CN103491535B - The general approximate enquiring method of secret protection of facing sensing device network - Google Patents
The general approximate enquiring method of secret protection of facing sensing device network Download PDFInfo
- Publication number
- CN103491535B CN103491535B CN201310389413.9A CN201310389413A CN103491535B CN 103491535 B CN103491535 B CN 103491535B CN 201310389413 A CN201310389413 A CN 201310389413A CN 103491535 B CN103491535 B CN 103491535B
- Authority
- CN
- China
- Prior art keywords
- node
- vector
- data
- base station
- sensor node
- 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
Links
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Mobile Radio Communication Systems (AREA)
Abstract
The present invention proposes a kind of general approximate enquiring method of secret protection of facing sensing device network, includes the following steps:The data structure shared by base station and sensor node, the number of sensor node and gathered data are hidden among vector;Aggregation node is transmitted vector to base station by tree-shaped routing;System of linear equations is constructed in base station, solves the histogram with global statistics information and corresponding sensor node number;The statistical information being had according to histogram completes required approximate query.The present invention reduces approximate query energy expenditure in the case where not leaking privacy information, using the strategy such as data aggregation and filter in net;In the case where model need not be changed, the precision controlling to secret protection approximate query is realized by parameter regulation.
Description
Technical field
The present invention relates to a kind of sensor network query methods more particularly to a kind of in the premise for meeting secret protection requirement
Under, the general approximate enquiring method of facing sensing device network belongs to internet of things field.
Background technology
Sensor network is the important component of Internet of Things, in environmental monitoring, health care, intelligent transportation, national defence troops
The fields such as thing have broad application prospects.With the application development of sensor network, the exposure during practical application deployment
Go out serious threat private data leakage or be tampered, causes the great attention of researcher.Currently, secret protection technology
Have become one of the research hotspot of Internet of Things field.
Secret protection technology can be mainly divided into three classes:Data perturbation technology, data encryption technology and data anonymization skill
Art.Wherein, the secret protection technology in data mining and data publication field provides technology for the secret protection of sensor network
It uses for reference and support, but often due to energy expenditure is big or the factors such as unsuitable distributed environment are not directly applicable sensor
Network.Therefore, existing sensor network secret protection technology is concentrated mainly on the aggregation of secret protection data and secret protection number
It is investigated that asking two aspects.
It in terms of secret protection data aggregation, is had studied in the prior art in the case where not revealing privacy information, is completed
The aggregation operators such as SUM, MIX, MAX or Median, but these technical solutions cannot obtain truthful data value in aggregation node
In the case of realize aggregation operator, due to aggregation information in there is no carry sensors node serial number information, so can not achieve needs
The complex query of sensor node ID, such as inquiry or range query.In terms of secret protection data query, divide in the prior art
Secret protection technology in the inquiries such as range query (Range Query), inquiry is not discussed, but can not reveal privacy
In the case of information, it is completed at the same time the operation of the complex queries such as inquiry, range query.In addition, since approximate query is sensor network
Important inquiry mode in network can be balanced in the case of not high to required precision between precision and energy expenditure.
It also studied the approximate enquiring method in sensor network in the prior art, but do not account for Privacy Protection.
On the other hand, the energy expenditure in sensor network often directly affects network life, and it is to pass to net interior data aggregation
Sensor network commonly reduces energy expenditure technology.In practical applications, acquired data values highests had usually both been needed to inquire (or most
It is low) preceding k (k is positive integer) a sensor node, and need inquire acquired data values some range sensor node, only
It can realize that the secret protection agreement of single query type is limited in practical applications.So research is poly- using data in net
The secret protection general polling method of collection has positive effect.
Invention content
In view of the deficiencies of the prior art, technical problem to be solved by the present invention lies in provide a kind of facing sensing device network
The general approximate enquiring method of secret protection.This method can realize that Top-k is looked under the premise of meeting secret protection requirement
The approximate queries such as inquiry, range query, SUM, MAX/MIN.
For the goal of the invention for realizing above-mentioned, the present invention uses following technical solutions:
A kind of general approximate enquiring method of secret protection of facing sensing device network is used in the similar sensor network of node
In, include the following steps:
(1) the data structure shared by base station and sensor node, the number of sensor node and gathered data are hidden
Among vector;
(2) aggregation node is transmitted the vector to base station by tree-shaped routing;
(3) system of linear equations is constructed in base station, solve the histogram with global statistics information and corresponding sensor node is compiled
Number;
(4) the statistical information being had according to the histogram completes required approximate query.
Wherein more preferably, the step (1) in, base station is that each sensor node generates unique random vector as ginseng
The amount of being pointed into is issued to corresponding sensor node after encryption;The sensor node is calculated using gathered data and with reference to vector
Response vector among the number of sensor node and gathered data are hidden in response vector, is route according to tree-shaped by the sound
Vector is answered to be uploaded to the father node as aggregation node.
Wherein more preferably, the step (2) in, the aggregation node using summation aggregate function to all data received
Assembled.
Wherein more preferably, the step (3) in, the base station constructs system of linear equations, solution using the response vector after aggregation
System of linear equations obtains the data interval that the gathered data of each sensor node is fallen into, and then obtains with global statistics information
The number of histogram and corresponding sensor node.
Wherein more preferably, the step (4) in, by adjust the granularity of division of the histogram balance inquiry precision with
Energy expenditure.
Wherein more preferably, the general approximate enquiring method of the secret protection has H-PGAQ patterns;
In the H-PGAQ patterns, the base station is that each sensor node generates unique perturbation vector and single channel is breathed out
Uncommon function, aggregation node executes sum operation to the vector received, and plus the vector that this node generates, meet with a response vector,
It uploads and participates in summation aggregation;The institute's directed quantity received is summed to obtain aggregation result vector, the aggregation result vector in base station
It subtracts through the processed noisy data of single channel hash function, obtains the system of linear equations.
Wherein more preferably, the general approximate enquiring method of the secret protection has F-PGAQ patterns;
In the F-PGAQ patterns, the base station is that each sensor node is generated with reference to vector;In continuous-query,
Histogram is divided and is used as filter window by the sensor node, if epicycle gathered data is fallen into together with upper wheel gathered data
One histogram divides, then does not upload data.
