CN116033518A

CN116033518A - Heterogeneous cognitive sensor network-based double-cluster head clustering routing method

Info

Publication number: CN116033518A
Application number: CN202211369612.9A
Authority: CN
Inventors: 宋晓莹; 张启龙
Original assignee: Dalian Neusoft University of Information
Current assignee: Dalian Neusoft University of Information
Priority date: 2022-11-03
Filing date: 2022-11-03
Publication date: 2023-04-28

Abstract

The invention discloses a dual-cluster-head clustering routing method based on a heterogeneous cognitive sensor network, which comprises a plurality of nodes, namely cognitive nodes, common nodes and base stations, wherein the nodes comprise the cognitive nodes, the common nodes and the base stations, a multi-criterion decision method of an intuitive fuzzy analytic hierarchy process is adopted to determine the main cluster head, cluster units are established according to competition radiuses, slave cluster heads are determined, after the cluster units are established, in a data transmission stage, the nodes of each cluster unit directly transmit perceived data to the main cluster head of the cluster in a single-hop manner, and the slave cluster head and the master cluster head of the next hop transmit the data to the base stations in a multi-hop transmission manner, so as to construct a routing algorithm based on frequency spectrum detection until the data are transmitted to the base stations. The method solves the problem that the energy consumption of cluster heads and member nodes is seriously unbalanced because only one cluster head exists in a cluster unit constructed by the existing cluster routing algorithm. In addition, the existing cluster routing algorithm does not consider the problem of reliability of transmission data.

Description

Heterogeneous cognitive sensor network-based double-cluster head clustering routing method

Technical Field

The invention relates to the field of cognitive sensor networks, in particular to a dual-cluster head clustering routing method based on a heterogeneous cognitive sensor network.

Background

The cognitive sensor network is a novel network formed by introducing a cognitive radio technology into a wireless sensor network. The wireless sensor network has no cognitive module, and the lack of spectrum resources can lead sensor nodes to mutually compete for channel resources, thereby leading to transmission conflict, causing loss of data packets, leading to network delay and consuming additional energy. The cognitive sensor network is provided with a cognitive module, and idle authorized spectrum can be dynamically selected to transmit data. When the main user returns to the authorized channel, the cognitive sensor network can actively back off and switch to other idle frequency spectrums, so that the conflict can be effectively avoided, and the energy utilization efficiency of the network is improved. However, cognitive sensor networks are expensive to manufacture and therefore face the dual challenges of high cost and high energy consumption.

The clustering routing algorithm becomes a research hotspot by virtue of the advantages of flexibility, high energy efficiency, simplicity, convenience in management, strong expandability and the like. In wireless sensor networks, many, but no consideration is given to existing research cluster routing algorithms. In a clustering routing algorithm of a cognitive sensor network, most of researches are mainly focused on network scenes of node isomorphism. In the heterogeneous cognitive sensor network, the problems of cluster head overload and data transmission safety are not considered.

The concept of the LEACH algorithm is applied to heterogeneous cognitive sensor networks, and the modified LEACH algorithm is referred to herein as LEACH-G. LEACH is the first proposed routing algorithm with equal clustering. The idea is to randomly select cluster head nodes in a disturbance mode through preset same probability. After each member transmits the perceived data to the cluster head, the cluster head directly transmits all the data of the unit to the Sink. Although this protocol is significantly improved over planar routing algorithms in terms of performance, such as network lifetime, its random perturbation approach causes additional energy consumption. And the direct transmission mode of the data from the cluster head to the Sink can cause the energy consumption of the remote cluster head to be greatly reduced, because the energy consumption of the sensor nodes and the distance between the nodes are in an exponential relationship.

The idea of EEUC is applied to heterogeneous cognitive sensor networks, and the improved EEUC is called EEUC-G. EEUC is an energy-efficient unequal cluster data collection algorithm, which mainly builds an unequal cluster structure according to the distance between nodes and base stations. In EEUC, the cluster size is proportional to the distance of the node from the base station. The cluster size near the base station is smaller than the cluster size far away. The assumed conditions for EEUC are more realistic. For example: the base station and cluster head locations of EEUC need not be precisely deployed. However, EEUC has the following disadvantages: firstly, the c value of the defined cluster head competition radius cannot simply obtain an optimal value, particularly in a large-scale wireless sensor network; second, contention-based cluster preference increases the power consumption of the network because it mainly uses a message negotiation mechanism.

HLEACH is a LEACH algorithm for multi-hop heterogeneous cognitive sensor networks. Firstly, the optimal configuration of the network node quantity ratio and the network node initial energy ratio is deduced theoretically so as to achieve the maximization of energy utilization efficiency. Then, HLEACH considers a plurality of factors such as residual energy, node distance and the like in the process of constructing the cluster, so that the cluster head and the cognitive nodes are distributed more uniformly, and the energy consumption of the network is effectively reduced. However, the cluster unit constructed by HLEACH has only one cluster head, which causes serious unbalance of cluster head and member node energy consumption. And HLEACH does not consider the security of the transmitted data when constructing clusters and multi-hop routing transmissions.

In summary, the cluster unit constructed by the existing cluster routing algorithm has only one cluster head, so that the energy consumption of the cluster head and the member nodes is seriously unbalanced. In addition, the existing cluster routing algorithm does not consider the problem of reliability of transmission data.

Disclosure of Invention

The invention provides a double-cluster-head clustering routing method based on a heterogeneous cognitive sensor network, which aims to overcome the defect that energy consumption of cluster heads and member nodes is seriously unbalanced because only one cluster head exists in a cluster unit constructed by the existing clustering routing algorithm. In addition, the existing cluster routing algorithm does not consider the problem of reliability of transmission data.

In order to achieve the above object, the technical scheme of the present invention is as follows:

a dual-cluster-head clustering routing method based on a heterogeneous cognitive sensor network comprises the following steps:

step 1: a plurality of nodes are randomly deployed in a network monitoring area, wherein the nodes comprise cognitive nodes, sensing nodes and base stations, the cognitive nodes and the sensing nodes can be directly communicated with the base stations, the initial energy of the cognitive nodes is higher than that of the sensing nodes, and the cognitive nodes confirm a main cluster head based on a main cluster preferred comprehensive evaluation hierarchical structure model;

step 2: constructing a competition radius and building cluster units according to node information of cognitive nodes of the main cluster head, wherein the competition radius of the cluster units gradually increases from a base station to a far end, and the node information comprises residual energy of the cognitive nodes, the distance between the cognitive nodes and the base station and the number of neighbor nodes of the cognitive nodes;

determining a slave cluster head in the cluster unit, wherein the cluster unit comprises a master cluster head and a slave cluster head, and the master cluster head is used for performing spectrum sensing, receiving channel monitoring results of non-cluster head cognitive nodes, deciding available channels, broadcasting and receiving data of the cluster unit in the cluster unit and fusing the data;

The slave cluster head is used for relaying other cluster unit data and forwarding the cluster unit data;

step 3: when each cluster unit is built, taking the cluster unit at the far end of the base station as a first-hop cluster unit for data transmission, wherein the data transmission comprises an intra-cluster transmission sub-stage of single-hop transmission and an inter-cluster transmission sub-stage of multi-hop transmission;

the single-hop transmission is that all member nodes except the slave cluster head in the cluster unit directly transmit the perceived data to the master cluster head in the cluster unit;

the multi-hop transmission is that a master cluster head in a cluster unit receives sensing data of all nodes except a slave cluster head and data of the slave cluster head in a previous-hop cluster unit, compresses and acquires a data packet and sends the data packet to the slave cluster head node in the cluster unit;

step 4: the slave cluster head node transmits the data packet to a main cluster head with the minimum routing cost in other cluster units in a multi-hop transmission mode to serve as a relay node until the data packet is transmitted to a base station to finish one round of data transmission;

step 5: after a round of data transmission, judging whether the residual energy of the main cluster head of each cluster unit is lower than a preset threshold value based on an energy consumption model, and if the residual energy of the main cluster head of each cluster unit is greater than the preset threshold value, carrying out reconstruction of the cluster unit without carrying out the complete data transmission process;

And if the residual energy of the main cluster head of one of the cluster units is smaller than a preset threshold value, repeating the steps 1 to 4.

Further, the specific step of determining the master cluster head in the step 1 is:

step 1.1: establishing a multi-criterion decision model in a cognitive sensor network, wherein the cognitive node establishes a main cluster first-choice comprehensive evaluation hierarchical structure based on the multi-criterion decision model, and obtains a main cluster first-choice attribute evaluation value h and a main cluster first-choice attribute importance degree value g;

the multi-criterion decision model comprises a main criterion layer and a sub-criterion layer, wherein the main criterion layer comprises a cognitive node service quality condition evaluation model, a cognitive node energy condition evaluation model and a cognitive node position condition evaluation model; the sub-criterion layer comprises a channel detection probability evaluation model, a transmission reliability evaluation model, a residual energy evaluation model, a message cost evaluation model, a neighbor node number evaluation model and a base station distance evaluation model;

step 1.2: in a network initialization stage, the cognitive node and the sensor node broadcast an INITIAL_MESSAGE MESSAGE to a base station, wherein the INITIAL_MESSAGE MESSAGE comprises ID data, communication radius data and residual energy data of the cognitive node and the sensor node, and the base station establishes an initialization node information table based on the INITIAL_MESSAGE MESSAGE;

The cognitive node also broadcasts the currently detected available channel information, and the cognitive node and the sensing node acquire the distance between the current cognitive node/sensing node and the neighbor node according to the initialized node information table, store the distance and construct a node neighbor table;

step 1.3: after the initialization stage is finished, each cognitive node obtains expected parameters of each cognitive node according to the node neighbor table of the cognitive node, wherein the expected parameters comprise the number of neighbor nodes, the maximum residual energy, the minimum residual energy, the maximum/minimum distance information between the cognitive node and a base station in a network, available channel information and buffer space information, and the number information of sending and receiving messages;

step 1.4: the cognitive node obtains an attribute evaluation value h based on the expected parameter, wherein the attribute evaluation value h comprises a channel detection probability evaluation value F _d Transmission reliability evaluation value R _CNi Residual energy evaluation value E _CNi Cognitive message cost assessment value M _CNi The neighbor cognitive node number evaluation value N and the cognitive node and base station distance evaluation value Di;

step 1.5: the cognitive node acquires an attribute importance degree value g of the comprehensive evaluation of the main cluster head according to an intuitive fuzzy analytic hierarchy process and the comprehensive evaluation hierarchical structure of the main cluster first choice, and integrates the attribute evaluation value h and the attribute importance degree value g based on a fuzzy integral calculation formula to acquire a comprehensive evaluation value PCS which becomes the main cluster head _i ；

Step 1.6: comprehensive evaluation value PCS of the main cluster head _i Obtaining time T of cognitive node becoming master cluster head based on time formula of broadcasting master cluster head _CNi The time formula is as follows:

wherein, is provided with

A random number of 0.8 to 1; PCS (PCS) _i Obtaining a comprehensive evaluation value which becomes a main cluster head for the cognitive node through a multi-criterion decision model; t (T) _O Indicating a preset clustering time.

Further, the specific step of obtaining, by the cognitive node in the step 1.5, the attribute importance degree value g of the primary cluster head comprehensive evaluation according to the intuitive fuzzy analytic hierarchy process and the primary cluster preferred comprehensive evaluation hierarchical structure includes:

obtaining comprehensive opinion score according to the primary cluster preferred comprehensive evaluation hierarchical structure, establishing an intuitionistic fuzzy complementary judgment matrix,

obtaining a fuzzy approximation judgment matrix according to the intuitionistic fuzzy complementary judgment matrix,

consistency test is carried out according to the fuzzy approximation judgment matrix to obtain a satisfactory consistency matrix;

and obtaining a combination weight value of each attribute of the sub-criterion layer relative to the target layer according to the satisfaction consistency matrix, wherein the combination weight value is an attribute importance degree value g.

Further, the calculation formula rcn_ic of the contention radius in step 2 is:

Wherein rcn_ic is the broadcast radius when competing for the master cluster head; ecurr_i is the current remaining energy of the cognitive node i; emax is the maximum residual energy of the cognitive node in the node neighbor table of the cognitive node i; dmax is the maximum distance between the cognitive node and the base station in the cognitive network; di, sink is expressed as the distance from the cognitive node i to the base station; ni is the number of neighbor cognitive nodes of the cognitive node i; n is the total number of nodes in the cognitive network, the parameter α is more than or equal to 0 and less than or equal to 1, β is more than or equal to 0 and less than or equal to 1, and α+β=1.

Further, in step 3, in each cluster unit, the slave cluster head is determined, specifically:

step 3.1: the cognitive node determines the broadcasting range through the time formula and the calculation formula of the competition radius, and then issues a message which becomes a main cluster head in the cluster unit;

step 3.2: other cognitive nodes in the cluster unit acquire the distance from the main cluster head according to the signal of the received main cluster head MESSAGE, and issue a SE_CH_MESSAGE MESSAGE to the main cluster head;

step 3.3: after the master cluster head receives the SE_CH_MESSAGE MESSAGE, determining that the cognitive node closest to the master cluster head in the cluster unit is a slave cluster head, wherein the SE_CH_MESSAGE MESSAGE comprises the current ID, the residual energy and the position information of the cognitive node;

Step 3.4: other member nodes of the cluster unit send a join_message JOIN MESSAGE to a main cluster head, wherein the join_message JOIN MESSAGE comprises a node ID and current residual energy;

the master cluster head determines member nodes according to the received JOIN_MESSAGE joining MESSAGE, the member nodes do not send the JOIN_MESSAGE joining MESSAGE to the master cluster heads of other cluster units after being determined, the master cluster head presets a data transmission time slot for the member nodes to generate a TDMA time table, the TDMA time table is broadcasted back to all members in the cluster, and the TDMA time table is used for the sequence of sending data to the master cluster heads by the member nodes of the cluster units in the data collection process;

if an isolated cognitive node exists in the network, namely, a cognitive node which does not belong to any cluster unit, the isolated cognitive node automatically becomes a master cluster head, and the master cluster head transmits data to a slave cluster head nearest to the isolated cognitive node as a relay node according to a node neighbor table;

if an isolated common sensor node exists in the network, the isolated common sensor node is automatically added into a cluster unit where the nearest sensor node is located, becomes a member node of the cluster unit, and transmits data to a main cluster head of the cluster unit.

Further, in step 4, a routing algorithm based on spectrum detection is constructed, specifically:

step 4.1: broadcasting an ne_message MESSAGE at twice the communication radius from the cluster head of each cluster unit;

step 4.2: the master cluster head receives the NE_MESSAGE MESSAGE and sends an AS_MESSAGE MESSAGE to the slave cluster head to establish a neighbor node information table, wherein the AS_MESSAGE MESSAGE comprises a master cluster head ID, the current available channel number of the master cluster head, the position of a base station, the current residual energy of the master cluster head and the number of neighbor cognitive nodes of the master cluster head;

step 4.3: the slave cluster head selects a master cluster head with the minimum routing cost value as a next-hop cognitive node of the slave cluster head according to the neighbor node information table, and a routing cost function formula is as follows:

the I channel T is the total number of available channels of all cognitive nodes in the neighbor node information table; the I channel cur_j is the current available information number of the primary cluster head j; dj, sink is the distance from the primary cluster head j to the base station; dmax_j is the maximum distance from the main cluster head to the base station in the neighbor information table; nj is the number of neighbor cognitive nodes of the master cluster head j; n is the total number of nodes in the cognitive network, the parameter is more than or equal to 0 and less than or equal to 1, and gamma+θ=1;

Step 4.4: in the cluster unit of the next hop node, repeating the steps 4.1 to 4.3 until the cluster unit is transmitted to the base station.

The invention has the beneficial effects that:

the invention provides a heterogeneous cognitive sensor network double-cluster head clustering routing algorithm based on multi-criterion decision. Firstly, determining a main cluster head by adopting a multi-criterion decision method under an intuitive fuzzy analytic hierarchy process in the process of competing the main cluster head by a cognitive sensor node, and enabling factors which become the main cluster head to be considered to be more comprehensive in the process of competing the main cluster head; secondly, in the process of constructing a non-uniform cluster unit, a selection scheme of a slave cluster head and an adjustable master cluster head competing radius are provided, so that the slave cluster head is suitable for a network with nodes distributed randomly and nodes heterogeneous; finally, in the cluster head transmission stage, a routing method based on spectrum detection is provided, so that the reliability of data transmission is improved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions of the prior art, the drawings that are needed in the embodiments or the description of the prior art will be briefly described below, it will be obvious that the drawings in the following description are some embodiments of the present invention, and that other drawings can be obtained according to these drawings without inventive effort to a person skilled in the art.

FIG. 1 is a flow chart of a dual cluster head clustering routing method based on a heterogeneous cognitive sensor network;

FIG. 2 is a flow chart of a dual cluster head clustering routing method based on a heterogeneous cognitive sensor network;

FIG. 3 is a block diagram of a primary cluster preferred comprehensive evaluation hierarchy of a dual cluster head cluster routing method based on a heterogeneous cognitive sensor network;

FIG. 4 is a flow chart of a dual cluster head clustering routing method based on a heterogeneous cognitive sensor network;

fig. 5 is a network life simulation experiment comparison chart of a dual cluster head clustering routing method based on a heterogeneous cognitive sensor network;

fig. 6 is a comparison chart of average channel detection probability simulation experiments of a dual cluster head clustering routing method based on a heterogeneous cognitive sensor network;

fig. 7 is a comparison diagram of network energy consumption simulation experiments of a dual cluster head clustering routing method based on a heterogeneous cognitive sensor network;

fig. 8 is a comparison diagram of stability simulation experiments of a dual cluster head clustering routing method based on a heterogeneous cognitive sensor network.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

The dual-cluster head clustering routing method based on the heterogeneous cognitive sensor network, as shown in fig. 1 to 2, comprises the following steps:

In a specific embodiment, as shown in fig. 3, the specific steps for determining the primary cluster head in step 1 are:

The fuzzy integral calculation formula is as follows:

the fuzzy measure of the sub-criterion set X is the set function g P (X) → [0,1] satisfying the following axiom.

(1) g (Φ) =0, g (X) =1 (boundary condition)

(2)If A,

the g (A) is less than or equal to g (B) (monotonicity)

Let g _λ A special fuzzy measure is defined for the lambda fuzzy measure and a power set P (X) of a finite set X, while also satisfying the finite lambda rule. Sugeno introduced a so-called lambda-fuzzy measure to satisfy the following additional properties:

g _λ (A∪B)＝g _λ (A)+g _λ (B)+λg _λ (A)g _λ (B),λ∈(-1,∞)

according to g _λ Is defined by a finite set { x } ₁ ,x ₂ ,…,x _n Mapping to function g _λ Can use the blurring strength g _i ＝g _λ ({x _i }) representation, noted as

Lambda-fuzzy measure g _λ The value lambda may pass the boundary condition g _λ (X) =1 solution;

let g denote the fuzzy measure on X, h is from X to [0,1]]Is a measurable function of (a). Let h (x) ₁ )≥h(x ₂ )≥…≥h(x _n ) Then the following Sugeno fuzzy integral calculation formula can be constructed:

wherein H is ₁ ＝{x ₁ },H ₂ ＝{x ₁ ,x ₂ },…,H _n ＝{x ₁ ,x ₂ ,…,x _n }＝X。

In practical applications, h may be considered as the performance of a particular attribute of the candidate, while g represents the relative importance of each attribute.

in order to prevent the messages broadcasted by a plurality of cognitive nodes as the main cluster head from being too close, and avoid message collision, a method is provided

In a specific embodiment, the specific steps of the attribute evaluation value h in the step 1.4 are:

step 1.3.1: the attribute of the channel detection probability is based on a fusion rule of channel perception results, if only one detection result of the cognitive node is that the current channel is occupied by a main user, the channel is judged to be unavailable, the cognitive node jointly perceives the detection probability of the channel, and the available channel information obtains a channel detection probability evaluation value Fd based on a channel detection probability evaluation model, wherein the channel detection probability evaluation value Fd is as follows:

wherein P is _d The method refers to the probability that a sensing node detects that the channel state is busy when a main user occupies the channel;

step 1.3.2: during data transmission, if cognitive nodes in the network are exhausted in energy or suffer physical damage to cause cognitionThe node failure can affect the receiving condition of the data packet, the cognitive node is used as a master cluster head and is mainly responsible for receiving the data of member nodes and other slave cluster heads, so that the probability of the cognitive node with high reliability as the master cluster head is higher, and the buffer space information obtains the evaluation value R of the transmission reliability of the cognitive node based on a transmission reliability evaluation model _CNi The method comprises the following steps:

wherein B is _ava (CN _i ) Representing buffer space available to the cognitive node i; b (B) _total (CN _i ) Representing the total space of the cache region of the cognitive node i, and R _CNi The larger the value of (2), the higher the reliability of the cognitive node i;

step 1.3.3: the cognitive nodes and the sensor nodes are provided with energy modules, and the battery modules are inconvenient to replace according to the characteristics of the network, so that the energy is the most important and scarce resource in the heterogeneous cognitive sensor network, and the number of the neighbor nodes, the maximum residual energy and the minimum residual energy are based on a residual energy evaluation model to obtain a cognitive node residual energy evaluation value, namely an ith cognitive node residual energy evaluation value E _CNi The method comprises the following steps:

wherein E is _max And E is connected with _min Maximum and minimum residual energy in neighbor cognitive nodes of the cognitive node i respectively, er is the residual energy of the cognitive node i, and E _CNi The higher the value of (2), the smaller the energy limit of the cognitive node;

step 1.3.4: in the competition process of the master cluster head, the master cluster head is generated in the cognitive node, the non-uniform cluster structure is realized by adopting a message exchange mode in the cognitive candidate node, according to the energy consumption model, the message cost directly influences the energy consumption of the cognitive node, and the sending and receiving of the message The number information is based on a message cost evaluation model to obtain an evaluation value M of cognitive message cost _CNi The method comprises the following steps:

wherein M is _C The total number of messages received and sent for the cognitive node i; m is M _T Total number of received and sent messages for all its neighbor nodes, and M _C The larger the value, the message cost M _CNi The lower;

step 1.3.5: the cognitive nodes and the sensor network are deployed in the network in a random mode, the characteristic is that some regional nodes are distributed more densely, some regional nodes are distributed more sparsely, so that the number of neighbor nodes in one cluster unit is inconsistent, when the number of the cognitive nodes in the neighbor nodes of one cognitive node is larger, the probability that the cognitive node becomes a cluster head is larger, the number of the neighbor nodes of the cognitive node is based on a neighbor node number evaluation model, and the neighbor node number evaluation value N is obtained as follows:

wherein, ntotal is the total number of cognitive nodes in the network, and Ni is the number of neighbor cognitive nodes of the cognitive node i;

step 1.3.6: the distance between the cognitive node and the base station is also a key factor to be considered in cluster preference, because the distance has a direct relation with energy consumption, the larger the distance between the cognitive node and the main cluster head is, the smaller the probability of becoming the main cluster head is relatively, the maximum/minimum distance information between the cognitive node and the base station in the network is based on a base station distance evaluation model, and the evaluation value Di of the distance between the cognitive node and the base station is obtained by:

Wherein dmax represents the maximum distance between a cognitive node in the network and a base station; dmin represents the minimum distance between a cognitive node in the network and a base station; d (i, sink) represents the distance between the cognitive node i and the base station; and when the distance between the cognitive node i and the base station is larger, the Di value is smaller.

In a specific embodiment, the step 1.5 of calculating the attribute importance degree value g of the primary cluster head comprehensive evaluation by using the intuitive fuzzy analytic hierarchy process and the primary cluster first comprehensive evaluation hierarchical structure according to the multi-criterion decision model includes the following specific steps:

performing consistency test and adjustment according to the fuzzy approximation judgment matrix to obtain a satisfactory consistency matrix; the consistency test and adjustment method of the fuzzy approximation judgment matrix adopts a consistency test adjustment and sequencing method of the fuzzy complementation judgment matrix given in the prior literature (consistency test and correction [ J ] of the fuzzy complementation judgment matrix, fuzzy system and mathematics, 2010, 24 (2): 105-111);

As shown in fig. 4, first, according to the primary cluster preferred comprehensive evaluation hierarchical structure, the importance of each attribute index of the primary criterion layer L2 and the secondary criterion layer L3 with respect to the target layer L1 (primary cluster head comprehensive evaluation layer) is compared in pairs, expert scoring is performed to obtain an expert comprehensive opinion score, and an intuitive fuzzy complementation judgment matrix is established, where the intuitive fuzzy complementation judgment matrix of the primary cluster head comprehensive evaluation layer L1 by the cognitive node service quality condition attribute L21, the cognitive node energy condition attribute L22 and the cognitive node position condition attribute L23 of the L2 criterion layer is:

the intuitive fuzzy complementary judgment matrix of the channel detection probability attribute L31, the transmission reliability L32, the residual energy attribute L33, the message cost attribute L34, the neighbor node attribute L35 and the cognitive node and base station distance attribute L36 of the sub-criterion layer L3 for the cognitive node service quality condition attribute L21, the cognitive node energy condition attribute L22 and the cognitive node position condition attribute L23 is as follows:

/>

secondly, obtaining a fuzzy approximation judgment matrix of the intuitional fuzzy complementation judgment matrix as follows:

Then, consistency test and adjustment are carried out on the fuzzy approximation judgment matrix to obtain a satisfactory consistency matrix, wherein the consistency matrix is as follows:

if P _ij For matrix of fuzzy consistency complementary judgment, P is _ij ＝β(w _i -w _j ) +0.5, (β. Gtoreq. (n-1)/2 and is constant), if P _ij If the values are inconsistent, the above formula is not established, and when β= (n-1)/2 solution:

s.t.

the weight formula of (2) is:

wherein w is _i The weight of the influence of the ith attribute on the factor of the last layer is given; p (P) _ij The element of the ith row and the jth column of the fuzzy consistency complementary judgment matrix; n is the order of the fuzzy consistency complementary judgment matrix.

In a specific embodiment, in the step 2, the range of the broadcast of the cognitive node is mainly controlled by the contention radius, and the division is established according to the relationship between the current remaining energy of the cognitive node, the distance between the cognitive node and the base station, and the number of owned neighbor cognitive nodes, wherein the greater the current remaining energy of the cognitive node, the farther the distance between the cognitive node and the base station, the fewer the number of owned neighbor cognitive nodes, the greater the contention radius, and vice versa, and the calculation formula rcn_ic of the contention radius is:

wherein rcn_ic is the broadcast radius when competing for the master cluster head; ecurr_i is the current remaining energy of the cognitive node i; emax is the maximum residual energy of the cognitive node in the node neighbor table of the cognitive node i; dmax is the maximum distance between the cognitive node and the base station in the cognitive network; di, sink is expressed as the distance from the cognitive node i to the base station; ni is the number of neighbor cognitive nodes of the cognitive node i; n is the total number of nodes in the cognitive network, the parameter alpha is more than or equal to 0 and less than or equal to 1, beta is more than or equal to 0 and less than or equal to 1, and alpha+beta=1;

And after receiving the message that the current cognitive node broadcasts itself to be the master cluster head, other cognitive nodes in the radius formula range automatically give up the behavior of competing itself for the master cluster head.

In a specific embodiment, in each of the cluster units described in step 3, the slave cluster head is determined, specifically:

step 3.1: after the cognitive node determines the broadcasting range by the time formula and the calculation formula of the competition radius, the cognitive node which issues the primary cluster head message earliest becomes the primary cluster head;

step 3.2: in the range of the competition radius, as long as one main cluster head is used, other cognitive nodes in the range of the competition radius can obtain the distance between the other cognitive nodes and the main cluster head according to the signal intensity of the received main cluster head MESSAGE, and release SE_CH_MESSAGE MESSAGE to the main cluster head;

step 3.3: after receiving the SE_CH_MESSAGE MESSAGE of the other cognitive nodes within the competition radius range, the master cluster head selects a nearest cognitive node from the master cluster head through comparison, wherein the nearest cognitive node is a slave cluster head, and the SE_CH_MESSAGE MESSAGE comprises the current ID, the residual energy and the position information of the cognitive node; if the distances between two cognitive nodes and the master cluster head are the same, the master cluster head randomly selects any one of the cognitive nodes as a slave cluster head after collecting data;

Step 3.4: the other member nodes of the cluster unit receive the main cluster head message; and transmitting a JOIN MESSAGE to the master cluster head, the JOIN MESSAGE including a node ID and a current remaining energy;

In a specific embodiment, in step 3, the main cluster head directly transmitted to the present cluster in a single-hop manner is specifically: the data transmission stage comprises two sub-stages, namely an intra-cluster transmission stage and an inter-cluster transmission stage, in the intra-cluster transmission stage, after each main cluster head collects the data of the member node, the data is sent to the main cluster head in a single-hop transmission mode, namely each non-cluster head node is distributed with a preset time slot, and when the time slot reaches the non-cluster head node, the member node transmits the data to the main cluster head node.

In a specific embodiment, the constructing a routing algorithm based on spectrum detection in step 4 specifically includes:

step 4.1: broadcasting an ne_message MESSAGE by the slave cluster head of each cluster at twice the communication radius;

step 4.3: the slave cluster head calculates the routing cost of each cognitive node according to the neighbor node information table, and selects the master cluster head with the minimum routing cost as the next-hop node of the slave cluster head, wherein the cost function formula is as follows:

The I channel T is the total number of available channels of all cognitive nodes in the neighbor node information table; the I channel cur_j is the current available information number of the primary cluster head j; dj, sink is the distance from the primary cluster head j to the base station; dmax_j is the maximum distance from the main cluster head to the base station in the neighbor information table; nj is the number of neighbor cognitive nodes of the master cluster head j; n is the total number of nodes in the cognitive network; the parameter is 0-gamma-1, 0-theta-1 and gamma+theta=1;

step 4.4: and repeating the steps 4.1 to 4.3 in the cluster of the next hop nodes until the cluster is transmitted to the base station.

In a specific embodiment, the data fusion model in step 2 is specifically:

if one node fuses 1bit data, the consumed energy is E _fusion When m data packets are fused into one data packet, the energy consumption formula is as follows:

E _f (m,l)＝mlE _fusion

in a specific embodiment, the energy consumption model in step 2 is specifically:

the energy consumption formula Et at the time of transmission is:

the energy consumption formula Er (l) at the time of reception is:

E _r (l)＝lE _elec

wherein E is _elec Indicating energy lost by the transmitting circuit if the transmission distance is less than the threshold d ₀ The power amplification loss adopts a free space model; if the transmission distance is greater than or equal to the threshold d ₀ In the process, a multipath attenuation model is adopted; epsilon _fs ，ε _amp Respectively representing the energy required for power amplification in the two models, and l represents the bit data received by the node.

Simulation experiment and analysis results

The running environment of the simulation experiment is an Intel Pentium dual-core (2.2 GHz) processor and a 2G memory, and four clustering routing algorithms are realized by adopting a Matlab2013b experiment platform, and the method comprises the following steps: the methods herein are HLEACH, EECU-G and LEACH-G. The four algorithms are compared in three performances of network service life, average channel detection probability and energy consumption of a main cluster head. Assume that in a 200m×200m network monitoring scenario, 200 cognitive nodes and common sensor nodes are deployed. Assuming that the available channel is 5, the initial energy of the cognitive node is 1J, the initial energy of the common sensor node is 0.5J, and other simulation parameters are shown in table 1:

TABLE 1 simulation parameters

Parameters (parameters)	Value of	Parameters (parameters)	Value of
				Base station coordinates	(250,100)	E _elec	50nj/bit
Cognitive node and common node	200	ε _fs	10pJ/(bit×m ² )
				Number of master users	5	E _fusion	5nj/bit
Number of available channels	5	ε _mp	0.0013pJ/(bit×m ⁴ )
				Initial energy of cognitive node	1J	P _d	0.7
Initial energy of common node	0.8J	d ₀	87m
				E _SEN	0.5J/bit	Data packet size	4000bits

As shown in fig. 5, the number of surviving nodes of four algorithms of cognitive nodes and common sensor nodes is compared (i.e. a network life simulation experiment comparison graph), and the number of surviving nodes of the algorithm of the present application is the largest along with the increase of the running time in the common sensor network, and the cognitive nodes also survive more, mainly because the cluster unit of the method has two cluster heads, namely, the main cluster head and the secondary cluster head, each cluster head bears the corresponding separation of tasks, thereby reducing the energy consumption. The number of common nodes and cognitive node survival nodes of the LEACH-G algorithm is the minimum, and mainly because after the cluster head of the algorithm collects data, the data is directly sent to the base station, so that the energy consumption is overlarge. EEUC-G has more surviving nodes than LEACH-G mainly because of non-uniform clustering, smaller cluster scale near the base station, larger cluster scale far from the base station, and multi-hop mode transmission of the inter-cluster route of EEUC-G. The number of common node and cognitive node survival nodes of HLEACH-G is greater than that of the first two algorithms, mainly because the initial energies of HLEACH-G and the methods herein are heterogeneous, which results in some nodes with high initial energies, resulting in an overall increase in the number of survival nodes over the first two. The number of common nodes and cognitive node survival nodes of HLEACH-G is not high, mainly because the cluster unit of HLEACH-G only has one cluster node, and the factors considered when the cluster head of the cluster node competes are not comprehensive.

As shown in fig. 6 and fig. 7, cluster average channel detection probability and network energy consumption (i.e., average channel detection probability and network energy consumption simulation experiment contrast graph) of four algorithms in 15 rounds of random selection are shown, respectively. As can be seen from fig. 7, the method herein has a higher average channel detection probability than the other three algorithms in the randomly selected 15 rounds of time. In 15 rounds selected randomly, the average channel detection probability of the method is greater than 80% and higher than that of other three algorithms, and mainly, the method not only considers the channel detection probability but also considers the distribution condition of cognitive nodes in the cluster when constructing cluster scale and multi-hop route transmission. Therefore, compared with other algorithms, the method provides better balance on the distribution of the cognitive nodes, so that the channel detection probability among clusters is higher and more stable. Although the HLEACH also considers the distribution situation of the cognitive nodes in the cluster, the factor of the channel detection probability of the cognitive nodes is not considered when selecting the cluster head and establishing the multi-hop route between clusters, and therefore, the average channel detection probability of the HLEACH is lower compared with the method. In addition, in the random 15 rounds, the average channel detection probability of EEUC-G is lower than that of the method and HLEACH, but higher than that of LEACH-G, mainly because EEUC-G considers the distance factor between the candidate cluster head and the base station and the multi-cluster head processing mechanism in the same cluster when competing for cluster heads, so that the distribution of CH in EEUC-G algorithm is more uniform. Therefore, the cognitive nodes of each cluster are distributed more uniformly, and a higher and stable channel detection probability is provided. The location of the cluster head in the LEACH-G algorithm is random, so the channel detection probability of the LEACH-G algorithm is extremely unstable.

As shown in fig. 6, the energy consumption of the algorithm in the random 15 rounds is the minimum, mainly because the dual cluster head strategy is adopted in the present invention, the task of the cluster head is effectively decomposed, so that the energy consumption of the whole network is reduced. Meanwhile, the reestablishment of unequal clusters is not needed in each round, and the energy consumed by reestablishing the cluster structure is saved. The energy consumption of HLEACH is slightly higher than that of HLEACH, mainly because HLEACH adopts a single cluster head mechanism, the cluster head is heavier in load and higher in energy consumption. While EEUC-G has lower energy consumption than LEACH-G, mainly because EEUC-G is a non-uniform clustering algorithm, while LEACH-G is a uniform clustering algorithm. The energy consumption of the non-uniform clustering algorithm is lower than that of the uniform clustering algorithm.

As can be seen from fig. 6 and 7, the cluster average channel detection probability of LEACH-G and the network energy consumption have large fluctuations. This is because the contention cluster heads in the LEACH-G use random numbers as thresholds, which are unevenly distributed, resulting in denser cluster heads in some areas and sparser cluster heads in some areas, thereby increasing the energy consumption in and among clusters. Moreover, the cluster size of LEACH-G is too small, so that the number of cognitive nodes is too small, and the channel detection probability of the cluster is too low, so that the communication requirement of the network cannot be met. Meanwhile, the channel detection error can lead to data retransmission, and further increase the network energy consumption.

As shown in fig. 8, the cluster head number of 15 rounds is randomly taken out for comparison of four algorithms, and the stability of the four algorithms is mainly examined (stability simulation experiment comparison chart). As can be seen from fig. 8, the cluster head number difference of the algorithm of the present invention has the best stability, mainly because the method adopts a timing broadcast mechanism when constructing unequal clusters, so that other candidate cluster heads abandon competition, thereby ensuring the stability of the cluster head number of the whole network. The stable cluster head number can also ensure that each cluster unit has high enough channel detection probability. HLEACH has slightly poorer stability mainly because HLEACH is more suitable for network scenes in which nodes are distributed uniformly, and when HLEACH is changed into network scenes which are distributed randomly and closer to actual nodes, the stability of the algorithm is slightly non-ideal. The EEUC-G algorithm has poor stability, and the cluster head is selected when the cluster head is in competition, and the factors of the residual energy and the distance between the base stations are considered, but the competition mechanism adopts information exchange among multiple nodes, so that the clustering algorithm is unstable. The LEACH-G algorithm has the worst stability, mainly because the determination of the candidate cluster head mainly adopts a random probability distribution mechanism, the residual energy of the candidate cluster head is not considered by factors such as distance from a base station, and the like, so that the energy consumption of the cluster head is too fast, the maximum fluctuation of the cluster head number of the LEACH-G is caused, and insufficient channel detection probability is also brought.

Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the invention.

Claims

1. A dual-cluster-head clustering routing method based on a heterogeneous cognitive sensor network is characterized by comprising the following steps:

Determining a slave cluster head in the cluster unit, wherein the cluster unit comprises a master cluster head and a slave cluster head, and the master cluster head is used for performing spectrum sensing, receiving channel monitoring results of non-cluster head cognitive nodes, deciding available channels, broadcasting in the cluster unit, receiving data of the cluster unit and fusing the data;

2. The dual-cluster-head clustering routing method based on the heterogeneous cognitive sensor network according to claim 1, wherein the specific steps of determining the master cluster head in the step 1 are as follows:

the multi-criterion decision model comprises a target layer, a main criterion layer and a sub-criterion layer, wherein the main criterion layer comprises a cognitive node service quality condition evaluation model, a cognitive node energy condition evaluation model and a cognitive node position condition evaluation model; the sub-criterion layer comprises a channel detection probability evaluation model, a transmission reliability evaluation model, a residual energy evaluation model, a message cost evaluation model, a neighbor node number evaluation model and a base station distance evaluation model;

step 1.4: the sub-criterion layer obtains an attribute evaluation value h based on the expected parameter, the attribute evaluation value h including a channel detection probability evaluation value F _d Transmission reliability evaluation value R _CNi Residual energy evaluation value E _CNi Cognitive message cost assessment value M _CNi The neighbor cognitive node number evaluation value N and the cognitive node and base station distance evaluation value Di;

wherein, is provided with

3. The dual-cluster-head clustering routing method based on the heterogeneous cognitive sensor network according to claim 2, wherein the specific steps of the cognitive node in the step 1.5 obtaining the attribute importance degree value g of the primary cluster head comprehensive evaluation according to the intuitive fuzzy hierarchical analysis method and the primary cluster preferred comprehensive evaluation hierarchical structure are as follows:

4. The method for clustering the double cluster heads based on the heterogeneous cognitive sensor network according to claim 1, wherein the calculation formula rcn_ic of the competition radius in the step 2 is as follows:

5. The method for dual-cluster-head clustering routing based on heterogeneous cognitive sensor network according to claim 1, wherein in each cluster unit in step 3, a slave cluster head is determined, specifically:

step 3.1: the cognitive node determines the broadcasting range of the broadcasting becoming the main cluster head and the competition radius through the time formula, and then issues the message becoming the main cluster head in the cluster unit;

the master cluster head determines member nodes according to the received JOIN_MESSAGE joining MESSAGE, the member nodes do not send the JOIN_MESSAGE joining MESSAGE to the master cluster heads of other cluster units after determining, the master cluster head presets a data transmission time interval for the member nodes, a TDMA time table is generated, the TDMA time table is broadcasted back to all members in the cluster, and the TDMA time table is used for the sequence of sending data to the master cluster heads by the member nodes of the cluster units in the data collection process;

if an isolated common sensor node exists in the network, the isolated common sensor node automatically joins the cluster unit where the nearest sensor node is located to become a member node of the cluster unit, and transmits data to the main cluster head of the cluster unit.

6. The dual-cluster-head clustering routing method based on the heterogeneous cognitive sensor network according to claim 1, wherein the step 4 is specifically: