CN109740029A - Multiple groups label parallel search method in a kind of extensive RFID system - Google Patents
Multiple groups label parallel search method in a kind of extensive RFID system Download PDFInfo
- Publication number
- CN109740029A CN109740029A CN201811622037.2A CN201811622037A CN109740029A CN 109740029 A CN109740029 A CN 109740029A CN 201811622037 A CN201811622037 A CN 201811622037A CN 109740029 A CN109740029 A CN 109740029A
- Authority
- CN
- China
- Prior art keywords
- label
- group
- time slot
- found
- reader
- 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
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention discloses multiple groups label parallel search methods in a kind of extensive RFID system, comprising the following steps: assuming that there is n set of tags to be searched；Server is that each group of label to be found distributes a group number and hash function collection Hi, for all labels to be found, mapped to obtain a binary bits string L using its corresponding hash function collection；Reader broadcasts L and Hi, cog region label utilizes the H that receivesiMapping obtains the mapping group number of oneself and is grouped on L；Test time slot and non-test time slot are calculated according to candidate target tally set Ci, according to non-targeted label is deactivated in the reply of non-test time slot, the target labels being not present and the existing target labels of confirmation are excluded from Ci according to the tag return situation in test time slot；It repeats the above process, until meeting termination condition.The present invention can a variety of label parallel searches, the grouping information of automatic identification label to be searched quickly checks non-label to be found, executes that the time is short and wrong classification probability is small.
Description
Technical field
The invention belongs to RFID label tag search field, in particular to multiple groups label is searched parallel in a kind of extensive RFID system
Suo Fangfa.
Background technique
Nowadays radio frequency identification (RFID) technology is widely used in many fields, including supply chain monitoring, storehouse management and
Storage controlling.It in certain scenes, does not need that labels all in system are identified or monitored, and only hopes to a certain
The interested specific label in part is monitored to determine them in systems and whether there is.For example, might have stored in system
Label (commodity) from multiple and different tissues (supplier), some supplier merely desire to be monitored simultaneously one's own commodity
It is further processed dependent merchandise.To be different from common monitoring, this operation is referred to as tag search: given one group specific
Label is known as label to be found, the purpose of tag search be these determining labels which at present in systems, which does not exist.
Tag search is a kind of very important operation in RFID system, but only has a few studies at present, and these
Research all only considers once to scan for a group of labels.If encountering following this scene: being stored in a bulk storage plant
The different commodity of multiple commercial vendors, if several manufacturers hold items list requirement and scan for simultaneously, due to being related to business
Competitive relation secret and that may be present, the information of these manufacturers commodity to be searched is reluctant disclosure mutually, so cannot this is several
The items list to be searched of manufacturer, family, which is put together, to be scanned for.Can only just a group of labels be waited to search for according to pervious method
At the search for carrying out lower a group of labels later, but due to there are a large amount of non-labels to be found in reader cog region, search for every time
It requires to exclude a large amount of identical non-labels to be found, greatly wastes the time, the present invention just considers simultaneously to these groups
Label, which synchronizes, scans for obtaining result to improve search efficiency.
In recent years, there are some efficient tag search researchs.For example, defining tag search problem and proposing base
(CATS) method is approached in tag search is compact, and CATS uses Bloom filter to combine closely between reader and label
Information exchange, this avoids the transmission of time-consuming tag ID, to reduce search time.Although it is to be found to execute sub-fraction
Tally set, still, when number of tags to be found increases, the performance of CATS is substantially reduced.When the overlay area in reader has more
More labels to be found, CATS do not work possibly even.Compared to using a single big Bloom filter, iteration mark
Label searching method (ITSP) are proposed to fall non-targeted label short across filter vector filter using a series of.With non-targeted label by by
Filtering out gradually, the size of filter vectors is gradually reduced, and this facilitate the high timeliness of ITSP.ITSP filters out one in each round
Half non-targeted label, in many cases, this is not optimal effectiveness, to reduce its performance.In STEP, read to be multiple
The radio-frequency recognition system for reading device proposes a tag search method.However, the reflecting using label to be found in STEP method
The reply of time slot and cog region label is penetrated to filter out non-targeted label, when single reader identification region number of labels is larger
When, it will be greatly reduced its performance.
All tag search methods proposed at present all do not account for existing simultaneously the case where multiple groups label to be found, according to
Pervious method carries out the search of lower a group of labels after can only just waiting a group of labels search to complete, but due to reader cog region
Middle to there are a large amount of non-labels to be found, search requires to exclude a large amount of identical non-labels to be found every time, greatly wastes
Time.