Wherein more preferably, when the scale of sensor network needs extension, the base station is that newly-increased sensor node is given birth to
At corresponding with reference to vector sum perturbation vector, and after the key encryption for using base station shared with sensor node, it is issued to newly-increased
Sensor node, while being added newly generated with reference to matrix with reference to vector, other sensors node is disturbed with reference to vector sum
Moving vector does not change.
Compared with prior art, the present invention propose it is a kind of it is novel can complete Top-k inquiries, range query, SUM,
The general approximate enquiring method of secret protection of the inquiries such as MAX/MIN.This method uses net in the case where not leaking privacy information
The strategy such as interior data aggregation and filter reduces approximate query energy expenditure;In the case where model need not be changed, pass through ginseng
Number is adjusted and can be realized to secret protection approximate query precision controlling.
Description of the drawings
Fig. 1 is the basic step schematic diagram of the general approximate enquiring method of secret protection provided by the present invention;
Fig. 2 is the schematic diagram that equi-depth histogram divides;
Fig. 3 is the schematic diagram that histogram mixing divides;
Fig. 4 is the traffic schematic diagram of system initialisation phase;
Fig. 5 is n=20, and in the case of t=100, figure is compared in influence of the interval division number to the traffic;
Fig. 6 is n=20, and in the case of t=100, figure is compared in influence of the gathered data wheel number to the traffic;
Fig. 7 is n=50, and in the case of t=100, figure is compared in influence of the interval division number to the traffic;
Fig. 8 is n=50, and in the case of t=100, figure is compared in influence of the gathered data wheel number to the traffic;
Fig. 9 is n=88, and in the case of t=100, figure is compared in influence of the interval division number to the traffic;
Figure 10 is n=88, and in the case of t=100, figure is compared in influence of the gathered data wheel number to the traffic;
Figure 11 is the relation schematic diagram of interval division number h and data value error threshold.
Specific implementation mode
The basic ideas of the general approximate enquiring method of secret protection (abbreviation PGAQ methods) provided by the present invention are to pass through
The data structure that base station is shared with sensor node is designed, the information such as sensor node number and the perception data of its acquisition are hidden
It is hidden in the random vector of construction.Intermediate aggregation node is transmitted the vector to base station by tree-shaped routing, is uploading this vector
During net interior summation aggregation to reduce energy expenditure.System of linear equations is constructed in base station, band global statistics is solved and believes
The histogram of breath and corresponding sensor node number.The statistical information being had according to histogram is completed with not revealing privacy
The approximate queries such as Top-k inquiries, range query, SUM, MAX/MIN, Median, Histogram.In above-mentioned PGAQ methods, make
With summation aggregation in net to reduce energy expenditure, and disappeared with energy to balance inquiry precision by adjusting histogram granularity of division
Consumption.Detailed specific description is unfolded in the following with reference to the drawings and specific embodiments.
PGAQ methods provided by the present invention are for the similar sensor network modelling of node.In this sensing
In device network model, net inner sensor node is all similar, all nets in terms of primary power, storage capacity, communication capacity
Interior nodes are likely to become the node that aggregation node etc. undertakes more multitask.By base station and n, (n is the sensor network model
Positive integer) a sensor node constitutes.The node serial number of each sensor is unique, is denoted as { s1,s2,...,sn}.Base station, which has, to be filled
Storage, calculating and the communication capacity of foot, and the topology information of the network overall situation can be obtained.Network can form tree-like road
By base station issues query demand by tree routing, and inquiry response data are uploaded to base by sensor node by tree routing
It stands.During uploading inquiry response, non-leaf nodes carries out aggregation to its child nodes data and this node gathered data and melts
It closes, only uploads data after aggregation.Aggregate function may be defined as y (t)=f (Vs1(t),...,Vsi(t),...,Vsn(t)), whereinFor node siIn the data that the t periods acquire.The present invention using summation aggregate function be
As shown in Figure 1, base station is each sensor node s firsti(si∈ G) unique random vector is generated as reference
Vector (a1i,a2i,…,ami)T, corresponding sensor node s is issued to after encryptioni.Sensor node siUse the number of its acquisition
According toWith reference vector (a1i,a2i,…,ami)T, calculate response vector (r1i,r2i,…,rmi)T, according to tree-shaped routing will respond to
Amount is uploaded to father node, and father node assembles all data received using summation aggregate function as aggregation node, raw
Then the response vector of cost sensor node sends aggregation result to its father node, is sequentially delivered to base station.Base station uses institute
There are the data received to calculate statistic histogram, secret protection approximate query is completed according to the histogram.
In PGAQ methods provided by the present invention, H-PGAQ and F-PGAQ both of which can be divided into.Wherein, H-PGAQ
Pattern uses disturbance of data technology and single channel hash function, to enhance Information Security.F-PGAQ patterns use filter
Reduce the continuous-query traffic.H-PGAQ mode expansions are described in detail first below.
In H-PGAQ patterns, it is divided into system initialization and two stages of inquiry.Hereinafter sensor network nodes are indicated with n
Number, to net inner sensor node carry out Unified number be expressed as { s1,s2,...,sn, with [vmin,vmax] indicate to pass
The codomain of sensor node gathered data.
(1) system initialisation phase
Base station 1,2,3 ..., 2γ(wherein γ is systematic parameter) middle m × n different positive integer a of random selectionij(i
=1,2 ..., m;J=1,2 ..., n), constitute m × n is with reference to matrix G:
Sensor node si(i=1,2 ..., n) and with reference to the i-th column vector (a in matrix G1i,a2i,…,ami)TIt is corresponding, it will
(a1i,a2i,…,ami)TAs sensor node siThe reference vector of (i=1,2 ..., n).
Base station is each sensor node si(i=1,2 ..., n) generate the perturbation vector (sn that unique and length is m1i,
sn2i,…,snmi)TWith single channel hash function h (), single channel hash function h () can map data into set 1,2 ...,
M }, wherein M=n × vmax×2γ, vmaxFor gathered data codomain [vmin,vmax] the upper bound.
In a base station, gathered data codomain [vmin,vmax] it is continuously divided into h section, it is expressed as (H1,…,Hh)。
Base station use and each sensor node si(i=1,2 ..., n) shared key is encrypted, in order to increase safety
Property, PGAQ uses k in such a way that the period replaces keyi,tIndicate period t inner sensor node siThe key shared with Sink, then
There is ki,t=hash (ki,t-1)。
WithCiphering process is indicated, to sensor node si(i=1,2 ..., n) the following message of transmission:
Sensor node siAfter (i=1,2 ..., n) receives message, using with base station shared key kiIt is decrypted, obtains
With the exclusive shared following information in base station:With reference to vector (a1i,a2i,…,ami)T, perturbation vector (sn1i,sn2i,…,snmi)T, it is single
Road hash function h () and gathered data codomain demarcation interval (H1,…,Hh)。
(2) inquiry phase
It is m that demarcation interval information where gathered data and sensor node number information are hidden in length by sensor node
Response vector in, and send aggregation node to.Aggregation node executes sum operation to the vector received, and is given birth to plus this node
At vector, obtain this node's length be m response vector, upload and participate in summation aggregation;Base station is to institute's directed quantity for receiving
Summation obtains aggregation result vector, and aggregation result vector subtracts noisy data, obtains using reference matrix as the linear of coefficient matrix
Equation group solves system of linear equations and obtains the histogram with statistical information and respective sensor node serial number, using histogram and
Respective sensor node serial number calculates approximate query result.
A. leaf node processing procedure
Leaf node siResponse vector algorithm is calculated by gathered data v using leaf nodei,tPlace demarcation interval information and
Sensor node number information is hidden in the response vector that length is m.The algorithm is described as follows:
Demand:With reference to vector (a1i,a2i,…,ami)T, perturbation vector (sn1i,sn2i,…,snmi)T, single channel hash function h
(), gathered data codomain demarcation interval (H1,…,Hh)
Leaf node siWhen having gathered data in polling cycle t, use above-mentioned algorithm calculate response vector for:
Wherein, NiFor sensor node siInstitute gathered data v in polling cycle ti,tPlace demarcation interval number, i.e.,
Leaf node siWhen there is no gathered data in polling cycle t, use algorithm 1 calculate response vector for:
(r1i,r2i,…,rmi)T=(h (sn1i),h(sn2i),…,h(snmi))T
Leaf node siBy response vector (r1i,r2i,…,rmi)TIt is uploaded to father node and participates in aggregation.
B. aggregation node accumulation process
Assuming that aggregation node sjThere is cnjA child nodes, child nodes number are respectively { j-cnj,j-cnj+1,...,j-
1 }, aggregation node sjGathered data is v in polling cycle tj,t.Aggregation node sjResponse vector algorithm is calculated using aggregation node
The vector of this node generation is summed and increased to the child nodes response vector received, generates aggregation node response vector simultaneously
It uploads.The algorithm is described as follows:
Demand:With reference to vector (a1j,a2j,…,amj)T, perturbation vector (sn1j,sn2j,…,snmj)T, single channel hash function h
(), gathered data codomain demarcation interval (H1,…,Hh)
Aggregation node sjBy response vector (r1j,r2j,…,rmj)TIt is uploaded to father node and participates in aggregation, if father node is not
Base station calculates response vector algorithm generation response vector using aggregation node and continues to upload, until being uploaded to base station.
C. base station processing procedure
After base station receives its all child nodes, following processing is carried out:
All child nodes response vectors are executed into sum operation, obtain aggregation result vector:
Aggregation result vector is subtracted into all nodes through the sum of processed noisy datas of single channel hash function h (), is obtained
To vector (b1,b2,…bn)T。
As
As following system of linear equations:
The coefficient matrix of equation group (1) is with reference to matrix G.When m is sufficiently large, set { N can be solved1,N2,…,
Nn}.It is n × n square formations with reference to matrix G, and have as m=n | G | ≠ 0.According to Cramer's rule (Cramer'sRule), work as m
Equation group (1) can be solved accurately when=n.Sensor node s1,s2,…,snArea in polling cycle t where gathered data
Between be respectivelyCount (H1,…,Hh) in each section sensor node gathered data for being included number and
Corresponding sensor node number, obtains with (H1,…,Hh) it is the statistic histogram divided.It can be solved according to histogram logical
With approximate query QGThe approximate query result for the query type T that middle user needs.
Above-mentioned H-PGAQ patterns use disturbance of data technology and single channel hash function, accordingly even when aggregation node can obtain
Sensor node and base station shared key, can decrypt and obtain the reference vector that base station issues, and aggregation node can not calculate
Go out perception data, enhance Information Security, still, H-PGAQ patterns need all the sensors node to upload data, continuously look into
Communication cost when inquiry is higher.
In F-PGAQ patterns, it is assumed that the key that aggregation node cannot obtain sensor node and base station is shared uses ginseng
The amount of being pointed into replaces true gathered data to carry out secret protection.Communication when in F-PGAQ patterns using filter reduction continuous-query
Histogram is divided and is used as filter window by amount, is drawn if epicycle gathered data falls into same histogram with upper wheel gathered data
Point, then it need not upload data.The processing procedure of F-PGAQ patterns and H-PGAQ patterns is substantially similar, and the main distinction is:(ⅰ)
Not to uploading vector addition noisy data in F-PGAQ patterns;Filter is used in (II) F-PGAQ patterns, in continuous-query,
If epicycle gathered data falls into same histogram with upper wheel gathered data and divides, data need not be uploaded.
In F-PGAQ patterns, it is also classified into system initialization and two stages of inquiry.It is described as follows:
(1) system initialisation phase
Base station uses the method demarcation interval (H in H-PGAQ patterns1,…,Hh), and be each sensor node si(i=
1,2 ..., n) it generates with reference to vector (a1i,a2i,…,ami)T.Using with sensor node siShared key ki,tIt is encrypted, base
It stands to sensor node si(i=1,2 ..., n) the following message of transmission:
Sensor node siDecryption obtains (a1i,a2i,…,ami)T(H1,…,Hh)。
(2) inquiry phase
A. leaf node processing procedure
In continuous-query, leaf node first determines whether the same section whether epicycle gathered data falls into upper wheel data
Divide (H1,…,Hh), if falling into same demarcation interval, data need not be uploaded.If epicycle gathered data and upper wheel number
According to falling in different sections, then calculates response vector algorithm (not adding noisy data) with reference to leaf node and calculate response vector
For:
(r1i,r2i,…,rmi)T=(a1i×Ni,a2i×Ni,…ami×Ni)
Leaf node siBy response vector (r1i,r2i,…,rmi)TIt is uploaded to father node and participates in aggregation.
