节点的融合时间和当前簇内节点到簇头节点传输时间计算得到距离下一轮数据传输的时间间隔。
具体地，将多个数据包在源节点进行融合后传输，在簇头再一次进行融合，比较明显的不足就是延迟较大。由于簇内成员节点到簇头的距离不同，所以传输数据所用的时间也不同，这就导致即使簇内成员节点的数据融合时间相当，但是数据到达簇头节点的时间却相差甚远。这种情况下，使得簇头节点或者等待剩余节点的数据包，这将增加延迟；或者严格控制时间，在规定时间内成员节点的数据包未到达，就将它列入到以后的数据融合中，这样的数据融合效果就会较差。
为了解决改进剩余能量和距离的低能耗分簇带来的延迟问题，引入一个时间心跳机制，具体做法如下：
(1)令簇内成员节点采集数据的时间序列长为s i，数据融合时间为t i(和时间序列长度有关，计算的时候可以用s i表示)，数据传输到簇头节点的时间为T i(和传输距离有关，即簇内成员节点到簇头节点的距离d i)，簇头节点分配给每个成员节点的传输时隙为w，这个传输时隙w表示簇头节点接收一个非簇头节点传送的数据包需要的时间，一个时隙内只允许和一个节点通信，其中i表示簇内成员节点的编号，i＝1,2,...,k，k表示一个簇内含有的成员节点数。簇内成员节点从数据采集到数据传输给簇头节点，所用的时间分配如图5所示。
(2)簇头节点先根据簇内成员节点的位置信息计算其与成员节点的距离d i(i＝1,2,...,k)，然后可以求得各成员节点将数据传送到簇头节点所需要的时间T i。
(3)找到距离最近的节点x，距离为d x，传输所用的时间计为T x，给它分配一个时间序列s x，由于传感器节点能量和存储空间有限，一般有实时性要求，因此时间序列不宜过长。由时间序列可得节点x需要的融合时间t x。这几个时间都确定之后，其它节点就以该节点x为参照对象。
(4)由图5所示，通过下列公式便可求得每个簇内成员节点所对应的时间序列s x,s 1,s 2,…,s k-1。
s 1+t 1+T 1＝s x+t x+T x+w
s 2+t 2+T 2＝s 1+t 1+T 1+w
...
s k-1+t k-1+T k-1＝s k-2+t k-2+T k-2+w
(5)簇内成员节点根据各自的时间序列开始采集数据并进行数据融合，然后传输给簇头节点。
(6)当一轮传输完毕后，由于每个簇内成员节点在时间安排上，都差了一个时隙，刚好能依次完成于簇头节点的数据传输。传输之后自动进入下一轮的采集信息，但是从这一轮开始的时间序列须按以下计算公式计算：
s 1+t 1+T 1＝s x+t x+T x
s 2+t 2+T 2＝s 1+t 1+T 1
...
s k-1+t k-1+T k-1＝s k-2+t k-2+T k-2
以后只需保持s i，t i，T i这三个时间和相等，簇内成员节点传输数据到簇头节点就能依次进行。也就是整个过程需要计算两次：起始和第一轮结束。
上述心跳机制可以保证簇内成员节点的数据依次传送到簇头节点，簇头节点在接收了这些数据之后能立刻进行数据融合，不仅提高了速度，同时也优化了数据融合的效果，大大降低了数据传输的能耗，并且也避免了簇内成员节点因同一时刻传输数据到簇头节点而产生信道争抢和数据丢失的严重后果。
进一步对网络进行定性分析，首先是能耗分析，设网络内有k个簇，每个簇内含有的n k个节点，即1个簇头节点和n k-1个簇内成员节点。采用簇内成员节点与簇头分别对网内数据进行融合，每个簇内成员节点可以在一段时间内采集多个数据包，然后根据基于时间序列的数据融合算法进行融合，合成一个数据包，因此输出的数据包只有一个；簇头节点接收到簇内成员节点的数据后再进行基于矩阵分析的数据融合，加入了时间心跳机制后，成员节点的数据可以依次送达簇头节点。每一轮网络所需发送的数据包的个数为
Figure PCTCN2021113733-appb-000014
与网络内部未采用数据融合或者只在簇头节点进行数据融合的情况相比，显然降低了网络的能量消耗。
其次是减少网络拥塞，通过簇内成员节点先进行时间序列的数据融合，使得原本在一个时段内需要多次传送数据改为了只传送一次，有效地 分散了数据单独传输对信道的争夺和资源浪费。
最后是延迟，由于簇头与簇内成员节点距离不等，造成数据到达簇头节点的时间也不同步，于是簇头节点上的数据融合效果受到影响，带来较大的延迟。不过后来引入了时间心跳机制，确保簇内成员节点能基本上在接近的时间内将数据包送达簇头，使得原本较大的延迟问题得以解决。
基于上述任一实施例，对本申请通过虚拟环境进行模拟仿真，采用NS2工具，即Network Simulator Version 2，是面向对象的、离散事件驱动的网络环境模拟器，主要用于解决网络研究方面的问题。NS2提供了在无线或有线网络上的TCP、路由、多播等多种协议的模拟，并可以完整地模拟整个网络环境，它使用一整套C++类库实现了绝大多数常见的网络协议以及链路层的模型，利用这些类的实例可以搭建起整个网络的模型，而且包括详尽的细节实现。
根据层次化聚类路由分簇算法，为了能进行实验和性能对比，利用层次化聚类路由算法协议，先将层次化聚类路由算法协议源码添成功加到NS2中，然后分别就DCHS分簇算法和EDCSA分簇算法对LEACH作相应的源代码修改，最后将LEACH协议以及修改后的改进剩余能量和距离的低能耗分簇协议进行比较分析。
仿真环境配置：
(1)在200m×200m的区域内随机分布100个同类型的节点，区域范围是横坐标x(0～200)，纵坐标y(0～200)，Sink点的坐标位置为Sink(50，200)；
(2)假设每个节点的初始能量一致，均为2焦耳(J)；
(3)采取连续发送模式，每个节点每一轮中有4000bit(500字节)的数据要传输；
(4)E elec＝50nJ/bit，代表接收机电路和发射机电路处理1比特数据的功耗；
ε fs＝10pJ/bit/m 2，表示发射机信号放大电路向单面积发射1比特数据的功耗；
ε amp＝0.0013pJ/bit/m 4，双路径模型下发射机信号放大电路向单面积发射1比特数据的功耗，适用于距离较远时发射机放大器的发射功率；
(5)数据融合进行时的能耗为E DA＝5nJ/bit/signal，表示簇头节点进行数据融合的时候，每处理1比特的数据需要的能量损耗，为了考虑通信开销和处理开销的折中问题，在我们的模型中，把它作为一个参量进行统计。
(6)节点数由50个至1000个，每次增加50个节点，便于了解算法性能与网络规模的关系，见表1所示。
表1
Sink节点位置 (50，200)
初始能量 2J
数据包长度 500Bytes
节点数目 50-1000
节点能量阈值 0.01J
E elec 50nJ/bit
E DA 5nJ/bit/signal
ε fs 10pJ/bit/m 2
ε amp 0.0013pJ/bit/m 4
d 0 82m
以下是算法涉及到的部分关键函数：
(1)节点初始化：Initsensor()；
(2)未死亡的节点数：LiveNum()；
(3)每轮中最佳的簇头数目：OptNum()；
(4)判断所有的节点是否都当选过簇头一次：Allelect()；
(5)簇头节点的能量消耗：EneCluster()；
(6)非簇头节点的能量消耗：EneNon()。
接下来是对业务进行多维度的验证：
(1)图6是不同节点数目的环境下(节点数目由50-1000)，采用层次化聚类路由算法、DCHS和EDCSA的网络第一个节点死亡的时间的比较。
实验表明，随着节点数目的增多，层次化聚类路由算法、DCHS和改进剩余能量和距离的低能耗分簇的第一个节点死亡的时间是呈缓慢减少的趋势，这是由于对基于分簇的协议，节点数目的增加意味着簇成员的增 多，簇头的能量消耗较大，故第一个节点死亡的时间是较短。但是在节点数为50的时候，改进剩余能量和距离的低能耗分簇的第一个节点死亡的时间是出现的较早，与DCHS较为接近，这主要是因为节点分布较稀疏的原因。
(2)图7为不同节点数目的环境下(节点数目由50-1000)，采用层次化聚类路由算法、DCHS和改进剩余能量和距离的低能耗分簇的网络全部节点死亡时间的比较。
实验表明，随着节点数目的增多，层次化聚类路由算法、DCHS和改进剩余能量和距离的低能耗分簇方法的FND时间呈缓慢增加的趋势，这是因为节点数目的增加，使得网络生命后期具有相对较高能量的节点的比率越高，故全部节点死亡时间长。
(3)图8场景为200m×200m，节点数量为100的节点死亡数量与网络工作时间(轮数)之间的关系图。
从图8中可以看出改进剩余能量和距离的低能耗分簇方法的第一个节点死亡的时间是层次化聚类路由算法的三倍多，全部节点死亡时间是LEACH的两倍多，而与DCHS的差距不是很大，但是改进之后也达到了一定的节能效果。这表明了改进剩余能量和距离的低能耗分簇方法在网络负载均衡上设计得较好，从而避免了节点因能量损耗过大而过早死亡的情况。
(4)图9是层次化聚类路由算法、DCHS和改进剩余能量和距离的低能耗分簇方法三种算法平均能量消耗在不同的网络规模(节点数)下的仿真结果，这是对每种算法在每个不同的网络规模(节点数目由50到1000)下运行10次取平均值得到的结果。
从图中可以看出，和其他的两个算法相比，改进剩余能量和距离的低能耗分簇方法的平均能量消耗是最小的，这是因为它不仅考虑了节点的剩余能量，还考虑了节点到基站的距离，从而较好的平衡网络负载，减少了节点间的能量差，从而延长了网络的生命周期，基本达到了无线传感器网络协议尽可能减少能耗、最大化网络生命周期的目的。
本申请通过控制簇的数目、选举剩余能量尽可能多距离基站尽可能近的节点为簇头，来达到网络能量消耗均衡的目的，最后利用现有的层次化 聚类路由算法协议进行模拟仿真和性能分析，实验结果表明，改进剩余能量和距离的低能耗分簇方法整体性能优于层次化聚类路由算法和DCHS，达到了节省无线传感器网络能耗的设计目的；分级数据融合策略将数据融合操作提前到了数据的采集节点上，累积一定时间的数据，并在本地进行数据融合，然后传输给簇头节点，簇头节点对来自不同成员节点的数据再一次进行融合；最后通过定性分析说明该数据融合策略可以降低数据的冗余性，减少数据传输量，从而延长了网络生命周期。
下面对本申请提供的分布式传感器网络数据融合系统进行描述，下文描述的分布式传感器网络数据融合系统与上文描述的分布式传感器网络数据融合方法可相互对应参照。
图10是本申请提供的一种分布式传感器网络数据融合系统的结构示意图，如图10所示，包括：获取模块1001、第一处理模块1002、第二处理模块1003和第三处理模块1004
获取模块1001用于在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；第一处理模块1002用于基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点；第二处理模块1003用于基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；第三处理模块1004用于基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
本申请通过基于改进剩余能量和距离的低能耗分簇算法，在每轮数据融合过程中由簇头随机选取协作节点，通过中间节点配合簇头进行数据的保护融合，以有效降低节点的计算量和通信量，最终在计算量、通信量和融合精度上都有较大的提升。
图11示例了一种电子设备的实体结构示意图，如图11所示，该电子设备可以包括：处理器(processor)1110、通信接口(Communications Interface)1120、存储器(memory)1130和通信总线1140，其中，处理器1110，通信接口1120，存储器1130通过通信总线1140完成相互间的通信。处理器1110可以调用存储器1130中的逻辑指令，以执行分布式传 感器网络数据融合方法，该方法包括：在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点；基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
此外，上述的存储器1130中的逻辑指令可以通过软件功能单元的形式实现并作为独立的产品销售或使用时，可以存储在一个计算机可读取存储介质中。基于这样的理解，本申请的技术方案本质上或者说对现有技术做出贡献的部分或者该技术方案的部分可以以软件产品的形式体现出来，该计算机软件产品存储在一个存储介质中，包括若干指令用以使得一台计算机设备(可以是个人计算机，服务器，或者网络设备等)执行本申请各个实施例所述方法的全部或部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、随机存取存储器(RAM，Random Access Memory)、磁碟或者光盘等各种可以存储程序代码的介质。
另一方面，本申请还提供一种计算机程序产品，所述计算机程序产品包括存储在非暂态计算机可读存储介质上的计算机程序，所述计算机程序包括程序指令，当所述程序指令被计算机执行时，计算机能够执行上述各方法所提供的分布式传感器网络数据融合方法，该方法包括：在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点；基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
又一方面，本申请还提供一种非暂态计算机可读存储介质，其上存储有计算机程序，该计算机程序被处理器执行时实现以执行上述各提供的分布式传感器网络数据融合方法，该方法包括：在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；基于预设节点数据融合算法，获得 若干簇头节点以及每个簇头节点对应的若干簇内节点；基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
以上所描述的装置实施例仅仅是示意性的，其中所述作为分离部件说明的单元可以是或者也可以不是物理上分开的，作为单元显示的部件可以是或者也可以不是物理单元，即可以位于一个地方，或者也可以分布到多个网络单元上。可以根据实际的需要选择其中的部分或者全部模块来实现本实施例方案的目的。本领域普通技术人员在不付出创造性的劳动的情况下，即可以理解并实施。
最后应说明的是：以上实施例仅用以说明本申请的技术方案，而非对其限制；尽管参照前述实施例对本申请进行了详细的说明，本领域的普通技术人员应当理解：其依然可以对前述各实施例所记载的技术方案进行修改，或者对其中部分技术特征进行等同替换；而这些修改或者替换，并不使相应技术方案的本质脱离本申请各实施例技术方案的精神和范围。

Claims (10)

  1. 一种分布式传感器网络数据融合方法，包括：
    在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；
    基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点；
    基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；
    基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
  2. 根据权利要求1所述的分布式传感器网络数据融合方法，其中所述基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点，具体包括：
    基于时间序列的数据融合算法，对所述若干簇内节点进行预处理；
    基于矩阵分析的数据融合算法，对所述若干簇头节点进行预处理。
  3. 根据权利要求2所述的分布式传感器网络数据融合方法，其中所述基于时间序列的数据融合算法，对所述若干簇内节点进行预处理，具体包括：
    获取预设时间序列，将所述预设时间序列划分为预设分段数，其中每个分段包含单个数据或相邻两个数据；
    提取所述预设分段数中任一两个相邻分段，分别计算所述任一两个相邻分段的合并代价；
    选择满足预设合并条件且所述合并代价最小的两个分段进行合并，直到将所有满足所述预设合并条件的分段全部进行合并。
  4. 根据权利要求2所述的分布式传感器网络数据融合方法，其中所述基于矩阵分析的数据融合算法，对所述若干簇头节点进行预处理，具体包括：
    确定置信距离测度和置信距离矩阵，以及分别确定关系矩阵与最佳融合数；
    基于所述置信距离测度、所述置信距离矩阵、所述关系矩阵和所述最 佳融合数，实现数据最优融合。
  5. 根据权利要求1所述的分布式传感器网络数据融合方法，其中所述基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据，具体包括：
    所述若干簇内节点和所述若干簇头节点分别进行初始化；
    执行第一轮选簇和组簇后，每个节点保存当前剩余能量，执行下一轮选簇和组簇，比较所述当前剩余能量和下一剩余能量，若存在任一个节点的所述下一剩余能量小于所述当前剩余能量和预设比例之积，则重新开始选簇和组簇，否则维持当前稳定状态。
  6. 根据权利要求5所述的分布式传感器网络数据融合方法，其中所述若干簇内节点和所述若干簇头节点分别进行初始化，具体包括：
    分别采用预设簇内标记对所述若干簇内节点进行标记，采用预设簇头标记对所述若干簇头节点进行标记；
    通过查询节点标记位，所述若干簇内节点和所述若干簇头节点分别获取各自对应的数据融合算法；
    每个簇头节点向对应的簇内节点发出查询命令，确认所述对应的簇内节点的状态；
    所述对应的簇内节点返回确认消息，进入监听状态，并打开计时器；
    每个簇头节点收到所述确认消息后，进行低能耗模式，等待所述对应的簇内节点进行数据传输。
  7. 根据权利要求1所述的分布式传感器网络数据融合方法，其中所述基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点，具体包括：
    基于时间心跳机制，获得所述若干簇内节点采集数据的时间序列长度、数据融合时间、数据传输到簇头节点时间和簇头节点分配给每个簇内节点的传输时隙；
    获取距离簇头节点最近的节点，计算最近节点距离和最近节点传输时间，给所述最近的节点分配时间序列，基于所述时间序列得到所述最近的 节点的融合时间；
    分别计算获得每个簇内节点对应的时间序列，每个簇内节点根据对应的时间序列采集数据并进行数据融合，将融合的数据传输给簇头节点；
    待所有簇内节点完成当前一轮数据传输后，根据当前簇内节点、当前簇内节点的融合时间和当前簇内节点到簇头节点传输时间计算得到距离下一轮数据传输的时间间隔。
  8. 一种分布式传感器网络数据融合系统，包括：
    获取模块，用于在分布式传感器网络节点集合中获取若干传感器节点和汇聚节点；
    第一处理模块，用于基于预设节点数据融合算法，获得若干簇头节点以及每个簇头节点对应的若干簇内节点；
    第二处理模块，用于基于改进剩余能量和距离的分布式分簇融合算法，将所述若干簇内节点和所述若干簇头节点分别进行数据融合，得到簇头融合数据和簇头融合数据；
    第三处理模块，用于基于预设融合机制，所述若干簇内节点传递所述簇内融合数据至对应的簇头节点，所述若干簇头节点传递所述簇头融合数据至所述汇聚节点。
  9. 一种电子设备，包括存储器、处理器及存储在所述存储器上并可在所述处理器上运行的计算机程序，其中所述处理器执行所述计算机程序时实现如权利要求1至7任一项所述分布式传感器网络数据融合方法的步骤。
  10. 一种非暂态计算机可读存储介质，其上存储有计算机程序，其中所述计算机程序被处理器执行时实现如权利要求1至7任一项所述分布式传感器网络数据融合方法的步骤。
PCT/CN2021/113733 2020-11-10 2021-08-20 一种分布式传感器网络数据融合方法及系统 WO2022100191A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202011248029.3A CN114071631A (zh) 2020-11-10 2020-11-10 一种分布式传感器网络数据融合方法及系统
CN202011248029.3 2020-11-10

Publications (1)

Publication Number Publication Date
WO2022100191A1 true WO2022100191A1 (zh) 2022-05-19

Family

ID=80233107

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/113733 WO2022100191A1 (zh) 2020-11-10 2021-08-20 一种分布式传感器网络数据融合方法及系统

Country Status (2)

Country Link
CN (1) CN114071631A (zh)
WO (1) WO2022100191A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115278819A (zh) * 2022-06-13 2022-11-01 中国科学院上海微系统与信息技术研究所 一种基于无线传感网的节点调度方法
CN116257892A (zh) * 2023-05-09 2023-06-13 广东电网有限责任公司佛山供电局 一种数字化档案的数据隐私安全性验证方法
CN116539096A (zh) * 2023-05-12 2023-08-04 广东康德威电气股份有限公司 一种基于物联网的变压器状态监测系统
CN116582910A (zh) * 2023-06-30 2023-08-11 中国通信建设第三工程局有限公司 一种数字信息传输方法及系统
CN116668993A (zh) * 2023-07-24 2023-08-29 西安深信科创信息技术有限公司 车路协同自动驾驶感知方法及装置
CN117062063A (zh) * 2023-09-01 2023-11-14 唐人通信技术服务股份有限公司 无线网络的安全防护方法、装置、设备及介质
CN117687578A (zh) * 2024-02-01 2024-03-12 南京信息工程大学 区块链中基于动态分簇的区块协同存储方法、介质及设备

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114650500A (zh) * 2022-03-08 2022-06-21 中国移动通信集团江苏有限公司 数据分发方法、装置、电子设备和存储介质
CN116709392B (zh) * 2023-08-08 2023-11-14 湖南天联城市数控有限公司 一种大规模无线传感器网络数据融合方法

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103037467A (zh) * 2011-09-30 2013-04-10 无锡物联网产业研究院 一种无线传感器网络的数据融合方法和装置

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105430598B (zh) * 2015-11-04 2019-01-18 刘运成 无线传感器网络的数据分层传输方法及装置
CN106658642B (zh) * 2016-12-29 2020-07-24 长春理工大学 一种能量异构无线传感器网络分簇方法和系统

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103037467A (zh) * 2011-09-30 2013-04-10 无锡物联网产业研究院 一种无线传感器网络的数据融合方法和装置

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
CHEN FUZENG: "Mathematical Method of Multi-Sensor Data Fusion", JOURNAL OF MATHEMATICS IN PRACTICE AND THEORY, vol. 25, no. 2, 15 May 1995 (1995-05-15), pages 11 - 16, XP055929774 *
LU YAO, ZHANG TAIHUA , HE ERBAO: "Multivariable WSN Hierarchical Data Fusion Method", MODERN ELECTRONICS TECHNIQUE, vol. 4318026, no. 101, 15 September 2020 (2020-09-15), pages 101 - 105+110, XP055929763, ISSN: 1004-373x, DOI: 10.16652/j.issn.1004-373x.2020.18.026 *
REN QIANQIAN: "Research on Sensing Data Compression Technology in Sensor Network", CHINESE DOCTORAL DISSERTATIONS & MASTER'S THESES FULL-TEXT DATABASE (MASTER), 15 May 2006 (2006-05-15), pages 1 - 103, XP055929770 *
XU JINGQI ,WANG MIN: "A Dynamic Clustering Routing Protocol for Grid-Based Wireless Sensor Networks", COMPUTER MEASUREMENT & CONTROL, vol. 20, no. 6, 25 June 2012 (2012-06-25), pages 1729 - 1732, XP055929767, ISSN: 1671-4598, DOI: 10.16526/j.cnki.11-4762/tp.2012.06.047 *

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115278819A (zh) * 2022-06-13 2022-11-01 中国科学院上海微系统与信息技术研究所 一种基于无线传感网的节点调度方法
CN115278819B (zh) * 2022-06-13 2024-06-07 中国科学院上海微系统与信息技术研究所 一种基于无线传感网的节点调度方法
CN116257892A (zh) * 2023-05-09 2023-06-13 广东电网有限责任公司佛山供电局 一种数字化档案的数据隐私安全性验证方法
CN116257892B (zh) * 2023-05-09 2023-08-29 广东电网有限责任公司佛山供电局 一种数字化档案的数据隐私安全性验证方法
CN116539096A (zh) * 2023-05-12 2023-08-04 广东康德威电气股份有限公司 一种基于物联网的变压器状态监测系统
CN116582910A (zh) * 2023-06-30 2023-08-11 中国通信建设第三工程局有限公司 一种数字信息传输方法及系统
CN116582910B (zh) * 2023-06-30 2024-01-23 中国通信建设第三工程局有限公司 一种数字信息传输方法及系统
CN116668993A (zh) * 2023-07-24 2023-08-29 西安深信科创信息技术有限公司 车路协同自动驾驶感知方法及装置
CN116668993B (zh) * 2023-07-24 2023-11-07 西安深信科创信息技术有限公司 车路协同自动驾驶感知方法及装置
CN117062063A (zh) * 2023-09-01 2023-11-14 唐人通信技术服务股份有限公司 无线网络的安全防护方法、装置、设备及介质
CN117687578A (zh) * 2024-02-01 2024-03-12 南京信息工程大学 区块链中基于动态分簇的区块协同存储方法、介质及设备
CN117687578B (zh) * 2024-02-01 2024-04-16 南京信息工程大学 区块链中基于动态分簇的区块协同存储方法、介质及设备

Also Published As

Publication number Publication date
CN114071631A (zh) 2022-02-18

Similar Documents

Publication Publication Date Title
WO2022100191A1 (zh) 一种分布式传感器网络数据融合方法及系统
Zhao et al. A reinforcement learning method for joint mode selection and power adaptation in the V2V communication network in 5G
Kour et al. Hybrid energy efficient distributed protocol for heterogeneous wireless sensor network
Tan et al. Federated machine learning in vehicular networks: A summary of recent applications
CN100373886C (zh) 基于自适应退避策略的无线传感器网络分布式分簇方法
CN103139862A (zh) 基于查询的无线传感器网络多源数据融合方法
Daanoune et al. BRE-LEACH: A new approach to extend the lifetime of wireless sensor network
Barai et al. Performance evaluation of LEACH protocol for wireless sensor network
Nithyakalyani et al. Data aggregation in wireless sensor network using node clustering algorithms—A comparative study
Ennaciri et al. Load balancing protocol (EESAA) to improve quality of service in wireless sensor network
CN109089273A (zh) 一种Ad-Hoc网络中基于状态转换概率的中继选择方法
Agoujil et al. Optimal cluster head in DTN routing hierarchical topology (DRHT)
Liu et al. QEGWO: Energy-efficient clustering approach for industrial wireless sensor networks using quantum-related bioinspired optimization
Wang et al. LIAA: A listen interval adaptive adjustment scheme for green communication in event-sparse IoT systems
CN113347096B (zh) 第三方支付数据的传输处理方法
CN110661566A (zh) 一种采用深度图嵌入的无人机集群组网方法与系统
CN116133082A (zh) 一种提高航空自组网拓扑持续时间的多跳分簇方法
Tahmassebpour Methods and algorithms of capacity calculation and increase throughput in wireless sensor networks base of ZigBee: A survey
Guo et al. Centralized Clustering Routing Based on Improved Sine Cosine Algorithm and Energy Balance in WSNs
Zhao et al. Cognitive and cache-enabled d2d communications in cellular networks
Sujithra et al. Genetic algorithm based energy efficient data gathering in wireless sensor networks
Wen et al. Research on LEACH routing algorithm based on fuzzy logic
Du et al. A communication scheme for delay sensitive perception tasks of autonomous vehicles
Khan et al. A new approach of energy efficient load balancing for wireless sensor networks
Luo et al. Hierarchical Collaboration Dynamic Resource Scheduling for Edge-Enabled Industrial IoT

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21890720

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21890720

Country of ref document: EP

Kind code of ref document: A1