Problem definition is as depicted in figs. 1 and 2, considers in an extensive RFID system, exists in reader cog region
Tag representation be Y={ y1,y2... }, the total f group of label to be found is expressed as G1={ g11,g12... }, G2={ g21,
g22... } ..., Gf={ gf1,gf2... }, and G1∩G2∩...∩Gf=Θ.All tally sets to be searched are expressed as X={ G1,
G2,...,Gf}.Traditional tag search problem is to find the intersection of a certain group of label to be found and cog region label.And in this hair
The intersection of multiple groups label to be found and cog region is found in bright simultaneously, i.e., finds G simultaneously1∩Y,G2∩Y,...,Gf∩Y.Cause
Intersection for cog region label and label to be found is unknown, so may label and cog region mark to be found under extreme case
Sign no intersection, i.e. X ∩ Y=Θ, it is also possible to which label to be found is all in cog region label, i.e. X ∩ Y=X.With | |
Indicate the radix of collection.For example, as shown in Figure 1, having 3 groups of tag set G to be found1, G2, G3, hypographous mark in each group
Label are represented and are now existed in reader cog region Y, and white is represented there is no in cog region, and different shade labels are right in X in Y
Shade label is answered to represent the different groups of labels for now existing in cog region.
Summary of the invention
It is an object of the present invention in view of the above shortcomings of the prior art, multiple groups in a kind of extensive RFID system are provided
Label parallel search method, can a variety of label parallel searches, the grouping information of automatic identification label to be searched, quickly check it is non-
Label to be found, executes that the time is short and wrong classification probability is small.
In order to solve the above technical problems, the technical scheme adopted by the invention is that:
Multiple groups label parallel search method in a kind of extensive RFID system, its main feature is that the following steps are included:
The first step, it is assumed that have n set of tags to be searched；Server be each group of label to be found distribute a group number and with
The corresponding hash function collection H of the group numberi(1≤i≤n) utilizes its corresponding hash function collection for all labels to be found
It is mapped to obtain a binary bits string L；
Second step, reader broadcast L and n hash function collection Hi(1≤i≤n), cog region label utilize n received
Hash function set HiMapping obtains the mapping group number of oneself on L；When the mapping group number is not 0, then corresponding cog region
Label, which is thought, oneself to be the member of mapping group and keeps active；When the mapping group number is 0, then corresponding cog region label is not
Label to be searched makes its silencing；Operation is repeated until without grouping failure label；
Third step calculates test time slot and non-test time slot according to candidate target tally set Ci, then according to it is non-test when
Non-targeted label is deactivated in the reply of gap, the target labels being not present according to the tag return situation in test time slot from Ci exclusion
And confirm existing target labels；It repeats the above process, until meeting termination condition.
As a preferred method, in the first step, if a shared f group label to be searched, for each group of mark to be found
Label distribute the group number that a length is f bit, which is formed by one 1 and f-1 0, wherein the group number of i-th group of tally set
Setting rule are as follows: i-th of bit is set as 1, and other f-1 bits are set as 0.
Third step treatment process includes: as a preferred method,
Firstly, reader calculates test time slot according to current candidate target labels collection；
Then, the frame length and hash function used when reader broadcast construction test time slot；Cog region label is receiving
After parameter information, oneself corresponding time slot is selected on frame with hash function identical with calculating test time slot, is then sent back
It should be to reader；
If reader receives response in test time slot, reader notifies corresponding label to keep active；
If reader receives response in non-test time slot, reader notifies corresponding label silencing；
Finally, reader updates candidate target tally set and is matched next time.
Binary bits string length as a preferred method,Wherein, M is label to be found
Number, f are that the group number of set of tags to be found is long, and α is specified classification failure rate.
Optimal frame sign is f as a preferred method,l=1.03*Nl；Wherein NlIndicate the active mark in the local of reader
Sign quantity.
Compared with prior art, the present invention can a variety of label parallel searches, automatic identification label to be searched grouping letter
Breath quickly checks non-label to be found, executes that the time is short and wrong classification probability is small.
Detailed description of the invention
Fig. 1 is problem model figure.
Fig. 2 is the expression title in each region in the case that one group of label to be found has intersection with cog region.
Fig. 3 is label configurations Bit String L and mapping group number schematic diagram.
Fig. 4 is system situation schematic diagram after cog region labeled packet.
Fig. 5 is that test time slot screens label process.
Fig. 6 is X, LiWith S (ri) between relationship.
Fig. 7 (a) and Fig. 7 (b) is that each method executes time comparison diagram when number of labels to be found changes under different scales.
Fig. 8 is each method execution time comparison diagram after the variation of target labels ratio.
Fig. 9 is each method execution time comparison diagram after group number variation.
Specific embodiment
The type of error that carrying out labeled packet using the present invention may occur can be divided into three classes, as shown in table 1.
1. false positive: some label of reader cog region and being not belonging to any one set of tags to be found, but pass through
It crosses screening and but belongs to certain a group of labels, this label is called false positive label.The probability that this thing happens is known as
False positive rate, this is similar to the false positive probability in Bloom filter.False positive probability is indicated with α.
2. erroneous packet: i.e. some label of reader cog region actually belongs to A group but by erroneous packet into B group.Also
There is a special case is, label actually belongs to some group, is but considered being not belonging to any one group, this feelings by screening
Condition is known as false negative.Some application programs, false negative and mistake classification are worthless or unacceptable.
3. grouping failure: i.e. some label of reader cog region shows to belong to multiple groups, in this case, nothing simultaneously
Method determines which group this label belongs on earth, or not belong to it is any group of, this label be known as be grouped unsuccessfully mark
Label.The probability that this thing happens is known as failure rate of classifying.Grouping probability of failure is indicated with β.
The citing of 1 labeled packet type of error of table
Multiple groups label parallel search method (MSTS) principle in extensive RFID system of the present invention are as follows:
One two is being constructed using tally set G to be found as map tags collection hash function collection H corresponding with its first
System Bit String L, the Bit String L constructed and all hash function collection H are broadcast to cog region label by reader later, are known
Other area label constructs " the mapping group number " of oneself according to the Bit String L and hash function collection H that receive, wherein having grouping failure
Label, map tags collection of these labels as next round.Then reader collects label information, according to time slot classification and returns
Complex information deactivates non-targeted label and excludes target labels existing for the target labels/confirmation being not present.Finally, being only left each
Group completes tag search operation in the label of cog region.
Detailed process of the present invention is as described below, it is assumed that this f group is assigned 1 first by a shared f groups to be searched respectively
To the sequence of f.In the first phase, one f binary strings are constructed, and give one group of corresponding hash function of each bit allocation
Collect H, each hash function collection has h hash function, and different.Due to there is f position, there be h Hash letter in each position
Number, a total of f*h hash function.For one hash function collection of each group of distribution, indicate for convenience, directly by primary Kazakhstan
Uncommon collection of functions distributes to first group, and so on, f hash function collection are distributed into f group.By each group of mark to be found
The group number of label collection is converted into corresponding f of binary string, corresponding position is marked 1, remaining position mark 0 obtains every group corresponding group
Number.Such as one share 5 groups, first group of primary hash function collection of correspondence, then first group of group number is 10000；With such
It pushes away, the hash function collection of the 5th group of correspondence the 5th, corresponding group number is 00001.
F set of tags to be found, the group number and " mapping group number " of label use g (t) and g'(t respectively in group) it indicates, according to
The present invention constructs rule, identical with the group number of a group of labels, but " mapping group number " probably due to the case where grouping failure without
Together.Key step of the present invention is as follows:
(1) it constructs Bit String L: utilizing each group of corresponding hash function collection, construct an a length of L includes all groups
The Bit String of label information.Detailed process is that is, the h hash function that the hash function that each group of label is organized using oneself is concentrated reflects
It is mapped on L, L is initially all 0, and the position mark 1 arrived by label mapping forms bit after all groups of all labels have all mapped
String L.
For example, as shown in Figure 3, it is assumed that now with 3 groups, first group of G1={ t1, group number g (t1)=100, it is corresponding
Hash function collection H1={ h1,h2}；Second group of G2={ t2, group number g (t2)=010, corresponding hash function collection H2={ h3,h4}；The
Three groups of G3={ t3, group number g (t3)=001, corresponding hash function collection H3={ h5,h6}；Bit String L is initially all 0, t1Use h1,
h2It is mapped on L, corresponding position mark 1；t2Use h3,h4It is mapped on L, corresponding position mark 1；t3Use h5,h6It is mapped on L, it is corresponding
Position mark 1；Binary bits string L is formed after the completion of all mapping etc. all labels.
(2) construction " mapping group number ": after having constructed L, then being each group of label acquisition " mapping group number ": again using all
The hash function collection of f group is mapped on Bit String L, constructs " the mapping group number " of an a length of f.Detailed process is that is, every
A group of label is successively mapped on L using the h hash function that the hash function of f group is concentrated, if this group of hash function collection
It is all mapped to 1, then correspondence mappings group number position is marked 1, as long as otherwise this group of hash function is concentrated with a hash function and reflects
It is mapped to 0, correspondence mappings group number position just marks 0.Form that label is corresponding " reflects after successively f hash function collection has been mapped
Penetrate group number ", this operation is carried out to each group of each label, corresponding distinctive " mapping group number " letter of each label can be obtained
Breath.Note that since L is corresponded to from the hash function collection mapping of group number position by label, so label corresponds to 1 of group number centainly
It is 1, because that all hash function can all be mapped to 1.For example, if first group of label group number is 10000, then
First of the correspondence " mapping group number " of this group of labels must be 0.
" mapping group number " why is constructed, is because when label to be found actually exists in cog region, according to the L of offer
With H map out come group number it is not necessarily identical with the group number of imparting, it is possible that grouping failure the case where, in this case without
Method determines which group it belongs to.So the label for being likely to occur such case is elected in advance on backstage in advance, in next round
It is excluded.It is observed that if " mapping group number " of label to be found can be successfully grouped, label if there is in
Cog region just centainly can also be successfully grouped, so only needing to pay close attention to the label of grouping failure in " mapping group number ".
The group number situation for illustrating grouping failure, as shown in figure 3, present first group of label t1To construct " mapping group
Number ", t1Successively use H1={ h1,h2, H2={ h3,h4, H3={ h5,h6Map L, wherein h1,h2,h5,h6It is mapped to 1, institute
With t1" mapping group number " first and third position mark 1；Wherein h3,h4It is mapped to 0, so t1" mapping group number " second mark
0, thus obtain its " mapping group number " g'(t1)=101.In this case, label t1It will occur point if there is in cog region
Group failure.
(3) reader broadcast data: reader broadcast Bit String L and all hash function collection H to cog region Y label,
Label successively maps L using f hash function collection, forms the corresponding group number of cog region label.The group number of cog region label has
Body construction process is identical with " mapping group number " construction process of label to be found, that is, cog region label successively utilizes the Hash of f group
H hash function in collection of functions is mapped on L, right only when all hash functions of every group of hash function collection are mapped to 1
Ying Weicai mark 1, as long as soon as otherwise thering is a hash function to be mapped to 0, corresponding position marks 0.
From the mapping result discussion of cog region Y label:
(1) there is 01 in label group number, illustrate that this label is not belonging to any one group, its silencing can be known with reducing
Other area number of tags.
(2) there is 11 in label group number, this label can be successfully grouped, and be no longer participate in the division operation of next round, only
Having into second stage can just reactivate.But it should be noted that label can have two kinds of situations at this time, the first situation is
This label really belongs to this group, and second situation is that this label is false positive label.For this two kinds of situations, whether for
False positive label is directly assigned in corresponding group, and next stage excludes false positive label again.
(3) there is more than one 1 in group number, then this labeled packet fails.In this case, label is uncertain belongs to which
Group, holding actively continue to participate in next round division operation.It so recycles until the label in tally set to be found is not grouped mistake
After the case where losing, the L of its construction is sent to cog region Y, after epicycle is screened, if still there is ungrouped label in Y, is said
This bright label is not belonging to label in tally set to be found certainly, can exclude.
As shown in figure 4, cog region number of labels can largely reduce after the first stage, all remaining participation second-orders
The label of section both knows about the group number of oneself, completes the division operation of cog region label.
Next second stage is individually carried out to every group of label to be found, second stage completes that the mesh of this group of label can be obtained
Label is marked, this operation successively is carried out to every a group of labels, until obtaining all groups of target labels, entire method is completed.
Test time slot screening label is formed more by taking turns, process are as follows: in order to without loss of generality, it is assumed that take turns in L, reader is first
First according to candidate target tally set ClTest time slot and non-test time slot are calculated, then from ClIt eliminates non-targeted label and confirms row
Except the non-label to be found in cog region.It repeats the above process, until meeting termination condition.
Start in L wheel, reader is first according to current candidate target labels collection ClTest time slot is calculated, then reader
The frame length L and hash function h used when broadcast construction test time slot.After receiving parameter information, cog region label is calculated
The identical hash function of test time slot selects their time slot on frame, then gives reader hair in their selected time slots
Send brief response.Reader is according to receiving the time slot of response from ClExclude non-targeted label:
In test time slot, if reader receives response, it replys a NAK and keeps enlivening for label.If reader
It is not responded in this time slot, it may determine that those ClThe middle candidate target label for selecting this time slot is not identifying now
Area must be non-target labels, and from ClExclude these labels to be found.
In non-test time slot, if reader receives some responses, non-label to be found must be from, it is logical that it replys ACK
These label silencings are known, to prevent them from participating in next screening wheel.
At the end of every wheel, reader will be updated candidate target tally set ClIt+1 and checks whether and reaches termination condition.Such as
It is unsatisfactory for termination condition to fruit, it will start a new wheel, and retest and elimination process.
As shown in figure 5, demonstrating how second stage works.In this illustration, dotted arrow indicates candidate mesh
It marks label and tests the mapping between time slot, and filled arrows indicate the transmission between cog region label and reader.Initial
Candidate target tally set C1=(t1,t2,t3,t4,t5).There are five cog region label, wherein t1And t3It is target labels.Fig. 5 (1)
Illustrate the execution of the first round of the invention with Fig. 5 (2).Reader is calculated in C first1Label test time slot and collect cog region
The response (Fig. 5 (1)) of label, then from C1It eliminates non-targeted label and confirms silencing in the non-label to be found of identification region
(Fig. 5 (2)).Reader replys NAK in 1, the 2 and 3 test time slots that it receives reply, keeps label active.Testing time slot 4 is
It is empty, so reader is from C1Exclude corresponding non-targeted label t4And t5.Meanwhile it time slot 5 reply ACK notify it is non-to be checked
Look for label Y2And Y3Silencing.Fig. 5 (3) and 5 (4) illustrate that the second wheel executes.The candidate target tally set C that reader updates2=
(t1,t2,t3), it recalculates test time slot and cog region label also reselects oneself time slot.Similar to the first round, read
It reads device and eliminates C2In be mapped to test time slot 3 non-targeted label t2, and notify to be mapped to the non-to be found of non-test time slot 4
Label Y1.If terminated after the second wheel of the present invention, this group searching result will be S (ri)=(t1,t3)。
The setting of grouping stage optimum bit string size:
The false positive rate of any hash function collection is calculated first
M is number of tags to be found, and f is that group number is long, and h is the hash function number of hash function collection H, and L is that Bit String is long.
The false positive probability of any Hash collection, i.e. a certain position false positive probability in f
In order to find the value for the hash function number h for minimizing false positive probability p, about the above-mentioned expression formula of h derivation and incite somebody to action
It is set as 0 to obtain p'=0, finds out optimal hash function number and is
A certain position is that the probability of false positive is minimum in group number at this time, is
Then the false positive rate of label is in the first stage
Pfp=fp (1-p)f-1≤fpbf
The classification failure rate of label is in first stage
Pcf=1- (1-p)f-1≤(f-1)p≤fpbf
Specifying the value of false positive rate to be equal to the value of classification failure rate at this time is α, enables Pfp< α and Pcf< α, i.e. fp < α and fp <
α, i.e.,It substitutes intoFinding out L is
L=Mlog0.6185 p
The long L of optimal bit string in the case where meeting specified false positive rate and the failure rate α that classifies
WhereinThen
Wherein α and f is known, it can thus be concluded that the long L of optimal bit string of first stage.
Test the best frame sign setting of time slot screening stage
In this part, the size of setting frame is determined how to minimize the non-targeted label excluded in candidate tally set
Average time cost CL。
Without loss of generality, consider L wheel.Allow NlIndicate reader riLocal active labels quantity, and allow flIndicate that frame is big
It is small.For a time slot in frame, it is that the empty probability cog region label that is equal to nothing selects the probability of the time slot, i.e.,
Obviously, a time slot is riThe probability of non-empty is (1-pe).Entirely the total duration of frame is
Wherein, teIt is the duration and t of an empty slotsIt is the duration of one short response time slot.
Allow ZLAll non-targeted tally sets, i.e. Z are taken turns in expression in LL=YL-(X∩Li), wherein X ∩ LiIt is to read ri's
Target labels collection.It therefore, can be from YLThe total number of labels of elimination is | ZL|。ZLThe selected test time slot of a label in if
Actually empty will be excluded.Because label to be found selects time slot together, in the expected non-targeted mark eliminated of this wheel
The quantity of label is about
Neli≈|Zl|*pe
From YLEliminate a non-targeted label average time cost be
In order to minimize Cl, allow
And ClMinimum value is obtained, when
Define load factor ρ=Nl/fl, the C when equation is set uplIt is minimized
It can be seen that ρ optimal value depends entirely on te/ts.In system model, te=0.184ms and ts=0.2ms.
In this case, the optimal value of ρ is 0.9697, and optimal frame sign is fl=Nl/ 0.9697=1.03*Nl。
Termination condition of the present invention
It determines how that termination condition of the invention is arranged now, to reach required required precision.Pay attention to (X ∩ Li)∈S
(ri).False positive number of tags is Nδ=| S (ri)|-|X∩Li|.Fig. 6 illustrates X, LiWith S (ri) between relationship.Target is high
The guarantee N of probabilityδP is not exceeded at leastreqThreshold value, that is,
P{Nδ≤Δ}≥preq
For example, if △=2, preq=0.95, meeting above-mentioned formula means that the probability more than or equal to 0.95 has at most
Two false positive labels are in search result.
In the present invention, the slave candidate target tally set C more taken turnslThe middle non-targeted label of iteration elimination.If in a certain wheel
(for example, L takes turns) does not have non-targeted label from ClIt excludes and is notified without non-wanted circular label, it may be speculated that false positive label
Number very little.In order to prove to infer and guarantee that false positive number of labels no more than a high probability, needs to observe continuous K
Wheel.If taking turns no non-targeted label in all K to eliminate, reader terminates the present invention.Otherwise, reader continues searching
Process.
Determine how setting K now to meet above-mentioned formula given △ and preq.Use EkExpression event is in all continuous K
A no non-targeted label of wheel is eliminated, and E is usedvExpression event just has V false positive label in the result.Then have
According to Bayes' theorem, probability P { Ev|EkCan be calculated by following formula
From the law of total probability, follow
Substituting into formula has
P{Ek|EvBe just have v false positive label and they no one of all K continuously take turns be excluded it is general
Rate.It is recognised that a non-targeted label cannot be arranged when being mapped to the test time slot an of non-empty from candidate target tally set
It removes.Therefore P { Ek|EvIt is equal to all v false positive labels in the probability of all K continuous wheel selection non-empty test time slots, i.e.,
P{Ek|Ev}=((1-pe)v)k=(1-pe)vk
(1-pe) it is the probability for testing time slot non-empty.
Because not knowing the distribution of v, it can be assumed that v follows univesral distribution, that is,
Formula is substituted into obtain
For the precision met the requirements, allow
1-(1-pe)(Δ+1)k≥preq
It can obtain
It means that
As front is analyzed, optimal ρ is In table
2 list to different △ and PreqCombined k minimum value.
2 difference △ and P of tablereqCombined k minimum value
Experiment simulation and analysis
In emulation experiment, channel keeps ideal situation, ignores the expenses such as redundant check.Four methods are realized to compare
Their performance: MSTS (i.e. of the invention), STEP, CATS, ITSP.The execution time (s) of application method compares as standard
The performance of distinct methods.Executing the time is the reaction most intuitive standard of searching method performance, under equal conditions, complete search when
Between shorter illustrate that searching method performance is better.Consider that three parameters of distinct methods performance may be influenced: (1) label to be found
Sum (| X |).(2) ratio of target labels and label to be found, be defined as η=| X ∩ Y |/| X |.(3) system label to be found
Group number (M).For each parameter setting, method 100 times of comparison are run, and calculate average metric data.
Simulating scenes and time setting
Simulating scenes: consider the RFID system for there are multiple readers.System default has 5 set of tags to be found, default portion
Affix one's name to 64 readers and 50000 labels.Reader is disposed with mesh model, and the distance between adjacent reader is set as √
2r.This leads to each reader, and there are about 1500 covering labels.The number of labels to be found of every group of default (| X |) it is arranged to
10000, the default scale (η) of every group of target labels is arranged to 0.2.For multiple reader scenes, graph coloring algorithm is used
To find the feasible schedule of reader.
Time setting: according to the duration of the different time slot of the time of EPC C1G2 UHF RFID label tag specified setting.
The rate that reader and label carry out data transmission is set to 62.5Kbps.Under this data rate, the time of different time-gap is such as
Under: tID=2.4ms, te=0.184ms, ts=0.2ms.Reader (is equal to a mark for 96 for binary bits string L points
Sign the length of ID) it is broadcasted, every section of transmission time is tID, then the transmission time of L is (L/96) * tID.It will be noted that
When data rate changes, absolute measure data may be different, but can achieve similar conclusion.
Different schemes performance compares
1) influence of every group of number of labels to be found
System is provided with the big system comprising 64 readers and 50000 labels, wherein label to be found has 5 altogether
Group indicates for convenience, reduces the influence of other factors, and it is identical to set different groups of number of labels to be found, and every group of target mark
The ratio of label is identical.As shown in Fig. 7 (a) and Fig. 7 (b), the execution of distinct methods when number of labels to be found increase is depicted
Time, wherein abscissa indicates the sum of 5 groups of labels to be found because calculate be all set of tags to be found search time it
With.Such as in Fig. 7 (a), every group of number of labels to be found is from 200 to 1200, then 5 groups of number of labels to be found in total are from 1000
To 6000.Wherein be divided to two kinds of situations to discuss again: one is small-scale label to be found, every group of number of labels to be found compared with
It is few；One is extensive label to be found, every group of number of labels to be found is more.Wherein Fig. 7 (a) indicates small-scale scene, Fig. 7
(b) extensive scene is indicated.
According to as a result, it can be observed that executing the time all with the increase of label to be found in the method for all considerations
And increase.Wherein in small-scale scene, the execution time of CATS method and ITSP method is better than the side STEP before 2500
Method, and when having arrived close to 5000, the execution time of CATS method alreadys exceed ITSP method, becomes and executes time longest side
Method, and said according to Fig. 7 (b) it is also seen that the execution time of CATS method is substantially increased with the growth of number of labels
Bright CATS method is only applicable to the tag search of smallest number, no matter and MSTS method is small-scale or extensive scene, protect always
It holds the optimal execution time, and as the increase of number of labels, the execution time of MTST method are slowly increased, stability is most
It is good.It is obtained according to Fig. 7 (b), more along with the change of number of labels, the method execution time is followed successively by CATS method from more to less,
ITSP method, STEP method and MTST method.
2) influence of every group of target labels ratio
The setting of this group of experiment scene with it is as before, wherein every group of number of labels to be found is 10000, it should be noted that change
Change is not change the quantity of 5 groups of labels to be found in the experiment of this group, only change the quantity of every group of target labels.Fig. 8 is depicted
The execution time of distinct methods when every group of target labels ratio η changes to 0.9 from 0.1.It can see from experimental result, when
When η becomes larger, MSTS method, STEP method, execution time of ITSP method all increases, and the execution time of CATS method not by
The influence of η.This is because the size of Bloom filter is according to the quantity of label to be found in CATS method | X | to be arranged
, this is unrelated with η.On the contrary, each reader region has more target labels when η increases, in this case, this
Inventive method and STEP method and ITSP method need to run more rounds to meet termination condition.But MTST of the present invention
Method always surpasses STEP method, CATS method and ITSP method.
3) influence of set of tags number to be searched
The setting of this group of experiment scene with it is as before, the ratio η of fixed target labels is 0.2, every group of number of tags to be found
Amount is 10000, then changes the group number of label to be found.As shown in figure 9, the group number for depicting label to be found becomes from 1 to 10
The execution time of distinct methods after change.It can see from experimental result, when set of tags number to be found becomes larger, MSTS method,
The execution time of STEP method, ITSP method, CATS method all increases.Wherein STEP method, ITSP method, CATS method
The execution time is substantially to be scaling up according to the quantity of group number, because these methods are unable to the knot of parallel computation multiple groups label
Fruit, can only single group search, repeat, so the time can be incremented by according to group number, and MSTS method in the first stage can be to all
It organizes while being operated, reduce cog region number of labels, and this operation comes into force to all groups of second stage, thus avoids
It repeats to cog region and carries out delete operation, so the group more big MSTS method advantage of the present invention of number is more obvious.
The embodiment of the present invention is described with above attached drawing, but the invention is not limited to above-mentioned specific
Embodiment, the above mentioned embodiment is only schematical, rather than limitation, those skilled in the art
Under the inspiration of the present invention, without breaking away from the scope protected by the purposes and claims of the present invention, it can also make very much
Form, within these are all belonged to the scope of protection of the present invention.
Claims (5)
1. multiple groups label parallel search method in a kind of extensive RFID system, which comprises the following steps:
The first step, it is assumed that have n set of tags to be searched；Server be each group of label to be found distribute a group number and with the group
Number corresponding hash function collection Hi, for all labels to be found, mapped to obtain using its corresponding hash function collection
One binary bits string L；
Second step, reader broadcast Bit String L and n hash function collection Hi(1≤i≤n), cog region label utilize n received
Hash function set HiMapping obtains the mapping group number of oneself on L；When the mapping group number is not 0, then corresponding cog region
Label, which is thought, oneself to be the member of mapping group and keeps active；When the mapping group number is 0, then corresponding cog region label is not
Label to be searched makes its silencing；Operation is repeated until all labels complete group number mapping；
Third step calculates test time slot and non-test time slot according to candidate target tally set Ci, then according to non-test time slot
Non-targeted label is deactivated in reply, the target labels being not present according to the tag return situation in test time slot from Ci exclusion are simultaneously true
Recognize existing target labels；It repeats the above process, until meeting termination condition.
2. multiple groups label parallel search method in extensive RFID system as described in claim 1, which is characterized in that first
In step, if a shared f group label to be searched, the group number that a length is f bit, the group are distributed for each group of label to be found
It number is formed by one 1 and f-1 0, wherein rule is arranged in the group number of i-th group of tally set are as follows: i-th of bit is set as 1, other
F-1 bit is set as 0.
3. multiple groups label parallel search method in extensive RFID system as claimed in claim 2, which is characterized in that third step
Treatment process includes:
Firstly, reader calculates test time slot according to current candidate target labels collection；
Then, the frame length and hash function used when reader broadcast construction test time slot；Cog region label is receiving parameter
After information, oneself corresponding time slot is selected on frame with hash function identical with calculating test time slot, then sending back should be extremely
Reader；
If reader receives response in test time slot, reader notifies corresponding label to keep active；
If reader receives response in non-test time slot, reader notifies corresponding label silencing；
Finally, reader updates candidate target tally set and is matched next time.
4. multiple groups label parallel search method in extensive RFID system as claimed in claim 2, which is characterized in that binary system
Bit-string lengthWherein, M is number of tags to be found, and f is that the group number of set of tags to be found is long, and α is to refer to
Fixed classification failure rate.
5. multiple groups label parallel search method in extensive RFID system as claimed in claim 3, which is characterized in that optimal frames
Size is fl=1.03*Nl；Wherein NlIndicate the local active labels quantity of reader.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811622037.2A CN109740029A (en) | 2018-12-28 | 2018-12-28 | Multiple groups label parallel search method in a kind of extensive RFID system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811622037.2A CN109740029A (en) | 2018-12-28 | 2018-12-28 | Multiple groups label parallel search method in a kind of extensive RFID system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109740029A true CN109740029A (en) | 2019-05-10 |
Family
ID=66361768
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811622037.2A Pending CN109740029A (en) | 2018-12-28 | 2018-12-28 | Multiple groups label parallel search method in a kind of extensive RFID system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109740029A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112650943A (en) * | 2020-12-24 | 2021-04-13 | 山东鑫泰洋智能科技有限公司 | Multi-cloud server collaborative data retrieval system and method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070126555A1 (en) * | 2005-12-06 | 2007-06-07 | Symbol Technologies, Inc. | Method and system for optimizing radio frequency identification (RFID) reader operation |
| CN103268465A (en) * | 2013-06-08 | 2013-08-28 | 无锡儒安科技有限公司 | Fast identifying method of tag type in radio frequency identification system |
| CN106096466A (en) * | 2015-04-27 | 2016-11-09 | 张大强 | Large-scale wireless radio system is lost the method for quick of label |
| CN106503603A (en) * | 2016-11-08 | 2017-03-15 | 湖南大学 | A kind of many reader RFID system fast tag recognition methods divided based on time slot |
-
2018
- 2018-12-28 CN CN201811622037.2A patent/CN109740029A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20070126555A1 (en) * | 2005-12-06 | 2007-06-07 | Symbol Technologies, Inc. | Method and system for optimizing radio frequency identification (RFID) reader operation |
| CN103268465A (en) * | 2013-06-08 | 2013-08-28 | 无锡儒安科技有限公司 | Fast identifying method of tag type in radio frequency identification system |
| CN106096466A (en) * | 2015-04-27 | 2016-11-09 | 张大强 | Large-scale wireless radio system is lost the method for quick of label |
| CN106503603A (en) * | 2016-11-08 | 2017-03-15 | 湖南大学 | A kind of many reader RFID system fast tag recognition methods divided based on time slot |
Non-Patent Citations (1)
| Title |
|---|
| 关点: "大规模RFID系统中高效标签信息收集协议研究", 《HTTPS://D.WANFANGDATA.COM.CN/THESIS/CHJUAGVZAXNOZXDTMJAYMJAZMJMSCUQWMTQ5OTUXOBOIEDNLDZLNNDK%3D》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112650943A (en) * | 2020-12-24 | 2021-04-13 | 山东鑫泰洋智能科技有限公司 | Multi-cloud server collaborative data retrieval system and method |
| CN112650943B (en) * | 2020-12-24 | 2022-07-26 | 厦门地铁创新科技有限公司 | Multi-cloud server collaborative data retrieval system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Kang et al. | Deriving stacking strategies for export containers with uncertain weight information | |
| Strehl et al. | Distance based clustering of association rules | |
| CN108446577A (en) | Multi-label anti-collision method in high-frequency/ultrahigh frequency RFID recognition system | |
| Liu et al. | Efficient polling-based information collection in RFID systems | |
| CN103268465B (en) | The method for quickly identifying of label classification in a kind of radio-frequency recognition system | |
| Pupo et al. | ReliefF-ML: an extension of ReliefF algorithm to multi-label learning | |
| CN105260742A (en) | Unified classification method for multiple types of data and system | |
| Liu et al. | Time efficient tag searching in large-scale RFID systems: A compact exclusive validation method | |
| Han et al. | A protocol for detecting missing target tags in RFID systems | |
| CN109740029A (en) | Multiple groups label parallel search method in a kind of extensive RFID system | |
| Liu et al. | On improving write throughput in commodity RFID systems | |
| Fu et al. | Active learning with optimal instance subset selection | |
| CN104166867B (en) | A kind of many HASH functions multiframe coupled mode RFID anti-collision algorithms (MHMFG) | |
| Zhang et al. | Seeding the kernels in graphs: Toward multi-resolution community analysis | |
| Winarko et al. | Discovering richer temporal association rules from interval-based data | |
| Brunello et al. | Towards interpretability in fingerprint based indoor positioning: May attention be with us | |
| Yang et al. | Fast RFID tag sorting at the edge for Internet of Things | |
| Kocher et al. | Scaling Density-Based Clustering to Large Collections of Sets. | |
| CN104700055A (en) | Method for detecting complex events on multi-probability RFID event flows | |
| CN109117687A (en) | The loss label Iceberg queries method of extensive grouping RFID system based on HZE | |
| Laha et al. | A new heuristic for minimizing total completion time objective in permutation flow shop scheduling | |
| CN113344140A (en) | Uncertain data sequence scanning method and system based on pruning conditions | |
| Portegys | A search technique for pattern recognition using relative distances | |
| Guo et al. | Efficient collision-slot utilization for missing tags identification in RFID system | |
| Giacinto et al. | Automatic design of multiple classifier systems by unsupervised learning |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20190510 |