B. aggregation node accumulation process
Aggregation node generates this knot vector using leaf node processing procedure.Aggregation node rings the child nodes received
Vector is answered to be summed and addedThe vector that this node generates, generates the response vector of aggregation node and upload.
If all child nodes do not upload data and aggregation node does not have gathered data, aggregation node not to need
Upload data.
C. base station processing procedure
Assuming that in epicycle inquiry, there is n in query regionG(nG≤ n) a sensor node uploads vector and participates in aggregation.For table
It states conveniently, it is assumed that uploading vectorial sensor node is.The child nodes response vector received is executed sum operation by base station, is obtained
To aggregation result vector:
As following system of linear equations:
When m is sufficiently large, set can be solvedWork as m=nGWhen equation group (2) can accurately ask
Solution.Base station is for no node s for uploading dataiUse last round of siCorresponding Ni, obtain adopting with query region whole node
Collect the histogram data { N of data information1,N2,…,Nn}.General approximate query Q is solved according to histogramGWhat middle user needed
The approximate query result of query type T.
F-PGAQ patterns are illustrated below.Assuming that sensor network is made of 3 nodes, tree-shaped routing is used
Transmission data, interior joint 2 and node 3 are leaf node, and node 1 is aggregation node.
In initial phase, base station randomly generates 3 × 3 matrix G and is:Using the 1st column vector in G as section
Point 1 is with reference to vector, i.e. (a11,a21,…,am1)T=(39,67,18)T;Using the 2nd column vector in G as node 2 with reference to vector, i.e.,
(a12,a22,…,am2)T=(92,25,59)T;Using the 3rd column vector in G as node 3 with reference to vector, i.e. (a13,a23,…,am3)T
=(72,82,30)T。
Assuming that gathered data codomain is [0,10], parameter h takes 10, using equivalent division methods, to gathered data codomain [0,
10] it is continuously divided, obtains (H1=[0,1), H2... ,=[1,2) H10=[9,10]).Base station use respectively with node 1,
Node 2 and 3 shared key of node are to (39,67,18)T、(92,25,59)T、(72,82,30)T(H1,H2…,H10) encryption, under
Respective nodes are sent to, intermediate node cannot be decrypted.Node 1 obtains this node with reference to vectorial (39,67,18) after decryptingTAnd division
(H1,H2…,H10), other nodes are same.If 1 acquired data values of node are 6.72, according to (H1,H2…,H10), because of 6.7 ∈ H7,
So N1=7, use (r1i,r2i,…,rmi)T=Ni×(a1i,a2i,…,ami)TCalculate response vector (r11,r21,…,rm1)T=
N1×(a11,a21,…,am1)T=7 × (39,67,18)T=(273,469,126)T;If 2 gathered data of node is 2.13, because
2.13∈H3, so N2=3, calculate (r12,r22,…,rm2)T=(276,75,177)T;If 3 gathered data of node is 8.29, because
For 8.29 ∈ H9, so N3=9, calculate (r13,r23,…,rm3)T=(648,738,270)T, node 2 and node 3 respectively will
(276,75,177)T(648,738,270)TIt is transmitted to node 1 (aggregation node), node 1 carries out summation aggregation operator and generates newly
Response vector (r11,r21,…,rm1)T=(273,469,126)T+(276,75,177)T+(648,738,270)T=(1197,
1282,573)T, node 1 is by (1197,1282,573)TIt is transmitted to base station.In a base station, (b is enabled1,b2,b3)T=(1197,1282,
573)T, according to the matrix G generated in initial phase
Following equation group is constructed,
It solves system of linear equations (3) and obtains N1=7, N2=3, N3=9.The sense of node 1, node 2 and node 3 is known in base station in this way
Primary data respectively falls in H7=[6,7), H3=[2,3), H9=[8,9) section.Base station counts histogram according to these data configurations
Figure, realizes general approximate query.
In the present invention, interrogation model is divided into interrogation model in general approximate query model and net.User uses general close
Querying command is sent to base station like interrogation model, sensor node is handed down in base station using interrogation model in net by routing tree,
Perception data is assembled and is uploaded by non-leaf nodes by sensor node, and histogram, root are derived in base station according to aggregation result
Go out the general approximate query of user's needs as a result, being returned to user according to histogram calculation.General approximate query model description
For:
QG=(query region=G) ∧ (epoch=t) ∧ (query type=T)
∧ (query parameter=P) ∧ (approximate parameter=ò)
Wherein G is query region, and t is polling cycle, and T is the query type set that user's needs return, and P is to be inquired in T
Parameter sets needed for type, ò are approximate query precision threshold parameter.
In PGAQ methods provided by the present invention, the query type supported has a Top-k inquiries, range query, MAX
MIN, SUM, Median, Histogram etc..It can be single query type in T, can also be that multiple PGAQ methods are supported to look into
Ask type.Parameter should be corresponding with query type in T in P.When query type is that Top-k is inquired, parameter is k values in P;Inquiry
When type is range query, parameter is query context [a, b] in P.
Interrogation model is in the net that base station is issued by tree routing to query region G inner sensor nodes:
Qt={ t, (H1,…,Hh)}
Wherein t is polling cycle, (H1,…,Hh) it is the histogram area calculated by approximate query precision threshold parameter ò
Between divide, if ò be default value, then need not transmit (H1,…,Hh)。
Secret protection approximate enquiring method provided by the present invention needs to meet following performance requirement and balance requires:
(1) precision is inquired
In the case of not high to required precision, approximate query can be balanced between precision and energy expenditure, with
Reduce energy expenditure.Meanwhile the precision of approximate query should be able to meet the needs of application, so approximate query precision should be able to lead to
It crosses systematic parameter to be controlled, to reach the balance met between application demand precision and energy expenditure.
(2) secret protection demand
It is attacked since attacker can be eavesdropped and be captured sensor node by Radio Link, secret protection approximation is looked into
Inquiry method should meet claimed below:In order to protect privacy information, arbitrary node s in sensor networki(si∈ G) acquisition data
It can not be known by Radio Link listener-in, while can not be by any other node s in networkj(sj∈ G and i ≠ j) know;
Arbitrary node cannot know query result data set in sensor network;Arbitrary node cannot know inquiry knot in sensor network
The corresponding sensor node ID of fruit data item.
(3) energy consumption
Energy expenditure often directly affects sensor network life, and it is that sensor network protocol is important to reduce energy expenditure
Performance requirement.Secret protection technology generally requires to increase sensor network communication amount and energy expenditure, secret protection vlan query protocol VLAN
The traffic and energy expenditure should be reduced to the greatest extent.Data aggregation and fusion can efficiently reduce volume of transmitted data in net, and then subtract
Few energy expenditure.Data aggregation and fusion application in net are of great significance in secret protection approximate enquiring method.
For this purpose, balancing inquiry precision and energy expenditure by adjusting histogram granularity of division in the present invention.Wherein, it inquires
Precision is mainly by systematic parameter h and to section (H1,...,Hh) dividing mode influence.Section (H1,...,Hh) dividing mode determines
It is scheduled on the type for the histogram that base station is calculated, systematic parameter h determines section (H1,...,Hh) granularity of division.In order to control
The needs of precision, the present invention use two kinds of histogram dividing modes of wide histogram and equi-depth histogram.Histogram dividing mode
It can be selected by the user of sensor network system.
By further investigation, inventor thinks that approximate query error can be described by two parameters, and respectively data value misses
Difference òvWith data item number error amount òn.Data value error amount òvThe middle data value range error upper limit is returned the result for inquiry, i.e.,
max{|vpmax-vrmax|,|vpmin-vrmin|}≤òv.Wherein, vpmaxAnd vpminRespectively RpMiddle maximum value and minimum value, vrmaxWith
vrminRespectively RrMiddle maximum value and minimum value, RpFor the approximate query result data item set that base station returns, RrTo meet inquiry
The accurate inquiry collection of data items of condition.Data item number error amount ònThe number of querying condition is not met in being returned the result for inquiry
According to the item number upper limit, i.e., | | Rp|-|Rr||≤òn.Wherein, RpFor the approximate query result data item set that base station returns, RrTo meet
The accurate inquiry collection of data items of querying condition.
On this basis, inventor has found that using wide histogram dividing mode approximate query result data values can be controlled
Error amount òv, so wide histogram dividing mode is more suitable for being used for range query type.When query type T is range query
When, parameter sets P is range query parameter [a, b].The minimum histogram that PGAQ methods calculate covering [a, b] divides set, will
It falls with the histogram and divides result set of the data item of set as approximate query, return to user.
On the other hand, inventor using equi-depth histogram dividing mode it has also been found that can control approximate query result data item
Number error amount òn.Assuming that statistical distribution functions f (x) is normal distribution f (x)=N (μ, σ), it is 7 to divide number h, equi-depth histogram
The schematic diagram of division is as shown in Figure 2.It is missed since equi-depth histogram dividing mode can control approximate query result data item number
Difference òn, so histograms dividing mode is waited to be more suitable for being used for Top-k query types.When query type T inquires for Top-k
When, parameter sets P is Top-k query arguments k.The data item number that the calculating of PGAQ methods falls into highest (low) demarcation interval is more than
(being less than) is equal to the minimum demarcation interval set of k, will fall with the histogram and divides the data item of set as approximate query
Result set, return to user.
Mixing dividing mode can be used in query result data value error and data item number error, histogram in order to balance.
Assuming that statistical distribution functions f (x) is normal distribution f (x)=N (μ, σ), it is 9 to divide number h, the schematic diagram that histogram mixing divides
As shown in Figure 3.Wherein, H1=H2=H8=H9,
Falling into the intensive region H of data itemconUsing the dividing mode of equi-depth histogram.In region HconInside haveDue in region HconIt is interior to control approximate query result data item number error
Value òn, simultaneously as region HconData item comparatively dense is inside fallen into, the dividing regions obtained using the dividing mode of equi-depth histogram
Between it is relatively narrow so that data value error is smaller, has taken into account query result data value error.On the other hand, to fall into data item sparse
Region HsparUsing the dividing mode of wide histogram, in region HsparInside haveDue in area
Domain HsparIt is interior to control approximate query result data values error amount òv, simultaneously as region HsparIt is diluter inside to fall into data item
It dredges, the data item number for falling into each demarcation interval will be kept less using wide histogram dividing mode so that data item number
Error is smaller, has taken into account query result data item number error.
In sensor network, the attack for private data can be divided into external attack and internal attack two kinds of attack moulds
Type.Since sensor network is using wirelessly communicating, attacker may be eavesdropped by link layer and be obtained when data are transmitted among the nodes
Sensitive data, this attack mode is taken to be known as external attack.When being internaled attack, attacker is by capturing or replicating sensor
The means such as node become the participant of network, can obtain all data of capture sensor node.The present invention is based on inventors
Previously used sincere but curious model is (referring to the paper of Fan Yongjian, Chen Hong《Secret protection is can verify that in two layers of sensor network
Top-k vlan query protocol VLANs》, it is published in《Chinese journal of computers》The 3rd phase of volume 35 in 2012), attacker is by recording or speculating sensitivity
Data carry out steal information, but it is abided by the agreement and not altered data, only destroy the privacy of data but do not destroy the complete of data
Whole property.On this basis, it is assumed that attacker steals sensitive information using two kinds of attack patterns:(1) external attack pattern, number are used
Sensitive data is obtained according to being eavesdropped by link layer when transmitting among the nodes;(2) pattern is internaled attack in use, captures sensor section
By recording or speculating that sensitive data carries out steal information after point.The corresponding security target of the present invention is:In sincere but curious mould
Under type, PGAQ methods cope with external attack and internal attack the attack of both of which, the case where not leaking privacy information
It is lower to complete general approximate Aggregation Query using data aggregation.Specifically, the general approximate query Q in sensor networkG=G,
T, T, P, ò } it is correct when executing, if attacker cannot obtain following sensitive information, inquire QGMeet the requirement of privacy:
(1) query region G inner sensors node si(i=1,2 ..., n) acquired data values vi,t;(2) include gathered data statistical information
Histogram information;(3) the corresponding sensor node number of query result data.
For this purpose, the present invention considers that attacker can carry out internaling attack and by eavesdropping link by capture aggregation node
Two kinds of attack patterns of external attack of layer.By further investigation, assuming that aggregation node can not obtain sensor node and base
It stands in the case of shared key, F-PGAQ patterns disclosure satisfy that the requirement of general approximate query privacy in sensor network.
It is described as follows:In the system initialisation phase of F-PGAQ patterns, base station reference corresponding to each sensor node is vectorial,
It is encrypted using itself and the key that sensor node is shared, sensor node is issued to, as it is assumed that aggregation node can not obtain
The key that sensor node and base station are shared, aggregation node cannot be obtained with reference to vector.Leaf node by gathered data information and
Node serial number Information hiding is in response vector and is uploaded to aggregation node, and aggregation node is not known with reference to vector, so cannot push away
Export gathered data information and node ID information.Simultaneously as the data transmitted in Radio Link are response vector, so external
Attacker can not obtain above-mentioned sensitive information.
On the other hand, no matter whether aggregation node can obtain the shared key in sensor node and base station, H-PGAQ moulds
Formula can meet the requirement of general approximate query privacy in sensor network.It is described as follows:In the base of F-PGAQ patterns
On plinth, H-PGAQ patterns use disturbance of data technology and single channel hash function.As the above analysis, aggregation node can not
In the case of the key that acquisition sensor node and base station are shared, F-PGAQ patterns meet general approximate query in sensor network
Privacy.When aggregation node can obtain the shared key in sensor node and base station, aggregation node can know with reference to
Amount, but since, comprising the processed noisy data of single channel hash function, aggregation node is same in response vector in H-PGAQ patterns
Sample cannot obtain above-mentioned sensitive information.
In addition, using PGAQ methods provided by the invention, in system initialisation phase, base station use is total with sensor node
For the delivering key enjoyed with reference to vector, each sensor nodes of m cannot obtain the reference vector of other sensors node, therefore even if obtain
Know that the response vector of other sensors node can not obtain other sensors node truthful data by collusion (ganging up).
In terms of energy expenditure, the energy expenditure of sensor network is mainly influenced by systematic parameter.In PGAQ methods, leaf
Child node is to the response vector that aggregation node conveying length is m, and aggregation node is by the response vector of all child nodes received
Summation aggregation is carried out, and this nodal information is added and generates the response vector that this node's length is m, upload and participates in the poly- of father node
Collection.So in PGAQ methods, what is uploaded in aggregation routing tree is the vector that length is m.Below by emulation data to being
System parameter and the energy expenditure relationship of sensor network are tested and are analyzed.
The cost of PGAQ methods provided by the invention can be divided into following two parts:In system initialisation phase, transmission adds
Close reference vector sum interval division cost;Inquiry phase be mainly sensor node gathered data, generate aggregation response to
The cost of amount and transmission response vector.Wherein network system performance is mainly influenced by inquiry phase cost.For sensor node,
Communication energy consumption evaluates the validity of algorithm using data traffic much larger than energy expenditure, the present invention is calculated.
The present invention uses publicly available real data set LUCE dataset, to the side PGAQ on OMNeT++ platforms
Method is tested.LUCE dataset data sets are that 11 dimension such as environment temperature, soil moisture of acquisition in 2006 and 2007 belongs to
Property data, use environment temperature data of the present invention tested, and the experimental data codomain used is [- 20,30].The present invention's
In one embodiment, using the node location data in LUCE dataset, 3 groups of experiments are divided into according to scale.450 × 300 is false
If sensor node effective propagation path is 60 meters, is established and route using TAG algorithms.1st group of experiment, 20 sensor nodes point
Cloth is in 200 × 200 square metres of rectangular areas, and sensor node averagely needs 2.5 jumps to unit header node transmission data, each
Sensor node averagely has 4.5 neighbor nodes;2nd group of experiment, 50 sensor nodes are distributed in 350 × 250 square metres of squares
In shape region, sensor node averagely needs 2.9 jumps, each sensor node averagely to have 6.4 to unit header node transmission data
A neighbor node;3rd group of experiment, 88 sensor nodes are distributed in square metre rectangular area, and sensor node is to unit header
Node transmission data averagely needs 3.1 jumps, each sensor node averagely to have 11.1 neighbor nodes.
Fig. 4 shows PGAQ methods in system initialisation phase, the experiment of 3 groups of heterogeneous networks scales, and base station is to sensing
The traffic that device node-node transmission encryption data needs.In above-mentioned experiment, the present invention is encrypted using 16 keys.
In order to investigate PGAQ methods inquiry when the traffic, the present invention by H-PGAQ patterns, F-PGAQ patterns and KIPDA,
End-to-end (end to end) encipherment scheme is compared.In 3 groups of different experiments of network size, section has been investigated respectively and has been drawn
The wheel number of the number and gathered data divided is to query communication amount CQInfluence.For ease of description, 1 chronomere of this experimental hypothesis
1 wheel data of sensor node acquisition, the i.e. interior acquisition t of polling cycle t take turns data.What this experiment used is equal to reference to vector length m
Net internal segment count n, with ensure base station can Exact Solutions linear equation group, obtain histogram information.It is set in KIPDA in this experiment
Parameter k is 5.
Fig. 5, Fig. 7 and Fig. 9 are respectively illustrated when polling cycle t is 100 (when i.e. gathered data wheel number is 100 times), area
Between divide influences of the number h to the sensor node continuous data transfer traffic.It can be seen that:KIPDA and H-PGAQ patterns are led to
Traffic is higher than the traffic of F-PGAQ patterns, and KIPDA is slightly above H-PGAQ patterns, this is because F-PGAQ patterns have used filtering
Device mechanism, reduces the traffic.In the 1st group of experiment, the end-to-end scheme traffic is less than KIPDA and H-PGAQ patterns, is higher than
F-PGAQ patterns.In the 2nd group and the 3rd group of experiment, the end-to-end scheme traffic is less than KIPDA and H-PGAQ patterns and F-
The traffic curve of PGAQ patterns has intersection, and the 2nd group of crosspoint is near h=175, and the 3rd group of crosspoint is near h=100.Area
Between divide number h to KIPDA and H-PGAQ mode influences very littles, and the traffic of F-PGAQ patterns increases with h and is increased, this be because
The width of filter window in F-PGAQ is determined for interval division number h, interval division number h is bigger, is filtered in F-PGAQ patterns
Device window is narrower, and the traffic is bigger, while it is higher to inquire precision.Data value error amount òvThe pass of number h is divided with wide histogram
System is òv=(vmax-vmin)/h.Figure 11 shows when gathered data codomain is [- 20,30], interval division number h and data value
The relationship of error threshold.
Fig. 6, Fig. 8 and Figure 10 are respectively illustrated when interval division number h is 100, gathered data wheel number t (i.e. polling cycles
T) to the influence of the sensor node continuous data transfer traffic.It can be seen that:The traffic is followed successively by from high to low:KIPDA,H-
PGAQ patterns, end-to-end scheme and F-PGAQ patterns.KIPDA is slightly above H-PGAQ patterns.KIPDA, H-PGAQ pattern and end are arrived
The end scheme traffic is higher than the traffic of F-PGAQ patterns, same because F-PGAQ patterns have used filter mechanism, reduces
The traffic.KIPDA, H-PGAQ pattern, end-to-end scheme and F-PGAQ patterns all increase with gathered data wheel number and are increased, because
KIPDA, H-PGAQ pattern and end-to-end scheme do not use strobe utility, so increasing with gathered data wheel number substantially linear
Increase.
As can be seen from the above analysis, H-PGAQ patterns use disturbance of data technology and single channel hash function, accordingly even when
Aggregation node can obtain sensor node and base station shared key, can decrypt and obtain the reference vector that base station issues, gather
Collection node can not calculate perception data, enhance Information Security, but H-PGAQ patterns need all the sensors node
Data are uploaded, communication cost is higher when continuous-query.F-PGAQ patterns assume aggregation node can not obtain sensor node and
The shared key in base station carries out secret protection using with reference to vector.F-PGAQ patterns reduce continuous-query using filter and communicate
Histogram is divided and is used as filter window by amount, is drawn if epicycle gathered data falls into same histogram with upper wheel gathered data
Point, then it need not upload data.By theory analysis and experimental verification, F-PGAQ patterns significantly reduce the traffic and energy disappears
Consumption.
On the other hand, in terms of network size, PGAQ methods can more easily adapt to the extension of network size.In PGAQ
In method, base station is each sensor node s in neti(i=1,2 ..., n) the different reference vector (a of distribution1i,a2i,…,
ami)T, base station needs generation m × n to need to be not less than net inner sensor node number with reference to matrix G, n.When sensor network scale
When needing extension, base station is that newly-increased sensor node generation is corresponding with reference to vector sum perturbation vector, and uses base station and biography
After the shared key encryption of sensor node, it is issued to newly-increased sensor node, while newly generated be added with reference to vector being joined
According to matrix, other sensors node need not change with reference to vector sum perturbation vector.In this way, generating new aggregation routing
Afterwards, PGAQ methods can execute on the network after expansibility of network size.
The general approximate enquiring method of sensor network secret protection provided by the present invention has been carried out specifically above
It is bright.For those of ordinary skill in the art, it is done under the premise of without departing substantially from true spirit any aobvious
And the change being clear to, it will all constitute to infringement of patent right of the present invention, corresponding legal liabilities will be undertaken.
Claims (9)
1. a kind of general approximate enquiring method of secret protection of facing sensing device network is used in the similar sensor network of node
In, it is characterised in that include the following steps:
(1) base station is that each sensor node generates unique random vector as with reference to vector, and corresponding biography is issued to after encryption
Sensor node;The sensor node calculates response vector, passes through base station and sensor section using gathered data and with reference to vector
It, will according to tree-shaped routing among the number of sensor node and gathered data are hidden in response vector by the shared data structure of point
The response vector is uploaded to the father node as aggregation node;
(2) aggregation node is transmitted the vector to base station by tree-shaped routing;
(3) system of linear equations is constructed in base station, solve the histogram with global statistics information and corresponding sensor node number;
(4) the statistical information being had according to the histogram completes required approximate query.
2. the general approximate enquiring method of secret protection as described in claim 1, it is characterised in that:
The step (2) in, the aggregation node using summation aggregate function all data received are assembled.
3. the general approximate enquiring method of secret protection as described in claim 1, it is characterised in that:
The step (3) in, the base station constructs system of linear equations using the response vector after aggregation, and solution system of linear equations obtains
The data interval that the gathered data of each sensor node is fallen into, and then obtain the histogram with global statistics information and corresponding
The number of sensor node.
4. the general approximate enquiring method of secret protection as described in claim 1, it is characterised in that:
The step (4) in, balance inquiry precision and energy expenditure by adjusting the granularity of division of the histogram.
5. the general approximate enquiring method of secret protection as claimed in claim 4, it is characterised in that:
Approximate query result data values error amount is controlled using wide histogram dividing mode.
6. the general approximate enquiring method of secret protection as claimed in claim 4, it is characterised in that:
Approximate query result data item number error amount is controlled using equi-depth histogram dividing mode.
7. the general approximate enquiring method of secret protection as described in claim 1, it is characterised in that:
The general approximate enquiring method of secret protection has H-PGAQ patterns;
In the H-PGAQ patterns, the base station is that each sensor node generates unique perturbation vector and single channel Hash letter
Number, aggregation node executes sum operation to the vector received, and plus the vector that this node generates, meet with a response vector, uploads
And participate in summation aggregation;Base station sums the institute's directed quantity received to obtain aggregation result vector, and the aggregation result vector subtracts
Through the processed noisy data of single channel hash function, the system of linear equations is obtained.
8. the general approximate enquiring method of secret protection as described in claim 1, it is characterised in that:
The general approximate enquiring method of secret protection has F-PGAQ patterns;
In the F-PGAQ patterns, the base station is that each sensor node is generated with reference to vector;It is described in continuous-query
Histogram is divided and is used as filter window by sensor node, if epicycle gathered data is fallen into upper wheel gathered data with always
Side's figure divides, then does not upload data.
9. the general approximate enquiring method of secret protection as described in claim 1, it is characterised in that:
When the scale of sensor network needs extension, the base station is that newly-increased sensor node generates accordingly with reference to vector
And perturbation vector, and after the key encryption for using base station shared with sensor node, it is issued to newly-increased sensor node, simultaneously
It is added newly generated with reference to matrix with reference to vector, other sensors node does not change with reference to vector sum perturbation vector.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310389413.9A CN103491535B (en) | 2013-08-31 | 2013-08-31 | The general approximate enquiring method of secret protection of facing sensing device network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201310389413.9A CN103491535B (en) | 2013-08-31 | 2013-08-31 | The general approximate enquiring method of secret protection of facing sensing device network |
Publications (2)
Publication Number | Publication Date |
---|---|
CN103491535A CN103491535A (en) | 2014-01-01 |
CN103491535B true CN103491535B (en) | 2018-10-30 |
Family
ID=49831427
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201310389413.9A Active CN103491535B (en) | 2013-08-31 | 2013-08-31 | The general approximate enquiring method of secret protection of facing sensing device network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN103491535B (en) |
Families Citing this family (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103945439B (en) * | 2014-04-21 | 2017-05-24 | 清华大学 | Precise query method for perception data in wireless sensor network |
EP3371994A4 (en) | 2015-11-03 | 2019-04-10 | Nokia Technologies OY | Apparatus, method and computer program product for privacy protection |
CN105515881A (en) * | 2016-01-29 | 2016-04-20 | 晏平 | Routing analysis method |
US10467433B2 (en) * | 2017-03-17 | 2019-11-05 | Mediasift Limited | Event processing system |
CN108616837A (en) * | 2018-04-13 | 2018-10-02 | 中南大学 | A kind of Top-k query method for the double-deck Sensor Network |
CN108737433B (en) * | 2018-05-29 | 2020-09-11 | 深圳市齐为信息科技有限公司 | Sensor node monitoring system and sensor node monitoring method based on Internet of things |
CN108833098B (en) * | 2018-07-05 | 2021-08-03 | 西北大学 | Wireless sensor network key management method based on linear equation block solution |
CN116821193B (en) * | 2023-08-30 | 2024-01-09 | 之江实验室 | Reasoning query optimization method and device based on proxy model approximation processing |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102196431A (en) * | 2011-05-13 | 2011-09-21 | 南京邮电大学 | Internet of things application scene-based protection method of privacy query and private identity verification |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7917517B2 (en) * | 2008-02-28 | 2011-03-29 | International Business Machines Corporation | Method and apparatus for query processing of uncertain data |
-
2013
- 2013-08-31 CN CN201310389413.9A patent/CN103491535B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102196431A (en) * | 2011-05-13 | 2011-09-21 | 南京邮电大学 | Internet of things application scene-based protection method of privacy query and private identity verification |
Non-Patent Citations (5)
Title |
---|
Verifiable Privacy-Preserving Sensor Network Storage for Range Query;Bo Sheng and Qun Li;《IEEE》;20111231;全文 * |
Verifiable Privacy-PreserviTwo-tiered Sensor Networksng Range Query in;Bo Sheng and Qun Li;《IEEE》;20081231;全文 * |
两层传感器网络中隐私与完整性保护的范围查询协议;李睿 林亚平 易叶青 胡玉鹏;《计算机学报》;20130630;第1-8节 * |
基于Z-O 编码的两层WSNs 隐私保护最值查询处理协议;戴 华*① 秦小麟② 刘 亮② 季一木① 付 雄① 孙 研②;《电子与信息学报》;20130430;全文 * |
无线传感器网络数据隐私保护技术;范永健1),2),3) 陈 红1),2) 张晓莹1),2);《计算机学报》;20120630;全文 * |
Also Published As
Publication number | Publication date |
---|---|
CN103491535A (en) | 2014-01-01 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN103491535B (en) | The general approximate enquiring method of secret protection of facing sensing device network | |
Liu et al. | Detection of multiple-mix-attack malicious nodes using perceptron-based trust in IoT networks | |
Al Hinai et al. | Internet of things: Architecture, security challenges and solutions | |
Liu et al. | Trace malicious source to guarantee cyber security for mass monitor critical infrastructure | |
Wei et al. | Research and simulation of queue management algorithms in ad hoc networks under DDoS attack | |
Cvitić et al. | CLASSIFICATION OF SECURITY RISKS IN THE IOT ENVIRONMENT. | |
Al Shuhaimi et al. | Software defined network as solution to overcome security challenges in IoT | |
CN107135072A (en) | A kind of wireless sensor network system based on quantum cryptography | |
Venkatasubramanian et al. | Iot malware analysis using federated learning: A comprehensive survey | |
Chandnani et al. | An analysis of architecture, framework, security and challenging aspects for data aggregation and routing techniques in IoT WSNs | |
Zakarya | DDoS verification and attack packet dropping algorithm in cloud computing | |
CN103441952B (en) | Based on multinuclear or the processing method of network data packets of many-core flush bonding processor | |
Narayana et al. | An efficient methodology for avoiding threats in smart homes with low power consumption in iot environment using blockchain technology | |
Rouzbahani et al. | Communication layer security in smart farming: A survey on wireless technologies | |
CN109858510A (en) | A kind of detection method for http protocol ETag value covert communications | |
Park et al. | PoAh-Enabled Federated Learning Architecture for DDoS Attack Detection in IoT Networks | |
Huang et al. | Trust mechanism-based multi-tier computing system for service-oriented edge-cloud networks | |
Liu et al. | A new Sybil attack detection for wireless body sensor network | |
CN110166243A (en) | Aggregated data method for secret protection based on elliptic curve in a kind of intelligent perception | |
Nour-El Aine et al. | Ensuring smart agriculture system communication confidentiality using a new network steganography method | |
Rajesh et al. | Fuzzy logic based performance optimization with data aggregation in wireless sensor networks | |
Zhang et al. | Link-based privacy-preserving data aggregation scheme in wireless sensor networks | |
Han et al. | A new private-preserving algorithm based on SMART | |
Cui et al. | Trillion Sensors Security | |
Lee et al. | A responsive probing approach to detect dynamic intrusion in a manet |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |