CN112512003B - Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network - Google Patents

Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network Download PDF

Info

Publication number
CN112512003B
CN112512003B CN202011305178.9A CN202011305178A CN112512003B CN 112512003 B CN112512003 B CN 112512003B CN 202011305178 A CN202011305178 A CN 202011305178A CN 112512003 B CN112512003 B CN 112512003B
Authority
CN
China
Prior art keywords
node
trust
nodes
network
communication
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202011305178.9A
Other languages
Chinese (zh)
Other versions
CN112512003A (en
Inventor
杜嘉欣
韩光洁
林川
王照辉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Dalian University of Technology
Original Assignee
Dalian University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Dalian University of Technology filed Critical Dalian University of Technology
Priority to CN202011305178.9A priority Critical patent/CN112512003B/en
Publication of CN112512003A publication Critical patent/CN112512003A/en
Application granted granted Critical
Publication of CN112512003B publication Critical patent/CN112512003B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
    • H04W4/30—Services specially adapted for particular environments, situations or purposes
    • H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W40/00—Communication routing or communication path finding
    • H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
    • H04W40/04—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources
    • H04W40/10—Communication route or path selection, e.g. power-based or shortest path routing based on wireless node resources based on available power or energy
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W40/00—Communication routing or communication path finding
    • H04W40/02—Communication route or path selection, e.g. power-based or shortest path routing
    • H04W40/20—Communication route or path selection, e.g. power-based or shortest path routing based on geographic position or location
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04W—WIRELESS COMMUNICATION NETWORKS
    • H04W40/00—Communication routing or communication path finding
    • H04W40/24—Connectivity information management, e.g. connectivity discovery or connectivity update
    • H04W40/32—Connectivity information management, e.g. connectivity discovery or connectivity update for defining a routing cluster membership

Abstract

The invention discloses a dynamic trust model based on a long-time memory network in an underwater acoustic sensor network, which is used for identifying malicious nodes in the network and ensuring the accuracy of node monitoring data in water pollution monitoring application.A base station collects relevant information representing communication behaviors and communication capacities among sensing nodes in each period; secondly, calculating four-dimensional trust representing the reliability of the nodes, and constructing a trust data set as the input of the LSTM network; and finally, realizing trust modeling and trust evaluation based on LSTM, detecting malicious sensing nodes in UASNs according to the trust evaluation value, effectively improving the security of network and node monitoring data, and being capable of being deployed in water quality pollution monitoring application to obtain more accurate and reliable water quality monitoring results.

Description

Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network
Technical Field
The invention belongs to the technical field of wireless sensor networks, and particularly relates to a dynamic trust model based on a long-time memory network in an underwater acoustic sensor network.
Background
In recent years, Underwater Acoustic Sensor Networks (UASNs) have become an important support technology in various fields of application in the ocean. Particularly, UASNs have a wide application prospect in water pollution monitoring, wherein a plurality of sensing nodes can be deployed at different positions in the ocean to monitor relevant data of water quality. However, UASNs are usually deployed in an unattended harsh environment, which makes the sensing node vulnerable to attacks, inaccurate or tampered with the monitoring data of the node.
To ensure the accuracy of water quality monitoring data in UASNs, trust models are used to identify malicious nodes under attack. The trust model is realized by firstly collecting trust data related to the communication behavior or the data monitoring capability of the sensing node, then analyzing and processing the trust data based on an abnormal detection algorithm to obtain the trust value of the sensing node, and finally judging the credibility of the sensing node according to the trust value of the node.
At present, researchers at home and abroad have extensively studied malicious node identification based on a trust model in UASNs, and relevant documents are as follows:
to achieve accurate energy-saving Trust assessment in UASNs, Han et al propose An anti-Attack Trust Model (ARTMM) Based on Multidimensional Trust measurement in Underwater Acoustic Sensor Network in An Attack-Resistant Trust Model Based on Multidimensional measurement. ARTMM includes three types of trust metrics, namely link trust, data trust and node trust. In the trust calculation process, the mobility of the underwater environment is considered, and the unreliability of the communication channel is analyzed in detail. In addition, trust updates in underwater mobile environments have also been investigated. Simulation results show that compared with other trust schemes suitable for a land sensor network, the ARTMM is more suitable for an underwater mobile environment, and has higher accuracy and energy consumption performance.
Jiang et al, A Dynamic Trust Evaluation and Update method for communication network Based on C4.5 Decision Tree in an underserver Wireless Sensor network, propose a C4.5 Decision Tree algorithm Based Trust Evaluation and Update Mechanism (TEUC) for an Underwater Wireless Sensor network. In TEUC, first a collection includes data-based, link-based and node-based trust evidence. The collected trust evidence is then used to train the C4.5 decision tree. In addition, reward and penalty factors are defined, updating trust based on a sliding time window. Simulation results show that the algorithm is superior to the traditional algorithm in the aspects of malicious node detection and energy consumption in the dynamic network environment. However, in TEUC, it is assumed that there is no malicious attack in the initial stages of network deployment, and in real underwater applications, this assumption may not make sense because it is difficult to know when an attacker launches an attack.
Because unreliable nodes may cause higher data packet loss, Rajendran et al propose a dynamic Bayesian game model based on a trust strategy in secure transmission using rule-based dynamic Bayesian gate in indirect access sensor networks for solving the problem of secure transmission among nodes in UASNs. By evaluating packet loss and behavioral anomaly activities occurring during data transmission, each node updates the trust value using bayesian rules and analyzes its neighbor nodes based on the trust value. The scheme can reduce packet loss attack and reduce improper communication behaviors of malicious nodes.
Aiming at the problem that a network is easy to attack due to node failure or damage, Arifeen et al put forward a Trust Management Model based on a Hidden Markov Model (HMM) in high Markov Model based Management Model for lower water Wireless Sensor Networks for measuring the credibility of a Sensor node against malicious or internal attacks. The HMM models the dynamic behavior of the sensor node, and the evaluation node can learn the behavior of the evaluated node from past interactions based on the attributes of the HMM. In addition, a lightweight stream cipher scheme is proposed to prevent external attacks.
Disclosure of Invention
In order to identify malicious sensing nodes in UASNs and ensure the accuracy of node monitoring data in water pollution monitoring application, the invention provides a dynamic trust model based on a long-time memory network in UASNs. Firstly, a base station collects relevant information representing communication behaviors among sensing nodes and representing the communication capacity of the sensing nodes in each period, and respectively calculates the communication trust, the data trust, the node trust and the environment trust of each sensing node; then, taking the four-dimensional trust data as the input of a long-time and short-time memory network to construct a trust model; and finally, obtaining a trust value of each sensing node based on the trust model, judging whether the nodes are abnormal or not according to the trust values, and judging whether the water quality is polluted or not by the base station according to the monitoring data of the normal nodes.
The technical scheme of the invention is as follows:
a dynamic trust model based on a long-time memory network in an underwater acoustic sensor network comprises the following steps:
(1) aiming at the considered water pollution monitoring application, different sensing devices are deployed in a monitoring range, node devices are clustered based on a KMeans clustering method, Underwater Acoustic Sensing Networks (UASNs) are constructed, and a network node communication and malicious attack mode is defined;
(1.1) considering an underwater pollution monitoring application scene, deploying different types of sensing equipment in water, wherein the underwater sensing equipment comprises 1 base station and n isomorphic sensing nodes, namely all the sensing nodes have the same energy and communication range, calculating and storing capacity, a deployment area is of a cubic structure, the base station is arranged in the center of the water surface, and the sensing nodes are randomly deployed in the water;
(1.2) dividing n underwater sensing nodes into k clusters based on a KMeans clustering method, wherein each cluster comprises 1 cluster head and a plurality of cluster members, each cluster head is a sensing node with the highest energy in the current cluster, the sensing nodes in a communication range can directly communicate based on an underwater acoustic link, and the sensing nodes beyond the node communication range are communicated in a multi-hop routing mode; due to the fact that the mobility of water flow can cause the change of the node position, the node position is updated by adopting an MCM (scanning Current mobility) model, wherein the node movement mainly comprises downstream flow and circular motion (namely the node is in a vortex);
(1.3) the sensor nodes are responsible for monitoring the water quality information of the area where the sensor nodes are located, wherein cluster members in each cluster periodically send the monitored water quality information to the cluster heads, the cluster heads are further gathered and sent to a base station for data analysis and processing, and malicious nodes intentionally send wrong data to the cluster heads or the base station or intentionally lose the monitored water quality information probabilistically, so that the base station cannot obtain a correct data analysis result;
(2) in order to identify and detect malicious nodes, the base station collects information (including communication results of the nodes, transmission data, residual energy, link characteristics and the like) representing communication behaviors and communication capacity between the sensing nodes in each period T, and respectively calculates trust factors with different dimensionalities, including communication trust, data trust, node trust and environment trust, for dynamically evaluating the credibility of the sensing nodes;
(2.1) the communication trust can represent the positivity of the sensing node in the water quality monitoring process and is used as an index for judging whether the node intentionally discards transmission data or reduces communication interaction between the nodes, the calculation of the communication trust is based on the communication result between the sensing nodes, the successful communication times and the failure communication times of the node i and the neighbor nodes are a and b respectively within a period T, wherein the neighbor nodes of the node i are all the sensing nodes which can directly communicate with the node i through a single-hop route, the higher the successful communication times between the node i and the neighbor nodes are, the stronger the communication reliability of the node i is represented, so the communication trust of the node i is defined as the ratio of the successful communication times and the total communication times of the neighbor nodes, and is represented as
Figure BDA0002788127510000041
(2.2) data Trust for Water quality monitoring to represent a nodeThe reliability of the measured data is that for the water quality pollution condition of a specific position, the monitored data of a plurality of nodes nearby should have certain correlation, and the water quality data monitored by the nodes are assumed to conform to normal distribution, namely, the probability density function of the water quality data is calculated as
Figure BDA0002788127510000042
Wherein x represents a random variable of the monitoring data, u and sigma represent an average value and a variance of the monitoring data, a node with the monitoring data as the average value u is set to have the highest credibility, correspondingly, the closer the monitoring data value is to the node trust value of u, and based on the node trust value, the monitoring data of the node i is represented as diThen the data trust calculation formula of the node i is defined as
Figure BDA0002788127510000051
(2.3) the node trust is used for representing the current monitoring task execution capacity and the importance degree of the node in a network range, and the node trust considers two trust factors, the node importance and the node energy;
in a changed network topological structure, due to the fact that node positions are different, network nodes bear different importance in monitoring tasks, generally speaking, a small number of nodes in a network bear most of data transmission and communication, therefore, calculating the importance of an underwater sensing node has a great effect on analysis and decision of water quality monitoring data, a communication model between the nodes in the network is abstracted into a connected graph, wherein the sensing nodes in the network correspond to vertexes in the connected graph, a direct communication link between two sensing nodes corresponds to edges of two vertexes in the connected graph, and correspondingly, the importance of the nodes in the network can be expressed as importance indexes related to the vertexes in the connected graph and comprise degrees, betweenness and neighbor degrees;
the degree of a node refers to the number of edges associated with the node; the node betweenness is the proportion of the number of paths passing through the node in all shortest paths in the network to the total number of the shortest paths; the neighbor degree of the node is defined as the average value of the degrees of all the neighbors of the node, and the use of the neighbor degrees can distinguish the importance of the nodes with the same degree under different network topological structures;
based on the degree, betweenness and neighbor degree of the node, the importance of the node is calculated as:
Figure BDA0002788127510000052
wherein b isiDenotes the betweenness of node i, diAnd dniRespectively representing the degrees of the node i and the degrees of the neighbors, and D represents the maximum value of the degrees in all the nodes in the network;
the higher the residual energy of the node is, the longer the life cycle of the node capable of executing the monitoring task is, the stronger the monitoring capability is, and the initial energy of the sensing node is set as EiniThe current residual energy of the node i is EresDefining the energy trust of the node i as
Figure BDA0002788127510000053
Wherein the sensing node knows its own residual energy;
node trust is defined as the product of energy trust and node importance, expressed as
Figure BDA0002788127510000054
(2.4) the environment trust is used for representing factors influencing normal communication of the nodes, and mainly considering different underwater environment noises; ambient noise is represented by the noise Power Spectral Density (PSD) in the acoustic channel, whose calculation depends on the frequency f, and is classified into four categories: turbulent noise NtTransport noise NsWind noise NwAnd thermal noise Nth(ii) a The environment trust is the total noise PSD under water, and the environment trust of the node i can be linearly modeled as:
Figure BDA0002788127510000061
(3) the method comprises the following steps of constructing four-dimensional trust (communication trust, data trust, node trust and environment trust) into a trust data set, and realizing trust evaluation based on an LSTM network model because a long-time memory network (LSTM) is suitable for processing problems highly related to time series, calculating a comprehensive trust value of a sensing node and judging whether the node is abnormal or not;
(3.1) the LSTM network adopts a gate structure to control the discarding or memorizing of information, and through the introduction of input gate control information, the filtering of the control state of an output gate and the forgetting degree of forgetting gate control information, the LSTM neural network can improve the memorizing capability of longer time sequence input;
(3.2) adopting four-dimensional trust data as input in a trust model based on the LSTM, collecting trust data of the next period for malicious node detection after training according to the trust data obtained when the network runs the current period, wherein in the LSTM algorithm, a loss function adopts binary cross entropy, the training frequency is 100, an output activation function is sigmoid, the final trust value of the node can be ensured to be in an interval (0, 1), wherein 0 represents that the node is not trusted, 1 represents that the node is completely trusted, the higher the trust value is, the more trusted the node is, a trust threshold value is determined according to the training process of the LSTM network, the node with the trust value higher than the threshold value is normal, and the node with the trust value lower than the threshold value is malicious.
And (3.3) identifying malicious nodes and filtering by the base station based on the evaluation result of the trust model, analyzing water quality monitoring data sent by normal nodes, and making a decision whether the water quality is polluted.
The invention has the beneficial effects that: the invention provides a dynamic trust model based on a long-time memory network in an underwater acoustic sensor network, which can be used as the input of an LSTM network by collecting and calculating four-dimensional trust metrics (communication trust, data trust, node trust and environment trust) representing the reliability of nodes, further realizes trust modeling and trust evaluation based on the LSTM, can detect malicious sensor nodes in UASNs according to the trust evaluation value, effectively improves the safety of network and node monitoring data, and can be deployed in water quality pollution monitoring application to obtain more reliable water quality monitoring results.
Drawings
FIG. 1 is a flow chart of one embodiment of the present invention;
FIG. 2 is a diagram of a water flow movement model according to an embodiment of the present invention;
FIG. 3 is a water quality monitoring data distribution graph according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of the UASNs communication model of an embodiment of the present invention;
FIG. 5 is a corresponding connectivity graph of the UASNs communication model of one embodiment of the present invention;
FIG. 6 is a graph illustrating neighbor degrees according to an embodiment of the present invention.
In the figure:
Figure BDA0002788127510000071
base station
Figure BDA0002788127510000072
Cluster head
Figure BDA0002788127510000073
Common sensing node
Figure BDA0002788127510000074
Acoustic communication link
Figure BDA0002788127510000075
And (4) a vertex.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The following detailed description of the principles of the invention is provided in connection with the accompanying drawings.
As shown in fig. 1, which is a flow chart of a dynamic trust model based on a long-time memory network in an underwater acoustic sensor network, the invention firstly clusters sensor nodes deployed in the network, selects cluster heads, and periodically sends water quality monitoring data to the cluster heads by the sensor nodes; secondly, collecting water quality monitoring data from the cluster head by the base station for making a decision on whether the water quality is polluted or not, and simultaneously collecting information related to the communication behavior and the communication capacity of the node for detecting a malicious node; the detection of the malicious node is based on a trust data set formed by four-dimensional trust data (including communication trust, data trust, node trust and environment trust), and the trust data set is used as the input of an LSTM network to obtain the comprehensive trust value of the sensing node; whether the sensing node is abnormal or not can be identified by comparing the comprehensive trust value with the trust threshold value, and the base station makes a correct decision by analyzing and processing the water quality monitoring data sent by the normal node. The dynamic trust model process based on the long-time memory network in the underwater acoustic sensor network specifically comprises the following steps:
step (1): aiming at the considered water pollution monitoring application, different sensing devices are deployed in a monitoring range, node devices are clustered based on a KMeans clustering method, an Underwater Acoustic Sensing Network (UASNs) is constructed, and a network node communication and malicious attack mode is defined, and the method specifically comprises the following steps:
(1.1) considering an underwater pollution monitoring application scene, deploying different types of sensing equipment in water, wherein the underwater sensing equipment comprises 1 base station and n isomorphic sensing nodes, namely all the sensing nodes have the same energy and communication range, calculating and storing capacity, a deployment area is of a cubic structure, the base station is arranged in the center of the water surface, and the sensing nodes are randomly deployed in the water;
(1.2) dividing n underwater sensing nodes into k clusters based on a KMeans clustering method, wherein each cluster comprises 1 cluster head and a plurality of cluster members, each cluster head is a sensing node with the highest energy in the current cluster, the sensing nodes in a communication range can directly communicate based on an underwater acoustic link, and the sensing nodes beyond the node communication range are communicated in a multi-hop routing mode; due to the change of the node position caused by the mobility of the water flow, the node position is updated by using an mcm (sequencing Current mobility) model shown in fig. 2, wherein the node movement mainly comprises downstream flow and circular motion (namely, the node is in a vortex);
(1.3) the sensor nodes are responsible for monitoring the water quality information of the area where the sensor nodes are located, wherein cluster members in each cluster periodically send the monitored water quality information to the cluster heads, the cluster heads are further gathered and sent to the base station for data analysis and processing, and malicious nodes intentionally send wrong data to the cluster heads or the base station or intentionally lose the monitored water quality information probabilistically, so that the base station cannot obtain a correct data analysis result.
Step (2): in order to identify and detect malicious nodes, a base station collects information (including communication results of the nodes, transmission data, residual energy, link characteristics and the like) representing communication behaviors and communication capacities between sensing nodes in each period T, and respectively calculates trust factors with different dimensions, including communication trust, data trust, node trust and environment trust, for dynamically evaluating the credibility of the sensing nodes, and the method specifically comprises the following steps:
(2.1) computing communication trust:
the communication trust can represent the positivity of a sensing node in the water quality monitoring process and is used as an index for judging whether the node intentionally discards transmission data or reduces communication interaction between the nodes, the calculation of the communication trust is based on the communication result between the sensing nodes, the successful communication and the failure communication times of a node i and a neighbor node thereof in a period T are respectively a and b, wherein the neighbor node of the node i is all the sensing nodes which can directly communicate with the node i through a single-hop route, the higher the successful communication times between the node i and the neighbor node is, the stronger the communication reliability of the node i is represented, so the communication trust of the node i is defined as the ratio of the successful communication times to the total communication times of the neighbor node, and is represented as
Figure BDA0002788127510000091
(2.2) computing data trust:
the data trust is used for representing the reliability of the water quality monitoring data of a certain node, and for the water quality pollution condition of a certain specific position, the monitoring data of a plurality of nodes nearby should have certain correlation, for example, as shown in fig. 3, it is assumed that the water quality data monitored by the nodes conform to normal distribution, that is, the probability density function of the water quality data is calculated as
Figure BDA0002788127510000092
Wherein x represents a random variable of the monitored data, u and sigma represent a mean and a variance of the monitored data, and the monitored data are processedThe node with the data as the average value u is set to be the highest credibility, correspondingly, the closer the monitoring data value is to the node trust value of u, and based on the node trust value, the monitoring data of the node i is represented as diThen the data trust calculation formula of node ii is defined as
Figure BDA0002788127510000093
(2.3) compute node trust:
the node trust is used for expressing the current monitoring task execution capacity and the importance degree of the node in a network range, and the node trust considers two trust factors, namely the node importance and the node energy;
in a changed network topology structure, due to different node positions, the network nodes have different importance in monitoring tasks, generally, a small number of nodes in a network undertake most of data transmission and communication, and therefore, calculating the importance of an underwater sensing node has a great effect on analysis and decision of water quality monitoring data, for example, as shown in fig. 4 and 5, a communication model (fig. 4) between nodes in the network is abstracted into a connected graph (fig. 5), wherein the sensing node in the network corresponds to a vertex in the connected graph, a direct communication link between two sensing nodes corresponds to edges of two vertices in the connected graph, and accordingly, the importance of a node in the network can be expressed as an importance index related to the vertex in the connected graph, including degrees, betweenness and neighbor degrees;
the degree of a node refers to the number of edges associated with the node; the node betweenness is the proportion of the number of paths passing through the node in all shortest paths in the network to the total number of the shortest paths; the neighbor degrees of a node are defined as the average value of the degrees of all the neighbors of the node, the use of the neighbor degrees can distinguish the importance of the node with the same degree under different network topological structures, as shown in fig. 6, the degrees of a vertex p and a vertex q are both 3, but due to the influence of the neighbor nodes in the network topological structures, q bears more data transmission tasks, and the importance of q is higher than that of p;
based on the degree, betweenness and neighbor degree of the node, the importance of the node is calculated as:
Figure BDA0002788127510000101
wherein b isiDenotes the betweenness of node i, diAnd dniRespectively representing the degrees of the node i and the degrees of the neighbors, and D represents the maximum value of the degrees in all the nodes in the network;
the higher the residual energy of the node is, the longer the life cycle of the node capable of executing the monitoring task is, the stronger the monitoring capability is, and the initial energy of the sensing node is set as EiniThe current residual energy of the node i is EresDefining the energy trust of the node i as
Figure BDA0002788127510000102
Wherein the sensing node knows its own residual energy;
node trust is defined as the product of energy trust and node importance, expressed as
Figure BDA0002788127510000103
(2.4) computing environment trust:
the environment trust is used for expressing factors influencing normal communication of the nodes, and mainly considers different underwater environment noises; ambient noise is represented by the noise Power Spectral Density (PSD) in the acoustic channel, whose calculation depends on the frequency f, and is classified into four categories: turbulent noise NtTransport noise NsWind noise NwAnd thermal noise Nth. The environment trust is the total noise PSD under water, and the environment trust of the node i can be linearly modeled as:
Figure BDA0002788127510000111
and (3): the method comprises the following steps of constructing four-dimensional trust (communication trust, data trust, node trust and environment trust) into a trust data set, and realizing trust evaluation based on an LSTM network model because a long-time memory network (LSTM) is suitable for processing problems highly related to time series, calculating a comprehensive trust value of a sensing node, judging whether the node is abnormal or not, and making a decision whether water quality is polluted or not by a base station based on a trust result, wherein the steps specifically comprise:
(3.1) the LSTM network adopts a gate structure to control the discarding or memorizing of information, and the LSTM neural network can improve the memorizing capability of longer time sequence input by introducing input gate control information, filtering the control state of an output gate and forgetting the forgetting degree of forgetting gate control information;
(3.2) adopting four-dimensional trust data as input in a trust model based on the LSTM, collecting trust data of the next period for malicious node detection after training according to the trust data obtained when the network runs the current period, wherein in the LSTM algorithm, a loss function adopts binary cross entropy, the training frequency is 100, an output activation function is sigmoid, the final trust value of the node can be ensured to be in an interval (0, 1), wherein 0 represents that the node is not trusted, 1 represents that the node is completely trusted, the higher the trust value is, the more trusted the node is, a trust threshold value is determined according to the training result of the LSTM network, the node with the trust value higher than the threshold value is normal, and the node with the trust value lower than the threshold value is malicious.
And (3.3) identifying malicious nodes and filtering by the base station based on the evaluation result of the trust model, analyzing water quality monitoring data sent by normal nodes, and making a decision whether the water quality is polluted.
In summary, the following steps:
the invention discloses a dynamic trust model based on a long-time memory network in an underwater acoustic sensor network, which is used for identifying malicious nodes in the network and ensuring the accuracy of node monitoring data in water quality pollution monitoring application.
The foregoing shows and describes the general principles and broad features of the present invention and advantages thereof. It will be understood by those skilled in the art that the present invention is not limited to the embodiments described above, which are described in the specification and illustrated only to illustrate the principle of the present invention, but that various changes and modifications may be made therein without departing from the spirit and scope of the present invention, which fall within the scope of the invention as claimed. The scope of the invention is defined by the appended claims and equivalents thereof.

Claims (1)

1. A method for constructing a dynamic trust model based on a long-time memory network in an underwater acoustic sensor network is characterized by comprising the following steps:
(1) aiming at the considered water pollution monitoring application, different sensing devices are deployed in a monitoring range, node devices are clustered based on a KMeans clustering method, an underwater acoustic sensing network is constructed, and a network node communication and malicious attack mode is defined;
(1.1) considering an underwater pollution monitoring application scene, deploying different types of sensing equipment in water, wherein the underwater sensing equipment comprises 1 base station and n sensing nodes which are isomorphic, namely all the sensing nodes have the same energy, communication range and calculation and storage capacity, a deployment area is of a cubic structure, the base station is arranged in the center of the water surface, and the sensing nodes are randomly deployed in the water;
(1.2) dividing n underwater sensing nodes into k clusters based on a KMeans clustering method, wherein each cluster comprises 1 cluster head and a plurality of cluster members, each cluster head is a sensing node with the highest energy in the current cluster, the sensing nodes in a communication range are directly communicated based on an underwater acoustic link, and the sensing nodes beyond the node communication range are communicated in a multi-hop routing mode; due to the fact that the mobility of water flow can cause the change of the node position, the node position is updated by adopting an MCM (multi-chip module) model, wherein the node movement mainly comprises downstream flow and circular motion, namely the node is in a vortex;
(1.3) the sensing nodes are responsible for monitoring the water quality information of the area, wherein cluster members in each cluster periodically send the monitored water quality information to the cluster heads, the cluster heads are further gathered and sent to a base station for data analysis and processing, and malicious nodes intentionally send wrong data to the cluster heads or the base station or intentionally lose the monitored water quality information probabilistically, so that the base station cannot obtain a correct data analysis result;
(2) in order to identify and detect malicious nodes, the base station collects information representing communication behaviors and communication capacity between the sensing nodes in each period T, and respectively calculates trust factors with different dimensions, including communication trust, data trust, node trust and environment trust, for dynamically evaluating the credibility of the sensing nodes;
(2.1) the communication trust can represent the positivity of the sensing node in the water quality monitoring process and is used as an index for judging whether the node intentionally discards transmission data or reduces communication interaction between the nodes, the calculation of the communication trust is based on the communication result between the sensing nodes, the successful communication times and the failure communication times of the node i and the neighbor nodes are a and b respectively within a period T, wherein the neighbor nodes of the node i are all the sensing nodes which can directly communicate with the node i through a single-hop route, the higher the successful communication times between the node i and the neighbor nodes are, the stronger the communication reliability of the node i is represented, so the communication trust of the node i is defined as the ratio of the successful communication times and the total communication times of the neighbor nodes, and is represented as
Figure FDA0003237987980000021
(2.2) the data trust is used for representing the reliability of the water quality monitoring data of a certain node, the monitoring data of a plurality of nodes nearby have certain correlation according to the water quality pollution condition of a certain specific position, and the water quality data monitored by the nodes are supposed to conform to normal distribution, namely the probability density function of the water quality data is calculated as
Figure FDA0003237987980000022
Wherein x represents a random variable of the monitoring data, u and sigma represent an average value and a variance of the monitoring data, a node with the monitoring data as the average value u is set to have the highest credibility, correspondingly, the closer the monitoring data value is to the node trust value of u, and based on the node trust value, the monitoring data of the node i is represented as diThen the data trust calculation formula of the node i is defined as
Figure FDA0003237987980000023
(2.3) the node trust is used for representing the current monitoring task execution capacity and the importance degree of the node in a network range, and the node trust considers two trust factors, the node importance and the node energy;
in a changed network topology structure, because the positions of nodes are different and the importance of the network nodes in the monitoring task is different, a communication model between the nodes in the network is abstracted into a connected graph, wherein the sensing nodes in the network correspond to vertexes in the connected graph, a direct communication link between two sensing nodes corresponds to edges of the two vertexes in the connected graph, and correspondingly, the importance of the nodes in the network is expressed as an importance index related to the vertexes in the connected graph and comprises degrees, betweenness and neighbor degrees;
the degree of a node refers to the number of edges associated with the node; the node betweenness is the proportion of the number of paths passing through the node in all shortest paths in the network to the total number of the shortest paths; the neighbor degree of the node is defined as the average value of the degrees of all the neighbors of the node, and the use of the neighbor degrees can distinguish the importance of the nodes with the same degree under different network topological structures;
based on the degree, betweenness and neighbor degree of the node, the importance of the node is calculated as:
Figure FDA0003237987980000031
wherein b isiDenotes the betweenness of node i, diAnd dniRespectively representing the degrees of the node i and the degrees of the neighbors, and D represents the maximum value of the degrees in all the nodes in the network;
the higher the residual energy of the node is, the longer the life cycle of the node capable of executing the monitoring task is, the stronger the monitoring capability is, and the initial energy of the sensing node is set as EiniThe current residual energy of the node i is EresDefining the energy trust of the node i as
Figure FDA0003237987980000032
Wherein the sensing node knows its own residual energy;
node trust is defined as the product of energy trust and node importance, expressed as
Figure FDA0003237987980000033
(2.4) the environment trust is used for representing factors influencing normal communication of the nodes, and mainly considering different underwater environment noises; ambient noise is represented by the noise power spectral density PSD in the acoustic channel, whose calculation depends on the frequency f, and is classified into four categories: turbulent noise NtTransport noise NsWind noise NwAnd thermal noise Nth(ii) a The environment trust is the total noise PSD under water, and the environment trust of the node i can be linearly modeled as follows:
Figure FDA0003237987980000034
(3) the method comprises the following steps of constructing a trust data set by using four-dimensional trust, namely communication trust, data trust, node trust and environment trust, and realizing trust evaluation based on an LSTM network model because a long-time memory network LSTM is suitable for processing problems highly related to time series, calculating a comprehensive trust value of a sensing node and judging whether the node is abnormal or not;
(3.1) the LSTM network adopts a gate structure to control the discarding or memorizing of information, and through the introduction of input gate control information, the filtering of the control state of an output gate and the forgetting degree of forgetting gate control information, the LSTM neural network can improve the memorizing capability of long-time sequence input;
(3.2) adopting four-dimensional trust data as input in a trust model based on the LSTM, collecting trust data of the next period for malicious node detection after training according to the trust data obtained when the network runs the current period, wherein in the LSTM algorithm, a loss function adopts binary cross entropy, the training frequency is 100, an output activation function is sigmoid, and the final trust value of the node is ensured to be in an interval (0, 1), wherein 0 represents that the node is not trusted, 1 represents that the node is completely trusted, the higher the trust value is, the more trusted the node is, the trust threshold is determined according to the training process of the LSTM network, the node with the trust value higher than the threshold is normal, and the node with the trust value lower than the threshold is malicious;
and (3.3) identifying malicious nodes and filtering by the base station based on the evaluation result of the trust model, analyzing water quality monitoring data sent by normal nodes, and making a decision whether the water quality is polluted.
CN202011305178.9A 2020-11-19 2020-11-19 Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network Active CN112512003B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011305178.9A CN112512003B (en) 2020-11-19 2020-11-19 Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011305178.9A CN112512003B (en) 2020-11-19 2020-11-19 Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network

Publications (2)

Publication Number Publication Date
CN112512003A CN112512003A (en) 2021-03-16
CN112512003B true CN112512003B (en) 2021-11-05

Family

ID=74958860

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011305178.9A Active CN112512003B (en) 2020-11-19 2020-11-19 Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network

Country Status (1)

Country Link
CN (1) CN112512003B (en)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113179136B (en) * 2021-04-23 2023-11-28 河海大学 Fluid boundary tracking method based on virtual node migration in underwater acoustic sensor network
CN113891275B (en) * 2021-10-12 2023-11-07 大连理工大学 Underwater wireless sensor network trust model based on transfer learning
CN114845306B (en) * 2022-04-21 2024-04-19 中南大学 Network trust state acquisition method based on active message detection

Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103095589A (en) * 2013-01-17 2013-05-08 电子科技大学 Sensor node trust value management method of wireless sensor network of tree topology structure based on Internet protocol (IP) v 6
CN103619009A (en) * 2013-10-25 2014-03-05 河海大学常州校区 A method for establishing a trust model in an underwater sensor network
CN103763705A (en) * 2014-01-22 2014-04-30 天津大学 Malicious anchor node detecting method for underwater sensor network
CN104469836A (en) * 2014-11-24 2015-03-25 河海大学常州校区 Method for building multi-dimension trust model in underwater sensor network
CN110536258A (en) * 2019-08-09 2019-12-03 大连理工大学 Trust model based on isolated forest in a kind of UASNs
CN110868714A (en) * 2019-11-15 2020-03-06 大连理工大学 Clustering-based weightless node trust evaluation method in UASNs
WO2020068141A1 (en) * 2018-09-26 2020-04-02 Google Llc Predicted variables in programming
CN111260029A (en) * 2020-01-13 2020-06-09 北京工业大学 Credibility analysis method for air quality data
CN111429718A (en) * 2020-03-20 2020-07-17 淮阴工学院 Intelligent detection system for road traffic safety

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9646330B2 (en) * 2012-12-05 2017-05-09 Loren Owens Non-metered wireless system and method to render usage based invoices for storage tank consumables
US11960981B2 (en) * 2018-03-09 2024-04-16 Zestfinance, Inc. Systems and methods for providing machine learning model evaluation by using decomposition

Patent Citations (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103095589A (en) * 2013-01-17 2013-05-08 电子科技大学 Sensor node trust value management method of wireless sensor network of tree topology structure based on Internet protocol (IP) v 6
CN103619009A (en) * 2013-10-25 2014-03-05 河海大学常州校区 A method for establishing a trust model in an underwater sensor network
CN103763705A (en) * 2014-01-22 2014-04-30 天津大学 Malicious anchor node detecting method for underwater sensor network
CN104469836A (en) * 2014-11-24 2015-03-25 河海大学常州校区 Method for building multi-dimension trust model in underwater sensor network
WO2020068141A1 (en) * 2018-09-26 2020-04-02 Google Llc Predicted variables in programming
CN110536258A (en) * 2019-08-09 2019-12-03 大连理工大学 Trust model based on isolated forest in a kind of UASNs
CN110868714A (en) * 2019-11-15 2020-03-06 大连理工大学 Clustering-based weightless node trust evaluation method in UASNs
CN111260029A (en) * 2020-01-13 2020-06-09 北京工业大学 Credibility analysis method for air quality data
CN111429718A (en) * 2020-03-20 2020-07-17 淮阴工学院 Intelligent detection system for road traffic safety

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
ITrust: An Anomaly-resilient Trust Model Based on Isolation Forest for Underwater Acoustic Sensor Networks;Jiaxin Du 等;《 IEEE Transactions on Mobile Computing ( Early Access )》;20201002;全文 *
Learning to Track by Bi-Directional Long Short-Term Memory Networks;Chen Pan 等;《 2019 IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computing, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI)》;20200409;全文 *
基于集对分析的无线传感器网络风险信任模型;赵金辉 等;《无线传感器技术》;20150807;全文 *

Also Published As

Publication number Publication date
CN112512003A (en) 2021-03-16

Similar Documents

Publication Publication Date Title
CN112512003B (en) Dynamic trust model based on long-time and short-time memory network in underwater acoustic sensor network
Otoum et al. Empowering reinforcement learning on big sensed data for intrusion detection
CN110536258B (en) Trust model based on isolated forest in UASNs
Alsheikh et al. Machine learning in wireless sensor networks: Algorithms, strategies, and applications
Dong et al. An Intrusion Detection Model for Wireless Sensor Network Based on Information Gain Ratio and Bagging Algorithm.
Coppolino et al. Applying data mining techniques to intrusion detection in wireless sensor networks
Yan et al. A hybrid intrusion detection system of cluster-based wireless sensor networks
Rida et al. EK-means: A new clustering approach for datasets classification in sensor networks
CN103533571B (en) Fault-tolerant event detecting method based on temporal voting strategy
Vijayakumar et al. An adaptive neuro-fuzzy logic based jamming detection system in WSN
Lu et al. Intrusion detection of wireless sensor networks based on IPSO algorithm and BP neural network
Sahoo et al. Guard against trust management vulnerabilities in wireless sensor network
Ayadi et al. Spatio-temporal correlations for damages identification and localization in water pipeline systems based on WSNs
Qu et al. A lightweight intrusion detection method based on fuzzy clustering algorithm for wireless sensor networks
Dagdeviren Weighted connected vertex cover based energy-efficient link monitoring for wireless sensor networks towards secure internet of things
Wang Packet traffic: a good data source for wireless sensor network modeling and anomaly detection
Del Rey et al. Malware propagation in Wireless Sensor Networks: global models vs Individual-based models
Karthik et al. Sensor data modeling for data trustworthiness
Şen et al. Power-aware intrusion detection in mobile ad hoc networks
Liang et al. Reliability indexes for multi-AUV cooperative systems
CN116916317A (en) Invasion detection method based on white shark and random forest
Islam et al. Data cleansing during data collection from wireless sensor networks
Hikal et al. Detection of black-hole attacks in MANET using adaboost support vector machine
Syed et al. Software-Fault Mitigation for Derivation of Quality of Services (QoS) in Wireless Sensor Networks (WSN).
JayaPratha et al. Transient failures detection in data transfer for wireless sensors based communication using stochastic gradient descent with momentum

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant