CN105634798B - A kind of sensor network event detecting method based on two-tier system - Google Patents
A kind of sensor network event detecting method based on two-tier system Download PDFInfo
- Publication number
- CN105634798B CN105634798B CN201510991979.8A CN201510991979A CN105634798B CN 105634798 B CN105634798 B CN 105634798B CN 201510991979 A CN201510991979 A CN 201510991979A CN 105634798 B CN105634798 B CN 105634798B
- Authority
- CN
- China
- Prior art keywords
- node
- cluster
- cluster head
- event
- network
- 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.)
- Expired - Fee Related
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/06—Management of faults, events, alarms or notifications
- H04L41/0604—Management of faults, events, alarms or notifications using filtering, e.g. reduction of information by using priority, element types, position or time
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W24/00—Supervisory, monitoring or testing arrangements
- H04W24/04—Arrangements for maintaining operational condition
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
Abstract
The invention discloses a kind of sensor network event detecting method based on two-tier system, its detection method is as follows: assuming that network is made of the node of many low price, all nodes are all static after deployment, each node knows the location information of oneself, whole network is logically divided into grid, and one cluster of a grid protocol is that each cluster selects a cluster head according to the dump energy of node, each node has a unique ID, there is a series of neighbor node；If 2 nodal distances are less than or equal to communication distance, they are exactly neighbours；If there is monitored link, node is in the grid closed on it could also be possible that neighbours；Each node directly or indirectly can report that then cluster head can use tree-shaped route to base station report to cluster head.The present invention is able to maintain lower rate of false alarm while embodying compared with high detection rate, accurately and in terms of rate of false alarm reaches well balanced in incident detection, can also effectively extend network lifecycle.
Description
Technical field
The present invention relates to network event detection technique field more particularly to a kind of sensor network things based on two-tier system
Part detection method.
Background technique
Wireless sensor network is that the network being made of in a distributed fashion a large amount of sensor nodes (hereinafter referred to as senses
Device network).Sensor network combines the technologies such as sensor, computer, communication, it may be convenient to carry out data sampling and processing
And transmission.Moment sensor network is answered in fields such as intelligent transportation, reading intelligent agriculture, environmental monitoring and military affairs extensively
With.
Compared to traditional network, sensor network has the following characteristics that 1) large scale deployment.Sensor network is often extensive
It is deployed in the complex environment for being difficult to reach, such as forest, the bottom.Complex environment and cheap price cause node to be easy to lose
Effect generates false readings, therefore is considered as Fault-Tolerant Problems in algorithm for design.2) self-organizing.By from group after node deployment
It knits algorithm to be communicated with surroundings nodes, self-organizing forms a network.During the network operation, node is it is possible that event
Barrier, this requires networks to have adaptivity, dynamically can configure and manage.
As described above, sensor network is that comparison is fragile, due to attacking, interfering, node can generate false readings, from
And the generation of false positive event.How while carrying out incident detection wrong report is told, is the critical issue for needing to solve.Due to
Node battery is difficult to replace, therefore saves key factor in need of consideration when energy is algorithm for design.
False readings are randomly generated, and without associated, and event has a spatial correlation, in certain area
Node can all detect event.Currently there is the incident detection algorithm of some tolerance mistakes, has had based on neighbours, also there is base
In what is clustered.In the case where sensor network nodes deployment is more relatively high than the degree of comparatively dense sensor node, based on neighbours'
Algorithm can achieve preferable performance.Algorithm based on cluster can preferably use resource, can be by better in the way of aggregation
It reduces network overhead, but if the region that occurs of an event is smaller, and across multiple clusters, is difficult to that suitable threshold is arranged
Value.So needing to reach a balance between detection accurate rate and rate of false alarm.
A kind of sensor network event detecting method TLEDA based on two-tier system is proposed based on this present invention, using facing
The spatial coherence of nearly node, give close on monitor the more weights of the node of event, effectively to distinguish event and wrong report；
Judge whether event occurs using cluster head, avoids all passing data back base station, communication-cost can be further reduced.
Summary of the invention
Technical problems based on background technology, the sensor network thing based on two-tier system that the invention proposes a kind of
Part detection method.
