阅读说明：本技术 基于fcm分簇拓扑的无线传感器网络节点休眠调度方法 (Wireless sensor network node dormancy scheduling method based on FCM (fuzzy c-means) clustering topology ) 是由 高洪元 陈世聪 杨洁 杜亚男 马静雅 孙贺麟 郭瑞晨 李慧爽 刘家威 于 2021-07-12 设计创作，主要内容包括：本发明提供一种基于FCM分簇拓扑的无线传感器网络节点休眠调度方法,规定网络按“轮”周期运行,在动态成簇阶段,根据当前网络状态判断是否需要动态成簇,若需要则采用模糊C均值聚类FCM实现成簇阶段；在冗余节点判断阶段,根据某节点覆盖区域内邻居节点的位置分布情况来判断该节点是否冗余；在动态簇首选举阶段,综合考虑节点剩余能量、节点与基站距离、节点与簇内其他节点距离三个因素并结合当前网络状态来选举最优簇首；在冗余节点休眠调度阶段,调度非簇首冗余节点休眠、簇首冗余节点工作；在稳定数据传输阶段,本发明优化了数据转发传递路径,减少了网络能耗,延长了整个网络的生命周期。(The invention provides a wireless sensor network node dormancy scheduling method based on FCM clustering topology, which comprises the steps that a network is specified to run according to a 'round' cycle, whether dynamic clustering is needed or not is judged according to the current network state in a dynamic clustering stage, and if the dynamic clustering is needed, fuzzy C mean value clustering FCM is adopted to realize the clustering stage; in the redundant node judging stage, whether a node is redundant is judged according to the position distribution condition of a neighbor node in the coverage area of the node; in a dynamic cluster head election stage, the optimal cluster head is elected by comprehensively considering three factors of node residual energy, the distance between a node and a base station and the distance between the node and other nodes in a cluster and combining the current network state; in the redundant node dormancy scheduling stage, scheduling non-cluster-head redundant node dormancy and cluster-head redundant node work; in the stable data transmission stage, the invention optimizes the data forwarding transmission path, reduces the network energy consumption and prolongs the life cycle of the whole network.)

1. A wireless sensor network node dormancy scheduling method based on FCM clustering topology is characterized in that: the method comprises the following steps:

step one, establishing a wireless sensor network system model;

step two: setting network parameters and initializing a network state, and starting the network to operate;

step three: determining the optimal clustering number of the network, and judging whether dynamic clustering is needed;

step four: the base station adopts FCM algorithm to realize network dynamic clustering, and the concrete steps are as follows:

(1) initializing FCM parameters and establishing a target function of the FCM parameters; setting the maximum iteration number of the FCM algorithm as G, wherein the iteration number is marked as G, and G belongs to [1, G ]]The FCM iteration termination factor is epsilon; setting a cluster center position vector set of the k-th round operation FCM in the g-th iteration of the network as For the location vector of the jth cluster center at the g-th iteration of FCM, j' is 1,2^kG1, 2,. G; FCM objective functionWherein m ∈ [1, ∞) ] is a blurring factor,respectively the Euclidean distance and the membership degree of the surviving node i 'relative to the clustering center j' at the g-th iteration of the FCM,s′_i′is the location vector of surviving node i'; randomly initializing a set of clustered center positions when g is 1Degree of membershipIt should be noted that: the sum of the membership of the surviving node i' to all the cluster centers is 1, i.e.

(2) Updating the membership degree; the membership degree is updated according to the following formula: labeling variables to be summed;

(3) updating the clustering center; the cluster center is updated according to the following formula:

(4) updating the relative Euclidean distance; the Euclidean distance between the surviving node and the cluster center is updated according to the following formula:

(5) judging evolution termination to complete network dynamic clustering; judging whether the FCM algorithm reaches an evolution termination condition, namely reaches the maximum iteration number G or meets the requirementIf yes, outputting the clustering centerRealizing network dynamic clustering according to the maximum membership principle; if not, making g equal to g +1, and returning to the step four (2) to continue execution;

(6) the base station packs network dynamic clustering messages and sends the network dynamic clustering messages and awakening beacons to a network in a flooding way, wherein the messages comprise node communication identification IDs, node residual energy, node positions, node working mode identifications, node attribution cluster IDs, other node communication identification IDs in attribution clusters, other node positions in attribution clusters and other node residual energy in attribution clusters, and each surviving node in the network acquires the clustering message belonging to the surviving node according to the communication identification IDs;

step five: awakening a dormant node in a network, and judging a redundant node;

when the network runs in the k-th round, if the nodes i 'and j' are alive and in the working mode, the situation is satisfied If | | | s is satisfied_i″-s_j″||≤r_netThen the nodes i 'and j' are adjacent nodes; the node i' detects all neighbor nodes of the node, if the following five conditions are met, the node is considered as a redundant node, the redundant node does not participate in the detection process as the neighbor node of other nodes to be detected, and therefore the situation of monitoring blind areas caused by the fact that the adjacent nodes are dormant at the same time is effectively avoided, and the conditions are as follows: the number of neighbor nodes of the node i' is more than or equal to 4; with an abscissa smaller than x_i″And the ordinate is less than y_i″The neighbor node of (2); with an abscissa smaller than x_i″And the ordinate is greater than y_i″The neighbor node of (2); with the abscissa greater than x_i″And the ordinate is less than y_i″The neighbor node of (2); with the abscissa greater than x_i″And the ordinate is greater than y_i″The neighbor node of (2);

step six: a wireless sensor network carries out competition election by a cluster head dynamically;

considering the residual energy of the node w in the cluster and the clusterThe ratio of the average residual energy of all surviving nodes in the cluster is taken as a reference and can be expressed asWherein W is the number of the memory active nodes in the cluster to which the node W belongs, E_res(w) is the residual energy of the node w, and the larger the residual energy of the node w is, the larger f₁The larger (w) the higher the probability of electing a cluster head; the distance between the candidate node and the base station also influences the cluster head election, the selection of the node closer to the base station as the cluster head is beneficial to reducing the energy loss of the cluster head, and the distance factor between the candidate node and the base station can be expressed asWherein d is_toBS(w) is the distance from the node w to the base station, the closer the node w is to the base station, f₂The larger (w) the higher the probability of electing a cluster head; the relative distance of the nodes can be expressed asWherein d is_a(w, w ') is the distance between node w and other nodes w' in the cluster, f₃The larger the' (w) is, the smaller the relative distance between the nodes is, and the smaller the communication cost between the nodes is; node base in clusterThe cluster heads are selected in a distributed mode;

step seven: the method comprises the steps of (1) wireless sensor network redundant node dormancy scheduling;

if a certain redundant node successfully selects a cluster head in a dynamic cluster head competition stage, scheduling the redundant node as a working mode; if a certain redundant node fails in the competition stage of the dynamic cluster head, the redundant node is scheduled to be in a sleep mode, the node only monitors a wakeup beacon in the network and sends a message to the cluster head of the home cluster to inform the node to enter the sleep state, and the message comprises: the method comprises the following steps that a node self communication identification ID, a node home cluster ID, node self residual energy and a node working mode identification are obtained;

step eight: the wireless sensor network stabilizes data transmission and updates the network state;

the data routing energy loss is specifically that if a certain node n is a cluster head, the energy loss of stable data transmission of the node n isWherein, in the step (A),the number of nodes which are in the working mode in the cluster to which the cluster head belongs and have the communication mode mark of 0; if a certain node n is a cluster working node and satisfiesχ_nWhen it is equal to 0, its stable data transmission energy lossWherein the content of the first and second substances,the cluster head is a cluster to which the node belongs; if a certain node n is a cluster working node and satisfiesχ_n1, its stable data transmission energy loss After a surviving node carries out stable data transmission, some nodes may die, if the condition is metThen the node working mode identification is updatedIdentifying the node death, and further updating the network state;

step nine: judging the operation termination of the wireless sensor network;

judging whether the network runs to the maximum running wheel number K or all nodes disappear, namely judging whether K +1 is equal to K or notWhether the current time is equal to zero or not, if yes, the operation of the wireless sensor network is terminated; if not, making k equal to k +1, returning to the step three, and performing the next round of operation by the wireless sensor network.

2. The FCM cluster topology based wireless sensor network node dormancy scheduling method of claim 1, wherein: the first step is specifically as follows: energy consumption of the wireless sensor network nodes is mainly generated by data grouping receiving and sending, a first-order radio model is adopted to simulate communication between the nodes, and the method specifically comprises the following steps:E_RX(l)＝lE_elec，E_DA(l)＝γlE_da(ii) a Wherein E is_TX(l,d)、E_RX(l)、E_DA(l) Respectively transmitting, receiving and fusing data energy consumption models of the nodes, the data volume transmitted by the nodes is lbit, the communication distance of the nodes is dm, E_elecEnergy consumed by a node per unit bit of data transmitted or received, E_daEnergy consumed by each fusion unit bit data of the node is shown, and gamma is a data fusion compression rate; threshold value for selecting which transmission model to communicate by nodeDetermination of epsilon_fsAnd ε_mpThe energy required for power amplification of the free space channel model and the multipath fading model is obtained when the communication distance of the nodes is less than a threshold valueAnd selecting a free space channel model, otherwise, selecting a multipath fading model.

3. The FCM cluster topology based wireless sensor network node dormancy scheduling method of claim 2, wherein: the second step is specifically as follows: setting N sensor nodes in a network to be deployed in an M multiplied by M monitoring area at random and static high density, wherein the maximum operation wheel number of the network is K, the operation wheel number is marked as K, and K belongs to [1, K ∈]Network coefficient of l_netThe node perception radius is r_netThe initial energy of the node is p_netJ, communication data packet is l_databit, control data packet is l_conbit, base station position vector isThe state of the nth sensor node in the kth round of network operationIt is shown that, among others,for the nth node position vector,is the residual energy of the nth node during the kth run, chi_nFor the nth node communication mode identification,identifying an operating mode when the kth node runs, wherein N is 1,2, and N, K is 1, 2.When x_nWhen the node is 1, the node is used as an intra-cluster node or a cluster head to communicate with the base station in a single hop; when x_nWhen the node is 0, the node is used as an intra-cluster node to communicate with a cluster head in a single hop manner or used as the cluster head to communicate with a base station in a single hop manner;when in useWhen the node is in the working mode; when in useWhen the node is in the sleep mode; when in useWhen the node dies, at this timeInitializing the current running wheel number k of the network to be 1, and remaining energy of the nodesN, the base station assigns a unique communication identification ID to each node in the network, and the network starts to operate.

4. The FCM cluster topology based wireless sensor network node dormancy scheduling method of claim 3, wherein: the third step is specifically as follows: optimal clustering number when network runs to k roundWherein the content of the first and second substances,storing the number of the movable nodes for the kth round, wherein round (·) is an integer function, when the network runs in the first round, a base station floods an uplink beacon to the network, and after each survival node in the network receives the uplink beacon, the survival node marks the position, the residual energy, the communication identification ID and the working mode identificationThe packet is sent to the base station, and the base station determines the number of active nodes stored in the k-th network operation according to the packetWhen the network runs in other rounds, the base station reads the control packet in the network uploading data according to a certain specific protocol, determines the state of the network node, and further determines the number of active nodes stored in the k-th round of network operationWhen k ≠ 1 and c are satisfied^k＝c^k-1And in time, the network does not need to be dynamically clustered, the base station sends a wakeup beacon to the network in a flooding way, and the step five is carried out continuously.

Technical Field

The invention relates to a wireless sensor network node dormancy scheduling method based on FCM clustering topology, and belongs to the technical field of wireless sensor networks.

Background

The WSN is a multi-hop self-organizing network formed by a large number of micro sensors in a wireless communication mode, and along with the development of microelectronic technology, sensing technology and communication technology, the WSN has wide application and development prospects in the fields of science, medical treatment, commerce, national defense and the like. The sensor nodes have the capabilities of data acquisition, processing, wireless communication and self-organization, and the nodes cooperate with each other to jointly complete large-scale complex monitoring tasks. The energy of the wireless sensor network is limited due to the limited computing capability and energy of the wireless sensor nodes, the battery cannot be replaced and the like, and it is an important research branch of the WSN to extend the life cycle of the wireless sensor network as much as possible.

At present, the energy-saving research of the wireless sensor network is mainly developed from two levels, namely a network level and a node level. Good clustering rules and routing protocols can be designed at the network level, so that the energy consumption is reduced when data is forwarded and transmitted, and the life cycle of the network is prolonged; the wireless sensor power supply can be powered on and throttled at the node level, the wireless sensor power supply can be prolonged in endurance time through research on electronic components or can be charged midway through solar energy and the like, redundant data collection is reduced mainly through a node dormancy scheduling strategy in the throttling aspect, energy consumption of the sensor is reduced, and the life cycle of the network is further prolonged. The wireless sensor network clustering routing protocol generally stipulates that the network runs according to a 'round' period, each round comprises three stages of cluster head election, cluster establishment and stable data transmission, a node in a cluster is in single-hop communication with the cluster head, and the cluster head is in single-hop or multi-hop communication with a base station; the wireless sensor network redundant node dormancy scheduling strategy generally provides that a network node has two working modes of working and dormancy, the working modes of the network node are scheduled while certain coverage quality is met, and part of the nodes are put into a low-power-consumption dormancy state in turn. The design of combining the clustering routing of the wireless sensor network and the dormancy scheduling of the redundant nodes can further reduce data redundancy and balance network energy consumption, thereby prolonging the life cycle of the whole network.

Through the search discovery of the prior art documents, a dynamic random cluster head election mechanism related to the residual energy of nodes is adopted in the research on a wireless sensor network clustering topological coverage area node scheduling optimization algorithm published in the journal of the technical science of sensing (2010,23(01):116-121) of Sunbao et al, so that the probability of selecting a cluster head by a node is positively correlated with the residual energy of the node, whether the node is redundant is judged according to the position distribution condition of a non-cluster head neighbor node in the coverage area of the node, and the node is subjected to sleep scheduling, thereby balancing the network energy consumption, but a 'hot zone' occurs in the network operation process, namely a large cluster, particularly the number of cluster heads is rapidly reduced at the later stage of the network operation, and the normal requirement of a clustering routing protocol is difficult to meet; before a redundant node judgment link is promoted to a dynamic random cluster head election link in ' wireless sensor network node scheduling optimization clustering algorithm research ' published in ' computer application research ' (2011,28(04):1499-1501) ', by dawn and the like, not only more redundant nodes can be judged, but also the number of cluster heads can be effectively reduced, but a ' hot zone ' still occurs in the network operation process, and the network coverage quality is reduced by more dormant nodes; in a wireless sensor network redundant node dormancy scheduling algorithm published in the communication science report (2014,35(10):67-80), the Tao et al researches the boundary contraction problem caused by node dormancy scheduling in the later operation stage of the network, but the research does not combine a network clustering routing protocol and only comprises the node dormancy scheduling; in a wireless sensor network node dormancy scheduling strategy based on data similarity published in the journal of the technical science of sensing (2019,32(12):1881-1888), such as Li Wei et al, a distributed non-uniform clustering routing protocol is combined with a node dormancy scheduling strategy based on fuzzy clustering of data received by a base station to further prolong the life cycle of the network, and the accurate positions of the data received by the base station and the nodes are used as the prior knowledge of node dormancy scheduling, so that the network calculation amount is increased greatly, and the principle of simple protocol is violated.

The existing literature results show that the existing wireless sensor network redundant node dormancy scheduling method has the problems of 'hot zones', is high in computational complexity and poor in robustness, and lacks a method for optimally combining network clustering routing and redundant node dormancy scheduling to comprehensively reduce WSN (wireless sensor network) network level energy consumption and node level energy consumption, so that the wireless sensor network node dormancy scheduling method based on FCM (fuzzy c-means) clustering topology is provided.

Disclosure of Invention

Aiming at the defects and shortcomings of the existing redundant node dormancy scheduling method of the wireless sensor network, the invention provides a node dormancy scheduling method based on FCM (fuzzy c-means) clustering topology, so that network clustering routing and node dormancy scheduling are combined more optimally, and WSN (wireless sensor network) network-level energy consumption and node-level energy consumption are reduced comprehensively. The method provides that the network operates according to a 'round' period, and each round specifically comprises five stages: dynamic clustering, redundant node judgment, dynamic cluster head election, redundant node dormancy scheduling and stable data transmission. In the dynamic clustering stage, judging whether dynamic clustering is needed or not according to the current network state, and if so, adopting fuzzy C mean clustering (FCM) to realize the clustering stage; in the redundant node judging stage, whether a node is redundant is judged according to the position distribution condition of a neighbor node in the coverage area of the node; in a dynamic cluster head election stage, the optimal cluster head is elected by comprehensively considering three factors of node residual energy, the distance between a node and a base station and the distance between the node and other nodes in a cluster and combining the current network state; in the redundant node dormancy scheduling stage, scheduling non-cluster-head redundant node dormancy and cluster-head redundant node work; in the stable data transmission stage, the traditional data transmission mode of single-hop communication between the intra-cluster nodes and the cluster head and single-hop or multi-hop communication between the cluster head and the base station is not adopted, but the novel data transmission mode of single-hop communication between the intra-cluster nodes and the cluster head or the base station and single-hop or multi-hop communication between the cluster head and the base station is adopted, so that the data forwarding transmission path is optimized, the network energy consumption is reduced, and the life cycle of the whole network is prolonged.

The purpose of the invention is realized as follows: the method comprises the following steps:

step one, establishing a wireless sensor network system model;

step two: setting network parameters and initializing a network state, and starting the network to operate;

step three: determining the optimal clustering number of the network, and judging whether dynamic clustering is needed;

step four: the base station adopts FCM algorithm to realize network dynamic clustering, and the concrete steps are as follows:

(1) initializing FCM parameters and establishing a target function of the FCM parameters; setting the maximum iteration number of the FCM algorithm as G, wherein the iteration number is marked as G, and G belongs to [1, G ]]The FCM iteration termination factor is epsilon; setting a cluster center position vector set of the k-th round operation FCM in the g-th iteration of the network asFor the location vector of the jth cluster center at the g-th iteration of FCM, j' is 1,2^kG1, 2,. G; FCM objective functionWherein m ∈ [1, ∞) ] is a blurring factor,respectively the Euclidean distance and the membership degree of the surviving node i 'relative to the clustering center j' at the g-th iteration of the FCM,s′_i′is the location vector of surviving node i'; randomly initializing a set of clustered center positions when g is 1Degree of membershipIt should be noted that: the sum of the membership of the surviving node i' to all the cluster centers is 1, i.e.

(2) Updating the membership degree; the membership degree is updated according to the following formula: labeling variables to be summed;

(3) updating the clustering center; the cluster center is updated according to the following formula:

(4) updating the relative Euclidean distance; the Euclidean distance between the surviving node and the cluster center is updated according to the following formula:

(5) judging evolution termination to complete network dynamic clustering; judging whether the FCM algorithm reaches an evolution termination condition, namely reaches the maximum iteration number G or meets the requirementIf yes, outputting the clustering centerRealizing network dynamic clustering according to the maximum membership principle; if not, making g equal to g +1, and returning to the step four (2) to continue execution;

(6) the base station packs network dynamic clustering messages and sends the network dynamic clustering messages and awakening beacons to a network in a flooding way, wherein the messages comprise node communication identification IDs, node residual energy, node positions, node working mode identifications, node attribution cluster IDs, other node communication identification IDs in attribution clusters, other node positions in attribution clusters and other node residual energy in attribution clusters, and each surviving node in the network acquires the clustering message belonging to the surviving node according to the communication identification IDs;

step five: awakening a dormant node in a network, and judging a redundant node;

when the network runs in the k-th round, if the nodes i 'and j' are alive and in the working mode, the situation is satisfied If it satisfiesThe nodes i 'and j' are adjacent nodes; the node i ' detects all the neighbor nodes of the node i ' and if the following five conditions are met simultaneously, the node i ' is considered to be a neighbor nodeThe node is a redundant node, the redundant node does not participate in the detection process as a neighbor node of other nodes to be detected, so that the situation of monitoring blind areas caused by the fact that the adjacent nodes are dormant at the same time is effectively avoided, and the conditions are as follows: the number of neighbor nodes of the node i' is more than or equal to 4; with an abscissa smaller than x_i″And the ordinate is less than y_i″The neighbor node of (2); with an abscissa smaller than x_i″And the ordinate is greater than y_i″The neighbor node of (2); with the abscissa greater than x_i″And the ordinate is less than y_i″The neighbor node of (2); with the abscissa greater than x_i″And the ordinate is greater than y_i″The neighbor node of (2);

step six: a wireless sensor network carries out competition election by a cluster head dynamically;

considering the ratio of the remaining energy of the node w in the cluster to the average remaining energy of all surviving nodes in the cluster as a reference, it can be expressed asWherein W is the number of the memory active nodes in the cluster to which the node W belongs, E_res(w) is the residual energy of the node w, and the larger the residual energy of the node w is, the larger f₁The larger (w) the higher the probability of electing a cluster head; the distance between the candidate node and the base station also influences the cluster head election, the selection of the node closer to the base station as the cluster head is beneficial to reducing the energy loss of the cluster head, and the distance factor between the candidate node and the base station can be expressed asWherein d is_toBS(w) is the distance from the node w to the base station, the closer the node w is to the base station, f₂The larger (w) the higher the probability of electing a cluster head; the relative distance of the nodes can be expressed asWherein d is_a(w, w ') is the distance between node w and other nodes w' in the cluster, f₃The larger the' (w) is, the smaller the relative distance between the nodes is, and the smaller the communication cost between the nodes is; node base in clusterThe cluster heads are selected in a distributed mode;

step seven: the method comprises the steps of (1) wireless sensor network redundant node dormancy scheduling;

if a certain redundant node successfully selects a cluster head in a dynamic cluster head competition stage, scheduling the redundant node as a working mode; if a certain redundant node fails in the competition stage of the dynamic cluster head, the redundant node is scheduled to be in a sleep mode, the node only monitors a wakeup beacon in the network and sends a message to the cluster head of the home cluster to inform the node to enter the sleep state, and the message comprises: the method comprises the following steps that a node self communication identification ID, a node home cluster ID, node self residual energy and a node working mode identification are obtained;

step eight: the wireless sensor network stabilizes data transmission and updates the network state;

the data routing energy loss is specifically that if a certain node n is a cluster head, the energy loss of stable data transmission of the node n isWherein, in the step (A),the number of nodes which are in the working mode in the cluster to which the cluster head belongs and have the communication mode mark of 0; if a certain node n is a cluster working node and satisfiesχ_nWhen it is equal to 0, its stable data transmission energy lossWherein the content of the first and second substances,the cluster head is a cluster to which the node belongs; if a certain node n is a cluster working node and satisfiesχ_n1, its stable data transmission energy lossAfter a surviving node carries out stable data transmission, some nodes may die, if the condition is metThen the node working mode identification is updatedIdentifying the node death, and further updating the network state;

step nine: judging the operation termination of the wireless sensor network;

judging whether the network runs to the maximum running wheel number K or all nodes disappear, namely judging whether K +1 is equal to K or notWhether the current time is equal to zero or not, if yes, the operation of the wireless sensor network is terminated; if not, making k equal to k +1, returning to the step three, and performing the next round of operation by the wireless sensor network.

The invention also includes such structural features:

1. the first step is specifically as follows: energy consumption of the wireless sensor network nodes is mainly generated by data grouping receiving and sending, a first-order radio model is adopted to simulate communication between the nodes, and the method specifically comprises the following steps:E_RX(l)＝lE_elec，E_DA(l)＝γlE_da(ii) a Wherein E is_TX(l,d)、E_RX(l)、E_DA(l) Respectively transmitting, receiving and fusing data energy consumption models of the nodes, the data volume transmitted by the nodes is lbit, the communication distance of the nodes is dm, E_elecEnergy consumed by a node per unit bit of data transmitted or received, E_daEnergy consumed by each fusion unit bit data of the node is shown, and gamma is a data fusion compression rate; node selects what kind of transmission to transmitModel-communicated thresholdDetermination of epsilon_fsAnd ε_mpThe energy required for power amplification of the free space channel model and the multipath fading model is obtained when the communication distance of the nodes is less than a threshold valueAnd selecting a free space channel model, otherwise, selecting a multipath fading model.

2. The second step is specifically as follows: setting N sensor nodes in a network to be deployed in an M multiplied by M monitoring area at random and static high density, wherein the maximum operation wheel number of the network is K, the operation wheel number is marked as K, and K belongs to [1, K ∈]Network coefficient of l_netThe node perception radius is r_netThe initial energy of the node is p_netJ, communication data packet is l_databit, control data packet is l_conbit, base station position vector isThe state of the nth sensor node in the kth round of network operationIt is shown that, among others,for the nth node position vector,is the residual energy of the nth node during the kth run, chi_nFor the nth node communication mode identification,identifying an operating mode when the kth node runs, wherein N is 1,2, and N, K is 1, 2.When x_nWhen the node is 1, the node is used as an intra-cluster node or a cluster head to communicate with the base station in a single hop; when x_nWhen the node is 0, the node is used as an intra-cluster node to communicate with a cluster head in a single hop manner or used as the cluster head to communicate with a base station in a single hop manner;when in useWhen the node is in the working mode; when in useWhen the node is in the sleep mode; when in useWhen the node dies, at this timeInitializing the current running wheel number k of the network to be 1, and remaining energy of the nodes The base station distributes a unique communication identification ID to each node in the network, and the network starts to operate.

3. The third step is specifically as follows: optimal clustering number when network runs to k roundWherein the content of the first and second substances,storing the number of the movable joints for the kth round, round (DEG) is an integer function, when the network runs in the first round, the base station floods an uplink beacon to the network, each survival node in the network packages the position, the residual energy, the communication identification ID and the working mode identification after receiving the uplink beacon and sends the packaged information to the base station, and the base station determines that the movable joints are stored for the kth round of network running according to the packaged informationCounting numberWhen the network runs in other rounds, the base station reads the control packet in the network uploading data according to a certain specific protocol, determines the state of the network node, and further determines the number of active nodes stored in the k-th round of network operationWhen k ≠ 1 and c are satisfied^k＝c^k-1And in time, the network does not need to be dynamically clustered, the base station sends a wakeup beacon to the network in a flooding way, and the step five is carried out continuously.

Compared with the prior art, the invention has the beneficial effects that: the existing wireless sensor network redundant node dormancy scheduling method has the problems of 'hot zones', high calculation complexity and poor robustness, and lacks a method for optimally combining network clustering routing and redundant node dormancy scheduling so as to comprehensively reduce WSN (wireless sensor network) level energy consumption and node level energy consumption, so that the wireless sensor network node dormancy scheduling method based on FCM (fuzzy C-means) clustering topology is provided. The method provides that the network operates according to a 'round' period, and each round specifically comprises five stages: dynamic clustering, redundant node judgment, dynamic cluster head election, redundant node dormancy scheduling and stable data transmission. In the dynamic clustering stage, judging whether dynamic clustering is needed or not according to the current network state, and if so, adopting fuzzy C mean clustering (FCM) to realize the clustering stage; in the redundant node judging stage, whether a node is redundant is judged according to the position distribution condition of a neighbor node in the coverage area of the node; in a dynamic cluster head election stage, the optimal cluster head is elected by comprehensively considering three factors of node residual energy, the distance between a node and a base station and the distance between the node and other nodes in a cluster and combining the current network state; in the redundant node dormancy scheduling stage, scheduling non-cluster-head redundant node dormancy and cluster-head redundant node work; in the stable data transmission stage, the traditional data transmission mode of single-hop communication between the intra-cluster nodes and the cluster head and single-hop or multi-hop communication between the cluster head and the base station is not adopted, but the novel data transmission mode of single-hop communication between the intra-cluster nodes and the cluster head or the base station and single-hop or multi-hop communication between the cluster head and the base station is adopted, so that the data forwarding transmission path is optimized, the network energy consumption is reduced, and the life cycle of the whole network is prolonged.

Simulation experiments prove the effectiveness of the wireless sensor network node dormancy scheduling method based on the FCM clustering topology, and compared with the traditional method, the method enables the operation of the wireless sensor network to be more energy-saving, and the method can be used as a node dormancy scheduling protocol of a high-density static sensor network.

Drawings

Fig. 1 is a schematic diagram of a wireless sensor network node dormancy scheduling method based on an FCM cluster topology designed by the present invention.

Fig. 2(a) - (f) are diagrams showing network clustering and node dormancy scheduling results.

Fig. 3 is a graph of the number of surviving nodes of the network and the number of running rounds of the network.

FIG. 4 is a graph of network residual energy versus network operating turns.

Fig. 5 is a relationship curve between the number of network sleeping nodes and the number of network running rounds.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

The steps of the invention are as follows with reference to the attached drawings:

step one, establishing a wireless sensor network system model.

The wireless sensor network researched by the invention is a high-density static sensor network which is randomly deployed, and the following assumptions are made: the sensor nodes are static nodes, the nodes do not move along with time after being randomly deployed, high redundancy is achieved, the nodes form a network by themselves, but the topological structure of the nodes can be changed due to node energy exhaustion or node failure and other reasons; all the nodes are homogeneous nodes, namely all specification parameters are the same, especially initial energy, sensing radius, computing capacity and communication capacity are the same, but the method is not limited to a specific certain sensor node; the node communication power is adjustable, and the transmitting power can be automatically adjusted according to the distance; the position of the base station is known, the energy and the computing capacity are infinite, and all nodes can directly communicate with the base station; the node can calculate the relative position of the node according to the arrival angle and the signal strength of the signal; each node has two working modes of working and sleeping, and the energy consumption of the node in the sleeping mode can be ignored; the node can sense all the nodes which are positioned in the sensing radius and are in the working mode; each node has a unique identification ID, and nodes mutually sensed can obtain the identification ID of the other node; the network communication channel is a symmetric channel; the collision problem is not considered in the data transmission process among the nodes.

Energy consumption of the wireless sensor network nodes is mainly generated by data grouping receiving and sending, a first-order radio model is adopted to simulate communication between the nodes, and the method specifically comprises the following steps:E_RX(l)＝lE_elec，E_DA(l)＝γlE_da. Wherein E is_TX(l,d)、E_RX(l)、E_DA(l) Respectively transmitting, receiving and fusing data energy consumption models of the nodes, the data volume transmitted by the nodes is lbit, the communication distance of the nodes is dm, E_elecEnergy consumed by a node per unit bit of data transmitted or received, E_daGamma is the data fusion compression ratio, and gamma is the energy consumed by the node per fusion unit bit data. Threshold value for selecting which transmission model to communicate by nodeDetermination of epsilon_fsAnd ε_mpThe energy required for power amplification of the free space channel model and the multipath fading model is obtained when the communication distance of the nodes is less than a threshold valueAnd selecting a free space channel model, otherwise, selecting a multipath fading model.

Step two: setting network parameters and initializing the network state, and starting the network to operate.

Setting N sensor nodes in the network to be randomly and statically deployed in an MxM monitoring area in high density, wherein the network is the most advancedThe number of large running wheels is K, the number of running wheels is marked by K, and K belongs to [1, K ]]Network coefficient of l_netThe node perception radius is r_netThe initial energy of the node is p_netJ, communication data packet is l_databit, control data packet is l_conbit, base station position vector isThe state of the nth sensor node in the kth round of network operationIt is shown that, among others,for the nth node position vector,is the residual energy of the nth node during the kth run, chi_nFor the nth node communication mode identification,for the operation mode identification of the nth node in the kth round of operation, N is 1, 2.When x_nWhen the node is 1, the node is used as an intra-cluster node or a cluster head to communicate with the base station in a single hop; when x_nAnd when the node is 0, the node is in single-hop communication with the cluster head as the intra-cluster node or in single-hop communication with the base station as the cluster head.When in useWhen the node is in the working mode; when in useWhen the node is in the sleep mode; when in useWhen the node dies, at this timeInitializing the current running wheel number k of the network to be 1, and remaining energy of the nodes The base station distributes a unique communication identification ID to each node in the network, and the network starts to operate.

Step three: and determining the optimal clustering number of the network, and judging whether dynamic clustering is needed.

Reasonable clustering number is beneficial to forming more uniform clusters, and meanwhile, the overall energy consumption cost of the network is reduced. Optimal clustering number when network runs to k roundWherein the content of the first and second substances,the number of points of articulation is saved for the kth round, round (. cndot.) is an integer function. The number of surviving nodes is gradually reduced along with the operation of the network, so that clustering is a dynamic process, and when the first-round operation of the network or the optimal clustering number changes, clustering needs to be carried out again. When the network first-round operation is carried out, the base station floods an uplink beacon to the network, each survival node in the network packages and sends the position, the residual energy, the communication identification ID and the working mode identification to the base station after receiving the uplink beacon, and the base station determines the number of the survival nodes in the kth-round network operation according to the numberWhen the network runs in other rounds, the base station reads the control packet in the network uploading data according to a certain specific protocol, determines the state of the network node, and further determines the survival of the network running in the kth roundNumber of nodesWhen k ≠ 1 and c are satisfied^k＝c^k-1And in time, the network does not need to be dynamically clustered, the base station sends a wakeup beacon to the network in a flooding way, and the step five is carried out continuously.

Step four: the base station adopts FCM algorithm to realize network dynamic clustering, and the concrete steps are as follows:

(1) the FCM parameters are initialized and their objective function is established. Setting the maximum iteration number of the FCM algorithm as G, wherein the iteration number is marked as G, and G belongs to [1, G ]]The FCM iteration termination factor is epsilon. Setting a cluster center position vector set of the k-th round operation FCM in the g-th iteration of the network asFor the location vector of the jth cluster center at the g-th iteration of FCM, j' is 1,2^kG1, 2. At this time, the FCM objective functionWherein m ∈ [1, ∞) ] is a blurring factor,respectively the Euclidean distance and the membership degree of the surviving node i 'relative to the clustering center j' at the g-th iteration of the FCM,s′_i′is the location vector of the surviving node i'. Randomly initializing a set of clustered center positions when g is 1Degree of membershipIt should be noted that: the sum of the membership of the surviving node i' to all the cluster centers is 1, i.e.

(2) And updating the membership degree. The membership degree is updated according to the following formula: the variables to be summed are labeled.

(3) And updating the clustering center. The cluster center is updated according to the following formula:

(4) the relative euclidean distance is updated. The Euclidean distance between the surviving node and the cluster center is updated according to the following formula:

(5) and (5) judging evolution termination to finish network dynamic clustering. Judging whether the FCM algorithm reaches an evolution termination condition, namely reaches the maximum iteration number G or meets the requirementIf yes, outputting the clustering centerRealizing network dynamic clustering according to the maximum membership principle; if not, let g be g +1, return to step four (2) and continue execution.

(6) The base station packs network dynamic clustering messages and sends the network dynamic clustering messages and the awakening beacon together to a network in a flooding mode, the messages comprise node communication identification IDs, node residual energy, node positions, node working mode identifications, node attribution cluster IDs, other node communication identification IDs in attribution clusters, other node positions in attribution clusters and other node residual energy in attribution clusters, and all surviving nodes in the network acquire the clustering messages belonging to the surviving nodes according to the communication identification IDs.

Step five: and awakening the dormant node in the network, and judging the redundant node.

Each surviving node in the network can receive the awakening beacon, the node which is originally in the working mode does not process the awakening beacon after receiving the beacon, and the node which is originally in the sleeping mode enters the transient working mode after receiving the beacon. All the surviving nodes in the network sequentially detect the neighbor nodes in the coverage area of the nodes according to the sequence of the communication identification IDs from small to large. When the network runs in the k-th round, if the nodes i 'and j' are alive and in the working mode, the situation is satisfiedIf | | | s is satisfied_i″-s_j″||≤r_netThen nodes i ", j" are neighbors to each other. The node i' detects all neighbor nodes of the node, if the following five conditions are met, the node is considered as a redundant node, the redundant node does not participate in the detection process as the neighbor node of other nodes to be detected, and therefore the situation of monitoring blind areas caused by the fact that the adjacent nodes are dormant at the same time is effectively avoided, and the conditions are as follows: the number of neighbor nodes of the node i' is more than or equal to 4; with an abscissa smaller than x_i″And the ordinate is less than y_i″The neighbor node of (2); with an abscissa smaller than x_i″And the ordinate is greater than y_i″The neighbor node of (2); with the abscissa greater than x_i″And the ordinate is less than y_i″The neighbor node of (2); with the abscissa greater than x_i″And the ordinate is greater than y_i″Of the neighboring node.

Step six: and (4) carrying out competition election by the wireless sensor network dynamic cluster head.

In order to reduce the energy consumption of the cluster head and avoid network failure caused by premature death of the node, the three factors of node residual energy, the distance between the node and a base station and the distance between the node and other surviving nodes in the cluster are comprehensively considered, and the optimal cluster head is selected by combining the network characteristics of the current round. In this specification, a surviving node refers to a node with a residual energy greater than zero, and includes both a working node, a redundant node, and a sleeping node.Energy is the most important challenge faced by the nodes, and selecting the nodes in the cluster with high energy as the cluster head helps the load balance in the cluster, so the ratio of the residual energy of the nodes w in the cluster to the average residual energy of all the surviving nodes in the cluster is considered as the reference, which can be expressed asWherein W is the number of the memory active nodes in the cluster to which the node W belongs, E_res(w) is the residual energy of the node w, and the larger the residual energy of the node w is, the larger f₁The larger the' (w) the higher the probability of electing a cluster head. The distance between the candidate node and the base station also influences the cluster head election, the selection of the node closer to the base station as the cluster head is beneficial to reducing the energy loss of the cluster head, and the distance factor between the candidate node and the base station can be expressed asWherein d is_toBS(w) is the distance from the node w to the base station, and f 'is the closer the node w is to the base station'₂The larger (w) is, the higher the probability of electing the cluster head is. The relative distance of the nodes can be expressed asWherein d is_a(w, w ') is the distance, f', between node w and the other nodes w 'in the cluster'₃The larger (w) the smaller the relative distance between nodes, the less the communication cost between nodes. Node base in clusterAnd (4) maximally performing distributed cluster head election.

Step seven: and (4) performing dormancy scheduling on redundant nodes of the wireless sensor network.

If a certain redundant node successfully selects a cluster head in a dynamic cluster head competition stage, scheduling the redundant node as a working mode; if a certain redundant node fails in the competition stage of the dynamic cluster head, the redundant node is scheduled to be in a sleep mode, the node only monitors a wakeup beacon in the network and sends a message to the cluster head of the home cluster to inform the node to enter the sleep state, and the message comprises: the node self communication identification ID, the node home cluster ID, the node self residual energy and the node working mode identification.

Step eight: the wireless sensor network stabilizes data transmission and updates the network state.

In order to facilitate the forwarding transmission and energy consumption of analog data in a network, the single-hop communication between the nodes in the cluster and the base station or the cluster head is set, and the single-hop communication between the cluster head and the base station is set. The method comprises the steps that a working node meeting the condition that a communication mode identifier is 0 in a cluster sends a joining request to a cluster head, the request comprises self residual energy, a self communication identifier ID, an attribution cluster ID and a self working mode identifier, the cluster head node divides time slots according to the number of the received joining requests, namely, a TDMA mechanism is adopted to distribute corresponding time slots to each requesting member, each requesting member collects data in the time slot of the requesting member and beats the data into a communication data packet, and the communication data packet and a control data packet comprising the self communication identifier ID, the self residual energy and the self working mode identifier are sent to the cluster head together; the working nodes meeting the communication mode identifier 1 in the cluster directly send data to the base station; and the cluster head node receives the data sent by the request members, performs fusion compression processing, and packs and sends all the data to the base station after adding the communication identification ID, the node residual energy and the node working mode identification of all the survival members in the cluster into the last frame. The data routing energy loss is specifically that if a certain node n is a cluster head, the energy loss of stable data transmission of the node n isWherein, in the step (A),the number of nodes which are in the working mode in the cluster to which the cluster head belongs and have the communication mode mark of 0; if a certain node n is a cluster working node and satisfiesχ_nWhen it is equal to 0, its stable data transmission energy lossWherein the content of the first and second substances,the cluster head is a cluster to which the node belongs; if a certain node n is a cluster working node and satisfiesχ_n1, its stable data transmission energy lossAfter a surviving node carries out stable data transmission, some nodes may die, if the condition is metThen the node working mode identification is updatedAnd identifying the node death, and further updating the network state.

Step nine: and judging the operation termination of the wireless sensor network.

Judging whether the network runs to the maximum running wheel number K or all nodes disappear, namely judging whether K +1 is equal to K or notWhether the current time is equal to zero or not, if yes, the operation of the wireless sensor network is terminated; if not, making k equal to k +1, returning to the step three, and performing the next round of operation by the wireless sensor network.

The method comprises the steps that N-200 sensor nodes in a network are randomly and statically deployed in a monitoring area of 200m multiplied by 200m in a high-density mode, the maximum running wheel number K of the network is 200, and the sensing radius r of the nodes is set to be r_net10m, node initial energy p_net0.04J communication data packet l_data4000bit control packet l_con100bit, base station position vectorData fusion compression ratio gamma of 0.4, E_elec＝50nJ/bit、E_da＝5nJ/bit、ε_fs＝10pJ/(bit·m²)、ε_mp＝0.0013pJ/(bit·m⁴) The simulation experiment frequency is 50, and the experimental results are counted and averaged.

In fig. 2(a) -5, the wireless sensor network node dormancy scheduling method based on the FCM cluster topology designed by the present invention is denoted as FCNS; the node dormancy scheduling method proposed in the research of the wireless sensor network clustering topology coverage area node scheduling optimization algorithm published by Supper et al in the journal of the technical science of sensing is recorded as NSACT; the node dormancy scheduling method proposed in "wireless sensor network node scheduling optimization clustering algorithm research" published by dawn et al in "computer application research" is referred to as CANS. In the designed FCNS, the maximum iteration number G of the FCM algorithm is set to be 1000, and an iteration termination factor epsilon is set to be 10^-6Fuzzy factor m is 2, network coefficient l_net0.2; in NSACT and CANS, the cluster head probability is set to 0.08. FIG. 2(a), FIG. 2(b), and FIG. 2(c) are diagrams showing the results of network clustering and node dormancy scheduling, respectively, when the NSACT, CANS, and FCNS methods are run in the first round of the network; fig. 2(d), fig. 2(e), and fig. 2(f) are respectively a result of network clustering and node dormancy scheduling when the NSACT, CANS, and FCNS methods operate for 80 rounds of network k. In fig. 2(a) - (f), the base station, the cluster head, the intra-cluster working node, and the intra-cluster sleeping node are denoted by "green +", "red", "black", and "black", respectively, and the cluster head is linked with the intra-cluster working node by a solid line and with the intra-cluster sleeping node by a dotted line for distinction. The network lifecycle is defined as the number of running rounds of the network from the start of operation to the appearance of the first dead node. As can be seen from fig. 2(a) -5, the wireless sensor network node dormancy scheduling method based on the FCM cluster topology, which is designed by the present invention, can comprehensively reduce WSN network level and node level energy consumption, prolong network life cycle and total network operation time, and can be used as a node dormancy scheduling protocol of a high-density static sensor network, under the condition that certain coverage quality is satisfied.