CN107369125A - Robust watermarking detection method and device for data flow - Google Patents
Robust watermarking detection method and device for data flow Download PDFInfo
- Publication number
- CN107369125A CN107369125A CN201710600765.2A CN201710600765A CN107369125A CN 107369125 A CN107369125 A CN 107369125A CN 201710600765 A CN201710600765 A CN 201710600765A CN 107369125 A CN107369125 A CN 107369125A
- Authority
- CN
- China
- Prior art keywords
- data
- watermark
- groups
- data item
- difference
- 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.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
- G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0053—Embedding of the watermark in the coding stream, possibly without decoding; Embedding of the watermark in the compressed domain
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0065—Extraction of an embedded watermark; Reliable detection
Abstract
The present invention relates to a kind of robust watermarking detection method and device for data flow, wherein method of detecting watermarks comprises the following steps：The data item for forming data flow is obtained, the data item comprises at least perception data and timestamp, and the data item for carrying watermark is determined with Probability p according to timestamp；Hash function calculating is carried out to timestamp in the data item for carrying watermark using packet key, data item is divided into by A groups and B groups according to obtained cryptographic Hash；Calculate the accumulated value of the difference of the perception data of two groups of data of A groups and B groups, and when the data item number of two groups of data is more than predetermined number, the mean difference of two groups of data is calculated, judges that watermark is present when the mean difference is more than predetermined threshold value, otherwise it is assumed that watermark is not present.The present invention determines to carry the data item of watermark and data item is grouped according to timestamp, it is ensured that the successful detection of watermark, while the resistivity to attack has been ensured again.
Description
The application divides for entitled " robust watermarking embedding grammar and detection method and device for data flow "
Case application, applying date of original application are 2017.6.13, Application No. 201710441578.4.
Technical field
The present invention relates to the robust watermarking detection method and device of a kind of water mark method, more particularly, to data flow.
Background technology
With the development of technology of Internet of things, increasing small intelligent hardware is widely used, such as bracelet, environment
Detection means etc..This kind equipment can be by various sensor gathered datas, and by data buffer storage or transmission.These data
Often endlessly produce, and transmitted in the form of data flow.The number gathered under the epoch of big data, many application scenarios
According to possessing commercial value, therefore copyright protection is one and has to solve the problems, such as.However, this kind equipment often has one altogether
Same feature, that is, resource-constrained, be specifically exactly computing resource, storage resource and finite energy.The numeral of robust
Watermark is just highly suitable for the copyright protection for solving the flow data in resource constrained environment.
Digital watermarking is divided into fragile watermark and robust watermarking according to the power of its robustness.Fragile watermark to distorting sensitivity,
Therefore it is commonly used to detect the integrality of watermark carrier；Robust watermarking has stronger resistivity for attack, therefore is commonly used to deposit
Copyright information is stored up, makes it be not easy to be erased by common signal processing means, plays a part of copyright protection.
The data flow transmitted in resource constrained environment network is mainly numeric type data, and implementing robust watermarking may generally meet with
Meet following several attack patterns：
1. intercepting, i.e., intercept Partial Fragment from endless data flow；
2. sampling, the data item in data flow is uniformly chosen with certain sample rate and forms new sampled data stream；
3. deleting, data flow is in transmitting procedure because active attack or network environment lost part data item；
4. injecting, the data item of forgery is inserted in a stream；
5. distort, in a variety of ways the perception data or timestamp of altered data item；
The target of these attacks is consistent, i.e., its hiding water is destroyed under the premise for not influenceing data use value
Official seal ceases, and can not just judge the copyright information of data so that all rights reserved, realize the illegitimate target that data are usurped.
It would therefore be highly desirable to develop the robust watermarking insertion and inspection of a kind of data flow for being directed to and being transmitted in resource constrained environment network
Survey method, so as to improve the resistivity to attack, play a part of copyright protection.
The content of the invention
The technical problem to be solved in the present invention is, lacks be directed to what is transmitted in resource constrained environment network in the prior art
Data flow carries out the method that effectively robust watermarking is embedded and detects, there is provided a kind of robust watermarking embedding grammar for data flow
And method of detecting watermarks and device.
First aspect present invention, there is provided a kind of robust watermarking embedding grammar for data flow, comprise the following steps：
Data decimation step, obtain the data item s for forming data flowi, the data item siIncluding at least perception data and
Timestamp ti, wherein i is the sequence number of the data item in a stream, and selection meets the data item of below equation for for carrying water
The data item of print：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability；
Packet step, using packet key K2Hash function is carried out to timestamp in the data item for carrying watermark
Calculate, the data item is included in by A groups or B groups according to obtained cryptographic Hash；
Numerical value increases and decreases step, to the perception data in A group data item increase watermark modification amount δ computing, to B group numbers
According to the perception data in item reduce watermark modification amount δ computing, to realize the insertion of watermark.
Second aspect of the present invention, there is provided a kind of robust watermarking detection method for data flow, comprise the following steps：
Data determination, obtains the data item for forming data flow, and the data item comprises at least perception data with timely
Between stab ti, wherein i is the sequence number of the data item in a stream, it is determined that the data item for meeting below equation is for carrying watermark
Data item：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability；
Data buffer storage step, using packet key K2Hash function is carried out to timestamp in the data item for carrying watermark
Calculate, data item be divided into by A groups and B groups according to obtained cryptographic Hash, and according to it is affiliated be grouped data item is cached in corresponding to
In two queues；
Difference detecting step, the accumulated value of the difference of the perception data of two groups of data of A groups and B groups is calculated, and in two groups of data
Data item number when being more than predetermined number, the mean difference of two groups of data is calculated, when the mean difference is more than predetermined threshold value
Judge that watermark is present, otherwise it is assumed that watermark is not present.
Third aspect present invention, there is provided a kind of robust watermarking flush mounting for data flow, including：
Data decimation module, the data item s of data flow is formed for obtainingi, the data item siIncluding at least perception data
And timestamp ti, wherein i is the sequence number of the data item in a stream, and selection meets the data item of below equation for for holding
Carry the data item of watermark：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability；
Packet module, for using packet key K2Hash is carried out to timestamp in the data item for carrying watermark
Function is calculated, and the data item is included in into A groups or B groups according to obtained cryptographic Hash；
Numerical value swap modules, for the perception data in A group data item increase watermark modification amount δ computing, to B
Perception data in group data item reduce watermark modification amount δ computing, to realize the insertion of watermark.
Fourth aspect present invention, there is provided a kind of robust watermarking detection means for data flow, including：
Data determining module, for obtain form data flow data item, the data item comprise at least perception data with
And timestamp ti, wherein i is the sequence number of the data item in a stream, it is determined that the data item for meeting below equation is for carrying
The data item of watermark：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability；
Data cache module, for using packet key K2Hash is carried out to timestamp in the data item for carrying watermark
Function is calculated, and data item is divided into A groups and B groups according to obtained cryptographic Hash, and is cached in data item pair according to affiliated be grouped
In two queues answered；
Difference detection module, the accumulated value of the difference of the perception data for calculating two groups of data of A groups and B groups, and at two groups
When the data item number of data is more than predetermined number, the mean difference of two groups of data is calculated, is more than default threshold in the mean difference
Judge that watermark is present during value, otherwise it is assumed that watermark is not present.
Implement the present invention, have the advantages that：The present invention selectes the data item of carrying watermark according to timestamp,
Data item is grouped also according to timestamp, and timestamp is not repaiied during the insertion and detection of watermark
Change, this guarantees data flow both ends can obtain identical packet, it is ensured that the successful detection of watermark, version can be played
The effect of protection is weighed, while has ensured the resistivity to attack again.
Brief description of the drawings
Fig. 1 is the robust watermarking embedding grammar flow chart for data flow according to the preferred embodiment of the present invention；
Fig. 2 is the robust watermarking detection method flow chart for data flow according to the preferred embodiment of the present invention；
Fig. 3 is the module diagram according to the robust watermarking flush mounting for data flow of the preferred embodiment of the present invention；
Fig. 4 is the module diagram according to the robust watermarking detection means for data flow of the preferred embodiment of the present invention；
Fig. 5 is difference detection module in the robust watermarking detection means for data flow according to the preferred embodiment of the present invention
Schematic diagram；
Fig. 6 shows the one section of temperature data stream gathered by sensor；
Fig. 7 shows the temperature data stream after embedded watermark；
Fig. 8 shows loss and false alarm rate under different statistical magnitude N；
Fig. 9 shows loss and false alarm rate under the Probability p of different watermark insertions；
Figure 10 shows loss and false alarm rate under different watermark modified values；
Figure 11 shows loss and false alarm rate under different threshold values；
Figure 12 shows the watermark detection rate under different sample rates；
Figure 13 shows the watermark detection rate under different injection rates；
Figure 14 shows the watermark detection rate under different modification rates.
Embodiment
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
The part of the embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, ordinary skill people
The every other embodiment that member is obtained on the premise of creative work is not made, belongs to the scope of protection of the invention.
The present invention proposes 1 bit robust watermarking Copyright protection scheme in the resource constrained environment based on Patchwork.
The present invention simplifies to application scenarios, and first, collecting device end is regarded as sensor node gathered data, and embedded watermark, this
Invention only considers numeric type data, and this is also the main flow of such application.Data are transmitted by arbitrary network, and data item is formed
Data flow arrives at terminal, is regarded as aggregation node, is pooled to a number of data item and is verified, and initial data can be with if necessary
It is resumed in aggregation node.The model of this simplification highlights the resource-constrained feature of this kind equipment, does not consider go-between
Transmission then embodies the diversity of Internet of Things network transmission, while only considers that the source (i.e. collection terminal) of data flow and terminal (are converged
Poly- end) it enormously simplify problem.
Present invention assumes that a unlimited data flow S, its data item s from sensor node to aggregation nodeiCome from
The collection each time of sensor node, siIncluding at least perception data diAnd its timestamp ti。
Referring to Fig. 1, it is the robust watermarking embedding grammar flow chart for data flow according to the preferred embodiment of the present invention.
As shown in figure 1, the robust watermarking embedding grammar for data flow that the embodiment provides comprises the following steps：
First, in step S101, data decimation step is performed, according to timestamp tiChosen with Probability p from data flow
For carrying the data item of watermark.The step includes obtaining the data item s for forming data flow Si, data item siIncluding at least sense
Primary data and timestamp ti, wherein i is sequence number of the data item in data flow S, and selection meets the data item of below equation
For the data item for carrying watermark：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability, and judging if above formula is met need to be to data item siCarry out
Watermark is embedded in.Wherein Probability p determines the intensity of scheme watermark addition, means data item all in data flow if p=1
Watermark should all be added.
Discriminant function F can use arbitrary form, only need to realize data item siDiscriminant function is met with Probability p.Example
Such as pass through K1With tiAs certain randomizer of state modulator, generation is uniformly distributed in the random number f between 0~1, if f<p
Then discriminant function F value is 1.
Above-mentioned steps S101 shows in collection terminal, when sensor node is often completed once to sample, i.e., produces one in data flow
Individual new data item siWhen, then need to judge data item s according to conditions aboveiWhether it is watermark carrier.
Then, in step s 102, packet step is performed, using packet key K2To the data for carrying watermark
Timestamp carries out hash function calculating in, and the data item is included in into A groups or B groups according to obtained cryptographic Hash.
Preferably, the packet step specifically includes：
(1) cryptographic Hash is calculated by below equation：
ci=H (ti,K2)；
Wherein H is hash function, K2For packet key, ciHashed value for output is cryptographic Hash, is a Bit String；ti
For the timestamp for being used to carry the data item of watermark selected in step S101；
(2) by hashed value ciEach bit carry out XOR obtain biIf bi=1, then by data item siA groups are included in,
Otherwise B groups are included in.
Finally, in step s 103, numerical value increase and decrease step is performed, increase water is carried out to the perception data in A group data item
Modification amount δ computing is printed, to the perception data in B group data item reduce watermark modification amount δ computing, to realize watermark
It is embedded.
Preferably, numerical value increase and decrease step specifically includes：
(1) by below equation to perception data diCarry out the perception data d ' that numerical value increase and decrease computing obtains being embedded in watermarki：
d′i=di+ δ, if bi=1
d′i=di- δ, if bi=0；
(2) the data item s for having been inserted into watermark is senti', including the perception data d ' of embedded watermarkiIt is and unmodified
Timestamp ti。
Referring to Fig. 2, it is the robust watermarking detection method flow chart for data flow according to the preferred embodiment of the present invention.
As shown in Fig. 2 the robust watermarking detection method for data flow that the embodiment provides comprises the following steps：
First, in step s 201, data determination is performed, obtains the data item for forming data flow, the data item
Including at least perception data and timestamp ti, wherein i is the sequence number of the data item in a stream, it is determined that meeting below equation
Data item be data item for carrying watermark：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability；Discriminant function F and selection key K1With Probability p with
The function that robust watermarking uses when being embedded in is consistent with parameter.Step S201 shows in pool side, when aggregation node often receives one
Individual data item siWhen, also according to timestamp tiWith selecting key K1, i.e. above-mentioned formula judges whether it is watermark carrier.
Then, in step S202, data buffer storage step is performed, using packet key K2To the data for carrying watermark
Timestamp carries out hash function calculating in, and data item is divided into A groups and B groups according to obtained cryptographic Hash, and according to affiliated point
Data item is cached in corresponding two queues by group.
Preferably, the data buffer storage step specifically includes：
(1) cryptographic Hash is calculated by below equation：
ci=H (ti,K2)；
Wherein H is hash function, K2For packet key, ciFor the hashed value of output, tiIt is used for for what is selected in step S101
Carry the timestamp of the data item of watermark.Packet key K2The packet key used when being embedded in robust watermarking is consistent.
(2) by hashed value ciEach bit carry out XOR obtain biIf bi=1, then by data item siA groups are included in,
Otherwise B groups are included in.It is grouped according to affiliated by data item siIt is cached in queue QaOr QbIn.QaWith QbTwo established for aggregation node
Queue, it is respectively used to cache the data of A groups and B groups, is cached in QaWith QbIn data item be denoted as sa respectivelyjWith sbj。
Then, in step S203, difference detecting step is performed, calculates the difference of the perception data of two groups of data of A groups and B groups
Accumulated value, and when the data item number of two groups of data is more than predetermined number, calculate the mean difference of two groups of data, it is flat at this
Equal difference judges that watermark is present when being more than predetermined threshold value, otherwise it is assumed that watermark is not present.
Preferably, difference detecting step includes：
(1) accumulated value of the difference of the perception data of two groups of data of A groups and B groups is calculated by below equation：
D=D+daj-dbj；
Wherein dajThe data item sa for forming a team to arrange for AjPerception data, dbjFormed a team column data item sb for BjPerception data,
J is data to sequence.As long as that is, queue QaWith QbNeither for sky, then both sides all dequeued data items and calculate its perceive number
According to difference, add up this difference be denoted as D.
(2) number of two set of queue dequeued data item, when the quantity of difference reaches N, i.e. queue Q are countedaWith QbAll fall out
During N number of data item, the mean difference α of two groups of data is calculated by below equation：
Wherein N size is schema definition, if hope obtains preferable watermark detection effect, N value should be larger.If
N values are close just infinite, then
(3) it is predetermined threshold value to judge whether α ＞ T, wherein T, and T=2 δ θ, 0 ＜ θ ＜ 1, δ are watermark modification amount；Such as
Fruit is then to judge that watermark is present, otherwise it is assumed that watermark is not present.Because the mean difference α of the data flow of not embedded watermark
Close to 0.
In preferred embodiment of the invention, it is above-mentioned that the robust watermarking detection method for data flow is additionally included in completion
After the difference accumulation calculating of (1) step, initial data can be recovered by increasing and decreasing the opposite inverse operations of step S103 with numerical value.
Referring to Fig. 3, the module for the robust watermarking flush mounting for data flow according to the preferred embodiment of the present invention
Schematic diagram.As shown in figure 3, the robust watermarking flush mounting 100 for data flow that the embodiment provides includes：Data decimation mould
Block 110, packet module 120 and numerical value swap modules 130.Preferably, the robust watermarking flush mounting for being used for data flow
100 are located at the source of data flow, i.e. collection terminal or sensor node.
Wherein data decimation module 110 is used to obtain the data item s for forming data flow Si, data item siIncluding at least sense
Primary data and timestamp ti, wherein i is sequence number of the data item in data flow S, and selection meets the data item of below equation
For the data item for carrying watermark：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability.Discriminant function F can use arbitrary form, only need reality
Existing data item meets discriminant function with Probability p.Such as pass through K1With tiIt is raw as certain randomizer of state modulator
Into the random number f being uniformly distributed between 0~1, if f<P then discriminant function F value be 1.
Packet module 120 is connected with data decimation module 110, for using packet key K2To for carrying watermark
Data item in timestamp carry out hash function calculating, the data item is included in by A groups or B groups according to obtained cryptographic Hash.
Preferably, the packet module 120 is used to calculate cryptographic Hash by below equation：
ci=H (ti,K2)；
Wherein H is hash function, K2For packet key, ciFor the hashed value of output, tiIt is used for for what is selected in step S101
Carry the timestamp of the data item of watermark；
The packet module 120 is also by hashed value ciEach bit carry out XOR obtain biIf bi=1, then should
Data item siA groups are included in, are otherwise included in B groups.
Numerical value swap modules 130 are connected with packet module 120, for being carried out to the perception data in A group data item
Increase watermark modification amount δ computing, to the perception data in B group data item reduce watermark modification amount δ computing, to realize
The insertion of watermark.
Preferably, the numerical value swap modules 130 by below equation to perception data diNumerical value increase and decrease computing is carried out to obtain
The perception data d ' of embedded watermarki：
d′i=di+ δ, if bi=1
d′i=di- δ, if bi=0；
Then, numerical value swap modules 130 send the data item s for having been inserted into watermarki', including the perception of embedded watermark
Data d 'iAnd unmodified timestamp ti。
Referring to Fig. 4, the module for the robust watermarking detection means for data flow according to the preferred embodiment of the present invention
Schematic diagram.As shown in figure 4, the robust watermarking flush mounting 200 for data flow that the embodiment provides includes：Data determine mould
Block 210, data cache module 220 and difference detection module 230.Preferably, the robust watermarking detection means for being used for data flow
200 terminals for being located at data are pool side or aggregation node.
Data determining module 210 is used to obtain the data item for forming data flow, and the data item comprises at least perception data
And timestamp ti, wherein i is the sequence number of the data item in a stream, it is determined that the data item for meeting below equation is for holding
Carry the data item of watermark：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability.Discriminant function F and selection key K1With Probability p with
The function that robust watermarking uses when being embedded in is consistent with parameter.
Data cache module 220 is connected with data determining module 210, for using packet key K2To for carrying watermark
Data item in timestamp carry out hash function calculating, data item is divided into by A groups and B groups according to obtained cryptographic Hash, and according to
Data item is cached in corresponding two queues by affiliated packet.
Specifically, the data cache module 220 calculates cryptographic Hash by below equation：
ci=H (ti,K2)；
Wherein H is hash function, K2For packet key, ciFor the hashed value of output, tiIt is used for for what is selected in step S101
Carry the timestamp of the data item of watermark.Packet key K2The packet key used when being embedded in robust watermarking is consistent.
Then, the data cache module 220 is by hashed value ciEach bit carry out XOR obtain biIf bi=1, then
By data item siA groups are included in, are otherwise included in B groups.It is grouped according to affiliated by data item siIt is cached in queue QaOr QbIn.QaWith Qb
Two queues established for aggregation node, it is respectively used to cache the data of A groups and B groups, is cached in QaWith QbIn data item difference
It is denoted as sajWith sbj。
Difference detection module 230 is connected with data cache module 220, for calculating the perception number of two groups of data of A groups and B groups
According to the accumulated value of its difference, and when the data item number of two groups of data is more than predetermined number, the mean difference of two groups of data is calculated,
Judge that watermark is present when the mean difference is more than predetermined threshold value, otherwise it is assumed that watermark is not present.
Referring to Fig. 5, it is difference in the robust watermarking detection means for data flow according to the preferred embodiment of the present invention
The schematic diagram of detection module.As shown in figure 5, the difference detection module 230 includes：Difference summing elements 231, difference averaging unit
232 and threshold decision unit 233.
Wherein, difference summing elements 231 be used for by below equation calculating two groups of data of A groups and B groups perception data it
The accumulated value of difference：
D=D+daj-dbj；
Wherein dajThe data item sa for forming a team to arrange for AjPerception data, dbjFormed a team column data item sb for BjPerception data,
J is data to sequence.
Difference averaging unit 232 is connected with difference summing elements 231, for counting of two set of queue dequeued data item
Number, when the quantity of difference reaches N, the mean difference α of two groups of data is calculated by below equation：
Threshold decision unit 233 is connected with difference averaging unit 232, for judging whether that α ＞ T, wherein T are default threshold
Value, and T=2 δ θ, 0 ＜ θ ＜ 1, δ are watermark modification amount；If it is judge that watermark is present, otherwise it is assumed that watermark is not deposited
.
Embedded to the robust watermarking for data flow of the present invention below and effect of detection method is analyzed and verified.
(1) validity
The core concept of simple 1 bit robust watermarking scheme proposed by the present invention is derived from digital picture
Patchwork algorithms.Probability p determines the intensity of scheme watermark addition, means data all in data flow if p=1
Xiang Jieying adds watermark.And the cryptographic Hash according to caused by data item timestamp with packet key is grouped, may be regarded as one it is random
The process of selection.Two groups of data item of A, the B randomly selected in a stream, its perception data is clearly independent identically distributed, and two
Group perception data difference is desired for 0 so that watermark is embedded in successfully.
The present invention selectes the data item of carrying watermark according to timestamp, and data item is divided also according to timestamp
Group, and timestamp is not modified during the insertion and detection of watermark, this guarantees data flow both ends to obtain
Identical packet, it is ensured that the successful detection of watermark.Certainly the present invention is it cannot be guaranteed that the number of the data item in two groups of A, B
It is definitely equal, but randomly choose and make it that both quantity is substantially suitable, as long as the quantity N of statistics is larger, the detection of watermark
It is unaffected.
Although the watermark capacity of this programme is only 1 bit, i.e., can only distinguishes data stream whether have embedded watermark.Although watermark
Capacity very little, but for being feasible by main purpose robust watermarking of copyright protection, because the present invention can be effectively
Detect whether contain copyright side's information in flow data.
(2) robustness
Robustness is the most important performance indications of robust watermarking scheme, and the present invention in background technology by for enumerating
Common attack pattern analyzes the robustness of watermark of the present invention.
(1) intercept
If one section in attacker's data intercept stream, the timestamp in data flow is not tampered, therefore scheme is still
The data item for choosing carrying watermark can be stabbed with passage time, is grouped according to timestamp and calculates the difference of two groups of perception datas
Average.Obviously, the watermark detection process for intercepting the simple robust watermarking scheme proposed to this section has little to no effect.Certainly,
If the data flow of interception is too short, the lazy weight of difference statistics, the statistical result for also resulting in average is wrong and can not detect
The presence of watermark.But this too short data flow has no too big actual use value, therefore interception attack can hardly come into force.
(2) sample
By sampling, attacker can obtain some discontinuous independent data item, and combine and to form new data flow.
Similarly, the not original timestamp of altered data item is sampled also.And select data item carry out watermark insertion be by selection key K1
Protection, attacker can not judge which data item carries data item for watermark, and sampling can only be carried out at random.The number of stochastical sampling
Necessarily uniform according to the data item distribution in stream, belonging to two groups of A, B, quantity is also just substantially suitable.Therefore, feelings are attacked with interception
Condition is similar, as long as enough watermark carrying data item can be obtained, just can therefrom detect the presence of watermark.
(3) delete
In a sense, it is a kind of same attack pattern to delete with sampling, only generally deletes attack and retain
Data item it is more, and sample attack delete data item it is more.Except of course that the deletion attack of active, is also present by network rings
Packet loss caused by border, but the deletion of quovis modo, it is all random that data item, which is lost,.Therefore, this is similar with sampling, only
The quantity of the data item for the carrying watermark to be counted is enough, and watermark can be just detected.
(4) inject
The data item that injection is forged not is the attack meanses of a common robust watermarking being directed to, because data falsification
The injection rate of item can not be excessive, and excessive injecting data item will necessarily influence the use value of initial data.Assuming that data falsification
The injection rate of item is pI, and pI＞ 0, it means that just need to inject xp per x data itemIIndividual data item, and total data item
Number is x (1+pI), wherein the data item for carrying watermark information is xp.Obviously, it is original carry watermark information data item still
So can correctly it be counted, but Hash computings cause the data falsification item that newly injects of part to meet the condition of embedded watermark,
There is xpIP data meet the criterion of embedded watermark.
Packet is still uniform for injecting data item, then in two groups of data of A, B, carries data item and the forgery of watermark
The ratio of data item is 1:pI, then when N values are larger, mean difference is
Obviously, as long as meeting α ＞ T, i.e.,
Watermark remains able to be detected, and on the contrary then watermark detection fails.But above analyze, excessive injecting data item
The use value of initial data can be influenceed, therefore, one can consider that injection attacks influence less on watermarking project.
(5) change
First, this programme can not resist the modification to timestamp, because either the selection of watermark carrying data item is still
Two groups of division is all determined by timestamp.Secondly, the modification for perception data will not affect that statistical result.
If the modified values of perception data are fixed, that is, the perception data of each data item adds a fixation
Value, the average of this time difference value remain as
Wherein, da 'jWith db 'jFor the perception data of two groups of data item after modification, dt is fixed modified values.
If modified values are random, i.e., the modified values of each data item are different, are denoted as dti, now
Write following form
dtiIt is random modified values, therefrom randomly chooses two groups of necessarily independent identically distributed, two groups of difference expectations
It is necessarily just 0.Therefore, when N is sufficiently large
Obviously, the modification on perception data neither influences the detection of watermark.
Generally speaking, as long as timestamp is not tampered with, robust watermarking scheme proposed by the present invention can be subjected to common attack
Hit, realize the detection of watermark.
(3) expense
In terms of computing cost, most significant one is Hash computings, and this has been to realize the minimum of safety compared with encryption
Expense.In terms of storage overhead, sensor node is not except when preceding data item caches other data, and aggregation node then needs
Establish the data item that two queues cache two groups.But because as long as two queues are neither empty with regard to dequeued data calculating difference, and
The packet of data item is random, so while aggregation node is bigger than the storage overhead of sensor node, but it is most of when
And too many data item need not be cached.The present invention can also be to aggregation node queue set a upper limit, if caching
Data reach this upper limit, and next data item that should be cached just directly recovers the statistics that initial data does not participate in watermark.This
The detection of watermark is not interfered with equally.In terms of communication overhead, this programme is based on watermark, does not increase extra amount of communication data.
Generally speaking, the expense of this programme is fully acceptable.
(4) service quality
It was found from the flow of the robust watermarking embedding grammar of the present invention, sensor node does not cache any data item, therefore
The time delay of transmitting terminal all is from time delay caused by data calculating, and this is typically negligible.In pool side,
Aggregation node is needed by the data cached item of queue, but has been analyzed at the expense of memory space, and buffer memory is generally little, and
And the present invention limits the quantity of caching using the upper limit of queue length, caused time delay compared with existing scheme, and
Acceptable.Also, in resource-constrained application scenarios, present invention, avoiding the time delay of sensor node, ensures
The smoothness of data acquisition transmission and data flow it is smooth, and the limitation of generally pool side is generally few.
This programme needs to modify to gathered data, and watermark modification amount δ determines the transparency of watermark, and obvious δ
The effect of the bigger watermark of value is better.First, the data after the embedded watermark of the present invention can be recovered, therefore δ has no effect on
The use value of watermark.Certainly, if the data after embedded watermark are too prominent, attacker still can be caused in route of transmission
Attention, so it is higher to remain desirable to watermark transparency.Fig. 6 shows the one section of temperature data stream gathered by sensor,
Fig. 7 then shows the temperature data stream after watermark insertion.For temperature data stream, δ value is taken as 0.01 degree, and Fig. 7 watermark
Scheme watermark modified values value is 2 δ, and as seen from the figure, embedded watermark does not produce excessive influence, transparency to the value of data flow
It is still acceptable.
(5) simulated experiment
(1) Setup Experiments
Simulated experiment is carried out by Matlab, and original data stream used in experiment is gathered from IntelBerkeley research experiments
The sensor that room is disposed, periodically gather the humidity with timestamp, temperature, intensity of illumination and voltage data.
The data flow tested every time only retains one kind in four kinds of sampled datas, and 100,000 data should be included in data flow
, the experimental result of each project both is from the repetition of 1,000 times, and tests all use different data flows every time.
For the detection of watermark, most important two indices are loss and false alarm rate.For robust watermarking, leakage
Inspection, refers to embedded in watermark and aggregation node fails to detect；And false-alarm, then refer to not being embedded in watermark and converging section
Point but detects the presence of watermark.It would therefore be desirable to first loss of the testing scheme under the premise of without any attack and
False alarm rate, and experimental program performance and the relation of several important parameters.
Test to robustness is then carried out according to the foregoing analysis to robustness.Attack for robust watermarking is not
Influence to make watermark undetectable as far as possible on the premise of initial data use, therefore the test to watermark robustness will focus on
The verification and measurement ratio of watermark under different attack patterns.The present invention is firstly used in without the relatively good one group of ginseng of performance in attack laboratory
Number carrys out allocation plan, then tests one by one under various attack patterns, the situation of change of watermark detection rate.In addition, the present invention should also survey
Examination is with statistical magnitude N increase, and change of the watermark detection rate under different attack patterns, this aids in determining whether to increase statistical number
Whether amount can improve the verification and measurement ratio of watermark.
For the robust watermarking scheme that watermark capacity is 1 bit, each experimental result only has two kinds, that is, detect watermark or
Person can't detect watermark.Therefore the either repetition of loss, false alarm rate or verification and measurement ratio all by thousands of experiments could obtain
Go out.
(2) tested without attack
For watermarking project performance may influential parameter mainly have addition watermark Probability p, watermark modification amount δ,
Threshold coefficient θ and difference statistical magnitude N.Wherein watermark modification amount δ is a relatively small value, such as temperature number
According to δ value is taken as 0.01 degree, if 25 degree of room temperature average out to, δ value is only the 0.04% of initial data, is one to water
Printthrough lightness influences little numerical value.Therefore, the present invention represents the change of this parameter with watermark modification amount δ multiple.
Fig. 8 shows the trend that loss and false alarm rate change with statistical magnitude N, and other three parameters are arranged to watermark
Embedded Probability p=0.5, watermark modification measure 2 δ, and threshold coefficient θ=0.8, i.e. threshold value are T=1.6 δ.From figure 8, it is seen that
N has a great influence to loss.If N values are smaller, there is the presence that larger Probability Detection does not go out watermark, but as long as N-dimensional is held one
Individual larger value, such as more than 2000, hardly occur missing inspection.This is identical with analysis before, and the statistic property of scheme is exactly
Establish on the statistical basis to mass data.Although false alarm rate reduces also with N increase, false alarm rate change
Amplitude is smaller, and is held at less than 1%, therefore statistical magnitude N influences less on false alarm rate.
Fig. 9 shows influence of the Probability p to loss and false alarm rate for watermark insertion, and other three parameters are arranged to
Statistical magnitude N=1000, watermark modification measure 2 δ, threshold coefficient θ=0.8.It was found from experimental result, Probability p is also to loss
There is large effect, loss reduces with p increase.Reason is then still with statistical correlation, and p is more big, and then watermark carrier exists
Selection in data flow is more intensive, and the distance between data item as carrier is nearer.And flow data gradually changes,
The difference of perception data is just smaller between adjacent nearer data item, and statistics effect is better.And if p is too small, then carrier number
More remote according to the distance between item in a stream, statistic property will reduce.Although false alarm rate also presents identical with loss
Trend, but false alarm rate very little, therefore it is considered that influence of the Probability p to false alarm rate of watermark insertion is little.
Figure 10 shows the trend that loss changes with false alarm rate with watermark modification amount, and other three parameters are arranged to water
Print embedded Probability p=0.5, statistical magnitude N=1000, threshold coefficient θ=0.8, the change of watermark modification amount then by δ until
5 δ, it is all δ integral multiple.Experimental result shows that the increase of watermark modification amount can cause loss to be reduced with false alarm rate, but needs
It is noted that the modified values increase of data, the transparency of watermark just reduce, therefore this modified values can only be maintained reasonable at one
Scope, it is impossible to blindly increase.
Figure 11 then shows threshold coefficient θ influence, Probability p=0.5 for being arranged to watermark insertion of other three parameters,
Statistical magnitude N=1000, watermark modification measure 2 δ.Obviously, threshold coefficient θ needs to do a balance.Threshold coefficient θ increases, then
Threshold value increases, then the loss rise of watermark, but false alarm rate can decline therewith, it is believed that threshold existing for detection watermark improves
.But from absolute figure, false alarm rate always maintains less scope, therefore, we can slightly reduce threshold value, improve water
The verification and measurement ratio of print, makes loss further reduce.
The result of four groups of experiments more than, first, influence of all parameters for scheme false alarm rate will be far smaller than
Influence to loss, therefore should be preferentially to reduce loss as target in allocation plan.And for robust watermarking, dimension
The detectability of water holding print is only primary goal.Secondly, watermark modification amount δ and threshold coefficient θ is required for the value of a balance,
Maintain the transparency of relatively low false alarm rate and watermark.Finally, watermark insertion Probability p and watermark statistical magnitude N all directly with
Statistic property is related, especially for statistical magnitude N, as long as the statistical magnitude of increase data item, for most of rational ginseng
Robust watermarking scheme under number configuration can realize the detection that loss is less than 1%.This characteristic be very suitable for quantity compared with
The data flow of big even endless sensor collection, as long as data constantly gather, embedded watermark can be just detected.
(3) attack experiment
To common robust watermarking attack pattern, before more detailed analysis has been made.Attacked for interception, I
All experiments be built upon finite number according to the statistical result of item on, therefore to parameter watermark insertion Probability p and
Experiment of the statistical magnitude N experiment namely to interception attack, as a result shows, as long as statistical magnitude reaches certain degree, cuts
Attack is taken not constitute a threat to.
For sampling attack and deleting attack, a kind of same attack pattern can be regarded as, certainly, sampling attack is deleted
Data volume it is bigger.Figure 12 shown under different sample rate attacks, and under different statistical magnitudes, the verification and measurement ratio of watermark
Variation tendency.(i.e. each two data item retains one) is reduced to 1/5 to sample rate by 1/2, statistical magnitude N by 500 always on
5000 are raised to, other specification is the configuration of the better performances selected in being tested without attack.Experimental result shows that sampling attack is true
The verification and measurement ratio of watermark is reduced in fact, it is more obvious especially when statistical magnitude is relatively low.Once however, statistical magnitude increase, scheme
Verification and measurement ratio will rapidly increase, and be finally reached a higher level.
Injection attacks are not the common attack form of robust watermarking, because it is original to inject new data item meeting significant impact
The use value of data.Figure 13 shows the verification and measurement ratio of watermark under different injection rates with the trend that statistical magnitude changes.Note
Enter rate and rise to 40% by 10%, it is meant that every 10 data item will inject the data item of 4 forgeries, the data item of forgery
Perception data value data item adjacent thereto is close.Experimental result shows that, when injection rate is relatively low, the influence to watermark detection rate is not
Greatly.But injection rate, when rising to 30%, the verification and measurement ratio of watermark drastically declines, certainly with the increase watermark detection of statistical magnitude
Rate improves, and still shows detectable trend.This obviously meets the analysis of previous section, and injection rate 30% still meets formula：
Then watermark is still detectable.But when injection rate rises to 40%, no longer meet above-mentioned formula, with statistical magnitude
Increase, watermark detection rate rapidly have dropped on the contrary, and watermark just becomes detectable.
If wishing to maintain the availability of data certainly, the injection rate of data cannot be too high.Therefore, scheme is attacked for injection
The robustness hit is also acceptable.
Theory analysis shows that scheme has stronger robustness to data modification, particularly with fixed value modification hardly
The detection of watermark is influenceed, experiment also demonstrates that this analysis, is not repeated herein.Figure 14 shows modification of the scheme for random value
The robustness of attack.Modification rate represents the number for the data item changed in data flow, and modified values change at random between ± 4 δ,
Ensure not interfere with the use value of initial data.Experimental result is shown, when modification rate is higher, i.e., most of number in data flow
During according to all being changed at random, the verification and measurement ratio of watermark is barely affected.Because the difference of random modified values is desired for 0, because
As long as the quantity of this statistics is sufficiently large, this has little influence on the detection of watermark.Although relatively low modification rate significantly reduces on the contrary
The verification and measurement ratio of watermark, but with statistical magnitude N increase, the verification and measurement ratio of watermark remains able to rise to a rational level.
In a word, the attack form for most robust watermarking, this present invention can maintain higher watermark detection rate.
Part attack can cause the decline of watermark detection rate with the rising of attack frequency, but once increase the quantity of statistics, watermark
Verification and measurement ratio will rise, and be finally reached a higher level.This is the scheme advantage based on statistics, as long as the sample of statistics
This is enough, can detect watermark.
In summary, the present invention selects some data item with Probability p in data source, i.e. sensor node in data flow S
As the carrier of watermark, selected data item is according to its timestamp tiTwo groups are divided into key, are denoted as A and B.In A groups
Each data item its perception data increases a less numerical value δ, and the perception data of each data item is reduced identical in B groups
Numerical value δ.Aggregation node chooses carrier data item, subdivided A, B group in the same way, and cumulative two groups of data item perceive number
According to difference.If it is necessary, data item can after the completion of calculating difference recovers raw sensed data.When the data item of statistics
When reaching certain quantity, you can judge that watermark whether there is by the average of difference.
It should be appreciated that the robust watermarking embedding grammar and method of detecting watermarks and dress for data flow of the invention
The realization principle put is identical with process, therefore to the robust watermarking embedding grammar and the tool of detection method embodiment for data flow
Body description is also applied for robust watermarking flush mounting and watermark detecting apparatus for data flow.
Finally it should be noted that：The above embodiments are merely illustrative of the technical solutions of the present invention, rather than its limitations；Although
The present invention is described in detail with reference to the foregoing embodiments, it will be understood by those within the art that：It still may be used
To be modified to the technical scheme described in foregoing embodiments, or equivalent substitution is carried out to which part technical characteristic；
And these modification or replace, do not make appropriate technical solution essence depart from various embodiments of the present invention technical scheme spirit and
Scope.
Claims (8)
1. a kind of robust watermarking detection method for data flow, it is characterised in that comprise the following steps：
Data determination, obtains the data item for forming data flow, and the data item comprises at least perception data and timestamp
ti, wherein i is the sequence number of the data item in a stream, it is determined that the data item for meeting below equation is the number for carrying watermark
According to item：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability；
Data buffer storage step, using packet key K2Hash function calculating is carried out to timestamp in the data item for carrying watermark,
Data item is divided into by A groups and B groups according to obtained cryptographic Hash, and two teams corresponding to according to affiliated packet data item is cached in
In row；
Difference detecting step, the accumulated value of the difference of the perception data of two groups of data of A groups and B groups is calculated, and in the number of two groups of data
When being more than predetermined number according to item number, the mean difference of two groups of data is calculated, is judged when the mean difference is more than predetermined threshold value
Watermark is present, otherwise it is assumed that watermark is not present.
2. the robust watermarking detection method according to claim 1 for data flow, it is characterised in that the data buffer storage
Step includes：
(1) cryptographic Hash is calculated by below equation：
ci=H (ti,K2)；
Wherein H is hash function, K2For packet key, ciFor the hashed value of output；
(2) by hashed value ciEach bit carry out XOR obtain biIf bi=1, then by data item siA groups are included in, otherwise
It is included in B groups.
3. the robust watermarking detection method according to claim 1 or 2 for data flow, it is characterised in that the difference
Detecting step includes：
(1) accumulated value of the difference of the perception data of two groups of data of A groups and B groups is calculated by below equation：
D=D+daj-dbj；
Wherein dajThe data item sa for forming a team to arrange for AjPerception data, dbjFormed a team column data item sb for BjPerception data, j for number
According to sequence；
(2) number of two set of queue dequeued data item is counted, when the quantity of difference reaches N, two groups are calculated by below equation
The mean difference α of data：
<mrow>
<mi>&alpha;</mi>
<mo>=</mo>
<mfrac>
<mi>D</mi>
<mi>N</mi>
</mfrac>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>N</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mrow>
<mo>(</mo>
<msub>
<mi>da</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<msub>
<mi>db</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
(3) it is predetermined threshold value to judge whether α ＞ T, wherein T, and T=2 δ θ, 0 ＜ θ ＜ 1, δ are watermark modification amount；If
Then judge that watermark is present, otherwise it is assumed that watermark is not present.
4. the robust watermarking detection method according to claim 3 for data flow, it is characterised in that methods described is also wrapped
Include after the difference accumulation calculating of (1) step in completing difference detecting step, numerical value increase and decrease operation phase during by being embedded in watermark
Anti- inverse operation recovers initial data.
A kind of 5. robust watermarking detection means for data flow, it is characterised in that including：
Data determining module, the data item of data flow is formed for obtaining, and the data item comprises at least perception data with timely
Between stab ti, wherein i is the sequence number of the data item in a stream, it is determined that the data item for meeting below equation is for carrying watermark
Data item：
F(p,ti,K1)==1；
Wherein F is discriminant function, K1To select key, p is probability；
Data cache module, for using packet key K2Hash function is carried out to timestamp in the data item for carrying watermark
Calculate, data item be divided into by A groups and B groups according to obtained cryptographic Hash, and according to it is affiliated be grouped data item is cached in corresponding to
In two queues；
Difference detection module, the accumulated value of the difference of the perception data for calculating two groups of data of A groups and B groups, and in two groups of data
Data item number when being more than predetermined number, the mean difference of two groups of data is calculated, when the mean difference is more than predetermined threshold value
Judge that watermark is present, otherwise it is assumed that watermark is not present.
6. the robust watermarking detection means according to claim 5 for data flow, it is characterised in that the data buffer storage
Module calculates cryptographic Hash by below equation：
ci=H (ti,K2)；
Wherein H is hash function, K2For packet key, ciFor the hashed value of output；
And the data cache module is by hashed value ciEach bit carry out XOR obtain biIf bi=1, then this is counted
According to item siA groups are included in, are otherwise included in B groups.
7. the robust watermarking detection means for data flow according to claim 5 or 6, it is characterised in that the difference
Detection module includes：
Difference summing elements, the accumulated value of the difference of the perception data for calculating two groups of data of A groups and B groups by below equation：
D=D+daj-dbj；
Wherein dajThe data item sa for forming a team to arrange for AjPerception data, dbjFormed a team column data item sb for BjPerception data, j for number
According to sequence；
Difference averaging unit, the number of two set of queue dequeued data item is counted, when the quantity of difference reaches N, passes through following public affairs
Formula calculates the mean difference α of two groups of data：
<mrow>
<mi>&alpha;</mi>
<mo>=</mo>
<mfrac>
<mi>D</mi>
<mi>N</mi>
</mfrac>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mi>N</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>N</mi>
</munderover>
<mrow>
<mo>(</mo>
<msub>
<mi>da</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<msub>
<mi>db</mi>
<mi>j</mi>
</msub>
<mo>)</mo>
</mrow>
<mo>;</mo>
</mrow>
Threshold decision unit, for judging whether that α ＞ T, wherein T are predetermined threshold value, and T=2 δ θ, 0 ＜ θ ＜ 1, δ is water
Print modification amount；If it is judge that watermark is present, otherwise it is assumed that watermark is not present.
8. the robust watermarking detection means according to claim 7 for data flow, it is characterised in that the variance yields tires out
Add unit also after difference accumulation calculating is completed, during by being embedded in watermark numerical value increase and decrease operate opposite inverse operation and recover original
Data.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710600765.2A CN107369125A (en) | 2017-06-13 | 2017-06-13 | Robust watermarking detection method and device for data flow |
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710600765.2A CN107369125A (en) | 2017-06-13 | 2017-06-13 | Robust watermarking detection method and device for data flow |
| CN201710441578.4A CN107274331B (en) | 2017-06-13 | 2017-06-13 | Robust watermarking embedding grammar and device for data flow |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710441578.4A Division CN107274331B (en) | 2017-06-13 | 2017-06-13 | Robust watermarking embedding grammar and device for data flow |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107369125A true CN107369125A (en) | 2017-11-21 |
Family
ID=60067510
Family Applications (2)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710441578.4A Active CN107274331B (en) | 2017-06-13 | 2017-06-13 | Robust watermarking embedding grammar and device for data flow |
| CN201710600765.2A Pending CN107369125A (en) | 2017-06-13 | 2017-06-13 | Robust watermarking detection method and device for data flow |
Family Applications Before (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710441578.4A Active CN107274331B (en) | 2017-06-13 | 2017-06-13 | Robust watermarking embedding grammar and device for data flow |
Country Status (1)
| Country | Link |
|---|---|
| CN (2) | CN107274331B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109005175A (en) * | 2018-08-07 | 2018-12-14 | 腾讯科技（深圳）有限公司 | Network protection method, apparatus, server and storage medium |
| CN113626802A (en) * | 2021-08-23 | 2021-11-09 | 重庆第二师范学院 | Login verification system and method for equipment password |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108494775B (en) * | 2018-03-26 | 2020-12-15 | 四川长虹电器股份有限公司 | Method for preventing network attack by using legal data or tampering legal data |
| CN111651735B (en) * | 2020-05-07 | 2022-03-01 | 清华四川能源互联网研究院 | Time sequence data watermarking method with timestamp alignment function |
| CN113095988B (en) * | 2021-03-29 | 2022-02-01 | 贵州大学 | Dispersion tensor image robust zero watermarking method based on ORC sampling and QGPCE conversion |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102646259A (en) * | 2012-02-16 | 2012-08-22 | 南京邮电大学 | Anti-attack robustness multiple zero watermark method |
| CN102867286A (en) * | 2011-07-08 | 2013-01-09 | 索尼公司 | Method and device for generating, embedding and detecting digital watermarks |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3778009B2 (en) * | 2001-06-13 | 2006-05-24 | ソニー株式会社 | Data transfer system, data transfer device, data recording device, and data management method |
| GB0317571D0 (en) * | 2003-07-26 | 2003-08-27 | Koninkl Philips Electronics Nv | Content identification for broadcast media |
| EP2797333A1 (en) * | 2013-04-26 | 2014-10-29 | Nagravision S.A. | Method for watermarking media content and system for implementing this method |
| CN105263024B (en) * | 2015-10-15 | 2018-06-29 | 宁波大学 | A kind of registration of HEVC video flowing zero watermarkings of anti-quantization transcoding and detection method |
-
2017
- 2017-06-13 CN CN201710441578.4A patent/CN107274331B/en active Active
- 2017-06-13 CN CN201710600765.2A patent/CN107369125A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102867286A (en) * | 2011-07-08 | 2013-01-09 | 索尼公司 | Method and device for generating, embedding and detecting digital watermarks |
| CN102646259A (en) * | 2012-02-16 | 2012-08-22 | 南京邮电大学 | Anti-attack robustness multiple zero watermark method |
Non-Patent Citations (2)
| Title |
|---|
| XI SHI ETC.: "A reversible watermarking authentication scheme for wireless sensor networks", 《INFORMATION SCIENCES》 * |
| 石熙: "数字水印技术在无线传感器网络安全中的应用研究", 《中国博士学位论文全文数据库 信息科技辑》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109005175A (en) * | 2018-08-07 | 2018-12-14 | 腾讯科技（深圳）有限公司 | Network protection method, apparatus, server and storage medium |
| CN113626802A (en) * | 2021-08-23 | 2021-11-09 | 重庆第二师范学院 | Login verification system and method for equipment password |
| CN113626802B (en) * | 2021-08-23 | 2023-05-12 | 重庆第二师范学院 | Login verification system and method for equipment password |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107274331B (en) | 2018-03-13 |
| CN107274331A (en) | 2017-10-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107274331B (en) | Robust watermarking embedding grammar and device for data flow | |
| Gianvecchio et al. | Model-based covert timing channels: Automated modeling and evasion | |
| Sultana et al. | Secure provenance transmission for streaming data | |
| Zhang et al. | Secure data aggregation in wireless sensor networks: a watermark based authentication supportive approach | |
| CN111818103B (en) | Traffic-based tracing attack path method in network target range | |
| Walls et al. | Liquid: A detection-resistant covert timing channel based on IPD shaping | |
| Zhao et al. | Detecting covert channels in computer networks based on chaos theory | |
| CN105407096B (en) | Message data detection method based on flow management | |
| Luo et al. | Robust network covert communications based on TCP and enumerative combinatorics | |
| CN106375157B (en) | A kind of network flow correlating method based on phase space reconfiguration | |
| CN104636764B (en) | A kind of image latent writing analysis method and its device | |
| CN110300127A (en) | A kind of network inbreak detection method based on deep learning, device and equipment | |
| CN107248996A (en) | A kind of detection of DNS amplification attacks and filter method | |
| CN106878314A (en) | Network malicious act detection method based on confidence level | |
| CN107733883A (en) | A kind of method and device for detecting batch registration account | |
| CN110351291A (en) | Ddos attack detection method and device based on multiple dimensioned convolutional neural networks | |
| CN115150182B (en) | Information system network attack detection method based on flow analysis | |
| Xiao et al. | Digital watermark-based independent individual certification scheme in WSNs | |
| Lu et al. | An HTTP flooding detection method based on browser behavior | |
| CN106850658B (en) | The network malicious act detection method of real-time online study | |
| CN109583056A (en) | A kind of network-combination yarn tool performance appraisal procedure and system based on emulation platform | |
| CN109858510A (en) | A kind of detection method for http protocol ETag value covert communications | |
| CN110290110A (en) | A kind of recognition methods of encryption malicious traffic stream and system based on redundancy detection framework | |
| JP2005236862A (en) | Log analyzing device and program, and recording medium | |
| Huang et al. | Mining Web User Behaviors to Detect Application Layer DDoS Attacks. |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20171121 |