A kind of sensor network event detecting method based on two-tier system proposed by the present invention, detection method are as follows:
Assuming that network is made of the node of many low price, all nodes are all static after deployment, and each node is known
The location information in road oneself, whole network are logically divided into grid, one cluster of a grid protocol, according to the surplus of node
Complementary energy is that each cluster selects a cluster head, and each node has a unique ID, there is a series of neighbor node；If 2
Nodal distance is less than or equal to communication distance, they are exactly neighbours；If there is monitored link, node also has in the grid closed on can
It can be neighbours；Each node directly or indirectly can report that then cluster head can use tree-shaped route to base station to cluster head
Report；
As node viIt detects an exception, sends 1, v to neighbor nodeiStatistics receives 1 quantity and includes oneself
, the purpose of statistics is to s with higheriThe more weights of node, in addition the degree di of node is also very important weighing apparatus
Figureofmerit, the contribution of 1 node we use w i2Si-1It measures, wherein wiWeight is represented,
T in formula 21Represent upper threshold, T2Represent bottom threshold；If si/ (di+1) is greater than the upper limit, wiIt is set as 1；
If si/ (di+1) is between lower and upper limit, wiIt is set as si/(di+1)；If si/ (di+1) is lower than lower limit, then wiIf
It is set to 0；The meaning of above-mentioned setting is siMore big then its contribution is bigger；siThe ratio for accounting for its neighbour sum is higher, and that contributes can
Higher by spending, weight is bigger；
Finally determine whether event occurs by cluster head；
Wherein
Indicate the summation of the contribution of occurred abnormal nodes in cluster, ∑ (1-xi) indicate it is all go out
Now abnormal interstitial content.
Preferably, it is assumed that pwIndicate the probability that mistake occurs, then sampling normal node in primary sampling should reach
1-pw, should averagely have s for a nodei/ (di+1) > 1-pw, so we are by T1It is set as 1-pw；Work as si/(di+
1) > 1-pw, which is a normal node, and weight should be 1；Work as pw< si/(di+1)≤1-pw, part approval should
The judgement of node, by T2It is set as pw, work as si/(di+1)≤pw, which probably has occurred mistake, so by its weight
It is set as 0.
Event detection is the important application of sensor network, and node, which is interfered, can issue wrong report, in the same of detecting event
When to differentiate wrong report be to need the critical issue that solves.General description of the present invention is as follows: 1) nodal test to exception is sent to neighbours
1；2) it each detects abnormal node, counts its quantity for receiving 1, be then sent to cluster head；3) cluster head is determined according to formula 2
The weight of each report node；4) whether cluster head occurs according to 3 decision event of formula；5) the degree of belief mould of cluster head more new node
Type.
A kind of sensor network event detecting method TLEDA based on two-tier system proposed by the present invention is utilized in cluster head
The information of neighbor node is based on threshold testing, judges whether event occurs；Distinguish event and wrong report when, give close on it is same
When the more weights of node that sound an alarm；The detection method be it is adaptive, being embodied in error node can be by from network interval
From；When experiment is shown in lesser event area, which still has preferable efficiency；It compares and is thrown based on most of
The local decision-making algorithm Vote of ticket, the detection method are able to maintain lower rate of false alarm while embodying compared with high detection rate,
Incident detection is accurate and rate of false alarm aspect reaches well balanced, can also effectively extend network lifecycle.
Detailed description of the invention
Fig. 1 is the hierarchy network graph of inventive sensor network；
Fig. 2 is the verification and measurement ratio figure of Vote algorithm of the present invention；
Fig. 3 is the rate of false alarm figure of Vote algorithm of the present invention；
Fig. 4 is the rate of false alarm figure of TLEDA algorithm of the present invention；
Fig. 5 is the rate of false alarm figure of TLEDA algorithm of the present invention.
Specific embodiment
Combined with specific embodiments below the present invention is made further to explain.
A kind of sensor network event detecting method based on two-tier system proposed by the present invention, detection method are as follows:
Assuming that network is made of the node of many low price, all nodes are all static after deployment, and each node is known
The location information in road oneself, whole network are logically divided into grid, one cluster of a grid protocol, according to the surplus of node
Complementary energy is that each cluster selects a cluster head, and each node has a unique ID, there is a series of neighbor node；If 2
Nodal distance is less than or equal to communication distance, they are exactly neighbours；If there is monitored link, node also has in the grid closed on can
It can be neighbours；Each node directly or indirectly can report that then cluster head can use tree-shaped route to base station to cluster head
Report；
As shown in Figure 1, network is divided into the grid of 4*4, each cluster has a cluster head, and cluster head is dynamic select,
Other nodes pass through 1 jump or multi-hop and cluster head connection in cluster.
Various mistakes may occur in sensor network.We be primarily upon due to noise or node failure generation
Mistake sampled data.Assuming that any node all may generate wrong data with same probability.Each node knows normal sample
The range of data, " normal " refer to that there is no sampled datas when event.Data except normal sample we all
Referred to as abnormal data is based on this, and node is it may determine that the sampled data of oneself is normal or abnormal.
Abnormal data may be an event, it is also possible to as caused by certain mistake.If an exception is by thing
Caused by part, it is necessary to which the event is reported to base station；It is not have if an exception is reported as caused by certain mistake
Meaning, it needs to filter this out in net, to avoid consumption excessive power.
One event be a border circular areas (size of event area will be considered, with facilitate for event setup it is suitable
Threshold value).For the event of a large area, relatively easy given threshold, because at least one cluster can have enough numbers
Node observe the event, can effectively distinguish event and wrong report.But after event area reduces, each cluster may only have
A small amount of event node causes certain difficulty thus for event detection.
In order to obtain higher incident detection accurate rate, and lower rate of false alarm is kept, it is proposed that being based on two-tier system
Sensor network event detecting method (Two-layer based Event Detection Algorithm, TLEDA),
TLEDA uses a kind of mixed mode, the mode of mode and level of the integrated utilization based on neighbours.
When node finds a uncommon sampling, which sends 1 to his neighbours, alert node each in this way
Its 1 quantity s received can be countedi.Each node vi, the s that is countediIt reports to cluster head, such cluster head can be right
The event is assessed.
Degree of belief represents degree of reliability when node report event.For the cluster head that one has n member node,
He needs to keep its degree of belief for each node.Initial value is both configured to 1.When having event report every time, cluster head is according to node report
The accuracy of announcement updates its degree of belief.Failure node will be just confirmed as when the degree of belief of node reaches lower limit, and will
It is isolated from network behind.When not having event, node report 1, degree of belief will be reduced α；Report 0 is trusted
Degree will will be increased β.
When event occurs, if the node report 0 in event area, their degree of belief will be lowered, due to very
Hardly possible determines the exact boundry of event, so we do not go the degree of belief of more new node when event occurs.
With the increase of error node, mistake or incident detection in sensor network become complicated.Especially when one
In a cluster error node and event node than when, be difficult that a suitable threshold value is selected to distinguish event and wrong report.This causes
In the case where not increasing rate of false alarm, it is difficult to keep higher incident detection accurate rate.In order in incident detection accuracy and mistake
Reach balance between report, it is proposed that a kind of mixed model detecting event, gives and sound an alarm and node is more adjacent to one another
Weight.
As node viIt detects an exception, sends 1 to neighbor node.Then, viThe quantity that statistics receives 1 (includes certainly
Oneself).The purpose of statistics is to s with higheriThe more weights of node, in addition the degree di of node is also very important
Measurement index.The contribution of 1 node we use w i2Si-1It measures, wherein wiRepresent weight.
T in formula 21Represent upper threshold, T2Represent bottom threshold.If si/ (di+1) is greater than the upper limit, wiIt is set as 1；
If si/ (di+1) is between lower and upper limit, wiIt is set as si/(di+1)；If si/ (di+1) is lower than lower limit, then wiIf
It is set to 0.The meaning of above-mentioned setting is siMore big then its contribution is bigger.siThe ratio for accounting for its neighbour sum is higher, and that contributes can
Higher by spending, weight is bigger.Finally determine whether event occurs by cluster head.
Wherein
Indicate the summation of the contribution of occurred abnormal nodes in cluster, ∑ (1-xi) indicate it is all go out
Now abnormal interstitial content.
We discuss the setting of threshold value below, it is assumed that pwIt indicates the probability that mistake occurs, is then sampled just in primary sampling
Normal node should reach 1-pw, should averagely have s for a nodei/ (di+1) > 1-pw, so we are by T1If
It is set to 1-pw.Work as si/ (di+1) > 1-pw, it is believed that the node is a normal node, and weight should be 1.Work as pw<
si/(di+1)≤1-pw, it is believed that the judgement of the node can only be partially accepted, so we are by T2It is set as pw.Work as si/(di+
1)≤pw, it is believed that mistake probably has occurred in the node, so setting 0 for its weight.
The present invention carries out emulation experiment using OMNET++.320 sensor nodes are shared in network, random placement exists
The region of one 4*4, each cluster have 20 nodes.The side length of grid is L, and the radius of event area is 0.5L, and communication distance is set
It sets so that the degree of node is about 5.In an experiment it is contemplated that 2 Indexs measure rates and rate of false alarm.
Verification and measurement ratio: detect the round of event divided by total event round；
Rate of false alarm: the round of false alarm divided by total not event round.
The present invention has evaluated the verification and measurement ratio of Vote algorithm, as shown in figure 3, horizontal axis indicates pw, longitudinal axis expression verification and measurement ratio.It is different
pwLower verification and measurement ratio requires to keep higher, if because event, which is not reported, may bring larger problem.Vote algorithm is only
Can just verification and measurement ratio be made to reach 0.95 when threshold value is less than 0.25.Fig. 4 shows the rate of false alarm of Vote algorithm, it can be seen that for
0.25 threshold value, rate of false alarm is too high, and when error rate only has 0.15, rate of false alarm all can be more than 70%.So being calculated for Vote
For method, it is difficult to meet higher verification and measurement ratio and lower rate of false alarm simultaneously.
The present invention also has evaluated the verification and measurement ratio and rate of false alarm of TLEDA algorithm, and as shown in Figure 4 and Figure 5, horizontal axis also illustrates that pw。
Fig. 4 is illustrated in T1For 0.7, T2In the case of 0.3, T3The verification and measurement ratio of respectively 0.6,0.65 and 0.7, it can be seen that work as T3For
Verification and measurement ratio highest when 0.6, and T3Under different values, the verification and measurement ratio of TLEDA algorithm can be maintained at higher level.Figure
5 illustrate the rate of false alarm of TLEDA algorithm, it can be seen that compare Vote algorithm, and rate of false alarm is reduced there has also been apparent.
The foregoing is only a preferred embodiment of the present invention, but scope of protection of the present invention is not limited thereto,
Anyone skilled in the art in the technical scope disclosed by the present invention, according to the technique and scheme of the present invention and its
Inventive concept is subject to equivalent substitution or change, should be covered by the protection scope of the present invention.
Claims (2)
1. a kind of sensor network event detecting method based on two-tier system, which is characterized in that its detection method is as follows:
Assuming that network is made of the node of many low price, all nodes are all static after deployment, and each node is known certainly
Oneself location information, whole network are logically divided into grid, one cluster of a grid protocol, according to the residual energy of node
Amount is that each cluster selects a cluster head, and each node has a unique ID, there is a series of neighbor node；If 2 nodes
Distance is less than or equal to communication distance, they are exactly neighbours；If there is monitored link, node in the grid closed on it could also be possible that
Neighbours；Each node directly or indirectly can report that then cluster head can use tree-shaped route to base station report to cluster head；
As node viIt detects an exception, sends 1, v to neighbor nodeiStatistics receives 1 quantity and includes oneself, system
The purpose of meter is to s with higheriThe more weights of node, in addition the degree di of node is also very important measurement and refers to
Mark, the contribution of 1 node we use wi2Si-1It measures, wherein wiWeight is represented,
T in formula 21Represent upper threshold, T2Represent bottom threshold；If si/ (di+1) is greater than the upper limit, wiIt is set as 1, siIt is every
A alert node counts its 1 quantity received；If si/ (di+1) is between lower and upper limit, wiIt is set as si/(di+
1)；If si/ (di+1) is lower than lower limit, then wiIt is set as 0；The meaning of above-mentioned setting is siMore big then its contribution is bigger；siIt accounts for
The ratio of its neighbour sum is higher, and the reliability of contribution is higher, and weight is bigger；
Finally determine whether event occurs by cluster head；
Wherein
Indicate the summation of the contribution of occurred abnormal nodes in cluster, ∑ (1-xi) indicate occur it is different
Normal interstitial content, T3For preset threshold.
2. a kind of sensor network event detecting method based on two-tier system according to claim 1, which is characterized in that
Assuming that pwIndicate the probability that mistake occurs, then 1-p should be reached by sampling normal node in primary samplingw, for a node
For should averagely have si/ (di+1) > 1-pw, so we are by T1It is set as 1-pw；Work as si/ (di+1) > 1-pw, which is
One normal node, weight should be 1；Work as pw< si/(di+1)≤1-pw, the judgement of the node is partially accepted, by T2If
It is set to pw, work as si/(di+1)≤pw, which probably has occurred mistake, so setting 0 for its weight.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510991979.8A CN105634798B (en) | 2015-12-25 | 2015-12-25 | A kind of sensor network event detecting method based on two-tier system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510991979.8A CN105634798B (en) | 2015-12-25 | 2015-12-25 | A kind of sensor network event detecting method based on two-tier system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105634798A CN105634798A (en) | 2016-06-01 |
| CN105634798B true CN105634798B (en) | 2019-10-15 |
Family
ID=56049336
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510991979.8A Expired - Fee Related CN105634798B (en) | 2015-12-25 | 2015-12-25 | A kind of sensor network event detecting method based on two-tier system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105634798B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108932842A (en) * | 2018-08-06 | 2018-12-04 | 深圳市创艺工业技术有限公司 | Urban highway traffic pollutant emission intelligent monitor system |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103747537A (en) * | 2014-01-15 | 2014-04-23 | 广东交通职业技术学院 | Wireless sensor network outlier data self-adaption detecting method based on entropy measurement |
| CN103887886A (en) * | 2014-04-14 | 2014-06-25 | 杭州昊美科技有限公司 | Power network detection system and method based on sensor network |
| CN104318688A (en) * | 2014-10-31 | 2015-01-28 | 湘潭大学 | Multi-sensor fire early-warning method based on data fusion |
| CN104994535A (en) * | 2015-06-04 | 2015-10-21 | 浙江农林大学 | Sensor data flow abnormality detection method based on multidimensional data model |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8510234B2 (en) * | 2010-01-05 | 2013-08-13 | American Gnc Corporation | Embedded health monitoring system based upon Optimized Neuro Genetic Fast Estimator (ONGFE) |
-
2015
- 2015-12-25 CN CN201510991979.8A patent/CN105634798B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103747537A (en) * | 2014-01-15 | 2014-04-23 | 广东交通职业技术学院 | Wireless sensor network outlier data self-adaption detecting method based on entropy measurement |
| CN103887886A (en) * | 2014-04-14 | 2014-06-25 | 杭州昊美科技有限公司 | Power network detection system and method based on sensor network |
| CN104318688A (en) * | 2014-10-31 | 2015-01-28 | 湘潭大学 | Multi-sensor fire early-warning method based on data fusion |
| CN104994535A (en) * | 2015-06-04 | 2015-10-21 | 浙江农林大学 | Sensor data flow abnormality detection method based on multidimensional data model |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105634798A (en) | 2016-06-01 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Ding et al. | Localized fault-tolerant event boundary detection in sensor networks | |
| CN107872460B (en) | A kind of wireless sense network DoS attack lightweight detection method based on random forest | |
| Yuan et al. | A distributed Bayesian algorithm for data fault detection in wireless sensor networks | |
| Garofalo et al. | Enhancing intrusion detection in wireless sensor networks through decision trees | |
| CN103338451B (en) | Distributed malicious node detection method in a kind of wireless sensor network | |
| CN104618908B (en) | The method and apparatus that distributed cognition wireless network is attacked anti-distort perception data | |
| Jafarali Jassbi et al. | Fault tolerance and energy efficient clustering algorithm in wireless sensor networks: FTEC | |
| Kar et al. | Reliable and efficient data acquisition in wireless sensor networks in the presence of transfaulty nodes | |
| Mohapatra et al. | Artificial immune system based fault diagnosis in large wireless sensor network topology | |
| Jaint et al. | An efficient weighted trust method for malicious node detection in clustered wireless sensor networks | |
| CN105634798B (en) | A kind of sensor network event detecting method based on two-tier system | |
| Gao et al. | A robust and efficient cooperative spectrum sensing scheme in cognitive radio networks | |
| CN103929738B (en) | WSNs united intrusion detection method based on multiple danger agents | |
| Xu et al. | Distributed fault diagnosis of wireless sensor networks | |
| Saihi et al. | Decentralized fault detection in wireless sensor network based on function error | |
| Mahapatroa et al. | Transient fault tolerant wireless sensor networks | |
| CN109150623B (en) | Method for resisting SSDF attack of malicious user based on round robin reputation value | |
| Weber et al. | Diagnosability evaluation for a system-level diagnosis algorithm for wireless sensor networks | |
| Zhao et al. | A fault detection algorithm based on cluster analysis in wireless sensor networks | |
| Zhang et al. | Secure sensor localization in wireless sensor networks based on neural network | |
| Zhang et al. | A fuzzy-decision based approach for composite event detection in wireless sensor networks | |
| Laoudias et al. | Fault tolerant target localization and tracking in binary WSNs using sensor health state estimation | |
| Taggu et al. | Random-byzantine attack mitigation in cognitive radio networks using a multi-hidden markov model system | |
| Pongaliur et al. | Multi-fusion based distributed spectrum sensing against data falsification attacks and byzantine failures in cr-manet | |
| Surya et al. | Deployment of backup sensors in wireless sensor networks for structural health monitoring |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | 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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20191015 Termination date: 20211225 |