CN106503603A - A kind of many reader RFID system fast tag recognition methods divided based on time slot - Google Patents

Abstract

The invention discloses a kind of many reader RFID system fast tag recognition methods divided based on time slot, are comprised the following steps：In the first stage, when frame starts, reader group gives one querying command of tag broadcast simultaneously, the tag return to be received such as then, the label of certain reader inquiry is successfully received according to the parameter included in querying command, select to reply time slot using hash function, then the RN for carrying is returned to reader in its time slot for selecting by the label, in a slot, if reader is properly received RN, the reader records this RN, other confiscate the reader of RN or detect signal conflict reader unification wait for a period of time, then one QueryRep order of reader group synchronized transmission starts a new time slot；Second stage, in each time slot, in reader group, all readers send the ACK comprising the RN being previously respectively received, and the label with identical RN sends the ID of the label to reader.The present invention can fully eliminate RT conflicts.

Description

A kind of many reader RFID system fast tag recognition methods divided based on time slot

Technical field

The present invention relates to a kind of many reader RFID system fast tag recognition methods divided based on time slot.

Background technology

Nowadays RF identification (RFID) technology is widely used in many fields, including supply chain monitor, storehouse management and Storage controlling.In such applications, physical object is labeled with the RFID label tag being passively or actively, and each label is unique comprising one ID.RFID reader collects the presence of the ID detection objects of appended label by wireless channel.Because the inquiry of reader Region is limited, and large-scale RFID system is usually deployed multiple reader cooperations and covers all of label.However, in many readers In RFID system, three kinds of important conflicts between label and reader, are there are：Tag-tag (Tag-Tag, TT) conflicts, reads Read device-label (RT) conflict and reader-reader (RR) conflicts.Which is defined as follows (Fig. 1 is shown in by schematic diagram)：

Tag-tag (TT) conflicts：When in the interrogation of same reader two or more labels with When sending a signal to same reader (Fig. 1 (a)), there is TT conflicts.

READER-TO-TAG (RT) conflicts：RT conflicts occur when a reader is in the interference region of another reader When (Fig. 1 (b)).In this case, the signal of B can disturb the signal transmission from label 3 to A, and this causes A None- identifieds label 3. Worse, this conflict may prevent A to receive the response from any label in whole process.

Reader-reader (RR) conflicts：If a label is located at the read area of two (or multiple) adjacent reader , then, when the label is simultaneously received from the signal of two (or multiple) readers, there is RR conflict (Fig. 1 in the overlapping place in domain (c)).It should be noted that although RR conflicts can be such that the label of overlapping region is accessed, reader still can read Other labels in oneself interrogation.

Existing single reader-tag recognizer is primarily upon how solving the tag-tag punching in single reader system Prominent, such as document [1,7,15].Present invention is primarily concerned with other two kind conflict of the point in many readers：READER-TO-TAG conflicts (Fig. 1 (b)) and reader-reader conflict (Fig. 1 (c)).This respect has had an a few thing, such as document [9,20,23, 30].These have worked and have focused mainly on for adjacent reader scheduling avoiding RT conflicts and RR in different time operations Conflict.Although conflict is reduced, but the reader for being capable of management and running simultaneously in these systems is only all readings in system The sub-fraction of device, this significantly limit the identification handling capacity of label in system.The Season methods proposed in document [30] Label in by dispatching all of reader come identifying system simultaneously.However, the method can not solve RT conflicts, its performance pole The big deployment way for depending on reader.When reader random placement, its performance can decline to a great extent.

Existing tag identification technologies can be divided into two big class, and the tag recognition algorithm in single reader system is read with more Tag recognition algorithm in device system.Single reader-tag recognition methods mainly has two kinds of technology：Technology based on ALOHA [1, 6,7,15] technology [9] that and based on tree decomposes.However, the tag recognition algorithm in single reader system do not account for reader- Reader collision and READER-TO-TAG conflict, the effect that therefore can not be obtained in many reader systems.

Prior art in many reader systems mainly includes [21,22,23,30].This kind of algorithm is adjacent by dispatching The reader of conflict may be caused in different time services, so as to avoid reader-reader conflict and READER-TO-TAG Conflict.However, this method prevents adjacent reader from while work, being therefore capable of readding for management and running simultaneously in system Read the sub-fraction that device only accounts for all reader numbers in system, strongly limit the tag recognition handling capacity of system.Document [30] propose in and ignore conflict first, dispatch all readers to carry out the Season methods of tag recognition.However, the method R-T conflicts are not solved, (the i.e. situation of reader rule deployment in the case where non conflicting region is much smaller than conflict area is only capable of Under) obtain better effects.When reader random placement, its performance drastically declines.

EPC C1G2 agreements are one of single reader-tag identification protocols most widely used at present.It is using multiple frames Mode is recognizing label.When there is Unidentified label in identification region, reader is joined comprising frame sign by broadcasting one Starting a frame, wherein f indicates how many time slot in the frame for the order of number f.When label receives the order, each label A random value is selected from [0, f-1], and the value is loaded into the time slot counter of oneself.Meanwhile, each label by using Randomizer generates 16 random numbers, i.e. RN.The transmission QueryRep orders that reader subsequently repeats indicate next The beginning of individual time slot.Whenever a label receives a QueryRep order of reader, its time slot counter is subtracted by the label 1.When the value of the time slot counter of label is zero, RN is sent by the label by backward scattered method (Backscatter) Give reader.

As shown in Fig. 2 according to the difference for sending the number of the label of RN simultaneously, have in EPC C1G2 three kinds different when Gap：Single time slot, conflict time slot and empty slot, wherein have one respectively, more than one, and send RN to reader without label. The duration of different time-gap is different, consequently leads to the asynchronism between different readers, and reads just because of this The asynchronism that reads between device result in READER-TO-TAG conflict.For example, it is assumed that the time slot sequence of the reader A and B in Fig. 1 (b) Row are (conflict, single, sky) and (single, conflict, empty) respectively, then the QueryRep orders of reader B would interfere with label 3 and arrive A's ID is transmitted, so as to there is RT conflicts (because the order of reader B broadcast can interfere with reader A when label 3 sends ID to A Signal identification to label 3).As the time-slot sequence of each reader is random, based on traditional EPC C1G2 agreements it is READER-TO-TAG conflict can not possibly be solved.

RF identification, RFID (Radio Frequency Identification) technology, also known as radio frequency identification, is A kind of communication technology, can be recognized by radio signals and specific objective read and write related data, and without the need for identifying system with specific Machinery or optical contact is set up between target.

Reader：It is used for the equipment is identified by label, can be with computer by wirelessly or non-wirelessly network connection.

Label：Can adhere on object come the equipment that object is identified and records its information.

Frame：The mode communicated with label by reader.Each frame includes multiple time slots, and each time slot is carried out by reader Synchronous.In each time slot, reader is first to all tag broadcast querying commands, and label returns corresponding letter to reader Breath.Each label only selects that a time slot to send the information of oneself to reader in each frame.

RN16：The random number of one 16 bit.Channel contention is used only in conventional labels identification protocol, and at me Agreement in can be also used for judging whether label is overlapping tags and new label.

Content of the invention

The technical problem to be solved is, not enough for prior art, there is provided a kind of reader-tag identification side Method.

For solving above-mentioned technical problem, the technical solution adopted in the present invention is：A kind of many readings divided based on time slot Device RFID system fast tag recognition methods, comprises the following steps：

1) when in the first stage, frame starts, reader group simultaneously gives one querying command of tag broadcast, then etc. to be received Tag return, is successfully received the label of certain reader inquiry according to the parameter included in querying command, using hash function Select to reply time slot, then the RN for carrying is returned to reader in its time slot for selecting by the label, in a slot, if read Device is properly received RN, and the reader records this RN, and other are confiscated the reader of RN or detect the reading of signal conflict Device unification is waited for a period of time, and then one QueryRep order of reader group synchronized transmission starts a new time slot；

2) second stage, in each time slot, in reader group, all readers send to include and were previously respectively received RN ACK, the label with identical RN sends the ID of the label to reader.

The reader group selection process is：

1) initialization enlivens reader set A for sky；

2) when new reader is added, the subregion si of unlabelled maximum weight is found, and this sub-regions is set It is set to marked state；

3) all readers to associating above-mentioned subregion are tested in turn, if certain reader is added to active readding Read, in device collection A, to enliven the reading efficiency value of reader set and for E (rj)=eff ({ rj } ∪ A)；eff The reading efficiency of ({ rj }) for reader rj；For the reader group of subregion si can be covered；

4) pick out and system effectiveness can be caused to increase maximum reader rk so that enliven the reader of reader set A Efficiency increases, and the reader is added to A；

5) all subregions that rk is covered are arranged to flag state；

6) repeat step 1)～5), until adding without new reader.

The length of the frame is：

Wherein, n_riFor the ID numbers that reader ri is collected；t_rnLength for each time slot of RN collection phases；t_idReceive for ID The length of each time slot of collection stage.

Compared with prior art, the present invention had the advantage that for：The present invention can fully eliminate RT conflicts, with When propose a kind of dispatching method effectively to reduce RR impacts of the conflict to tag recognition efficiency.Compare existing tag recognition side Method, the present invention can effectively improve tag recognition handling capacity 198%.

Description of the drawings

Fig. 1 (a), Fig. 1 (b), Fig. 1 (c) are respectively the conflicts of the TT in many reader systems, RT conflicts, RR conflicts；

Fig. 2 is EPC C1G2 protocol implementations；

Schematic diagrames of the Fig. 3 for time slot partitioning technology；

Fig. 4 is subregion schematic diagram；

Fig. 5 is maximum throughput reader selection algorithm flow process；

Fig. 6 is the optimal frame length in 100 simulating scenes, and 100 readers, weights are between 100 to 200；

Scheduling times of the Fig. 7 (a) and Fig. 7 (b) for algorithms of different；

Handling capacities of the Fig. 8 (a) and Fig. 8 (b) for algorithms of different；

Fig. 9 (a), Fig. 9 (b) and the identification that Fig. 9 (c) is different-scale systems in random placement postpone to compare；

Figure 10 is that the identification in regular deployment scheme postpones；

Duration of the Figure 11 (a) for EPC C1G2 different time-gaps；Figure 11 (b) is the signal in time slot division；

Figure 12 selects example for enlivening reader.

Specific embodiment

The present invention observes and be required for carrying out same behavior being to collect RN in all of reader of each time slot：Read Device is in each time slot broadcast QueryRep, and waits label to send RN (s).Only when reader is successfully received a RN, read Read device by replying the ACK for including received RN, could require that label sends its ID.Single time slot has maximum duration, because ID is needed to transmit for it.If reader is only collected RN (s) in single time slot and is not replied ACKs (so as to not cause ID to transmit), Single time slot will have and conflict time slot identical duration.I.e. reader directly transmits one after receiving RN (s) from label QueryRep orders start next time slot.If additionally, the duration of empty slot is expanded to and the time slot that conflicts/mono- time slot phase With, then all of reader can all execute simultaneously operating, i.e., the transmission QueryRep orders that they can be simultaneously are simultaneously connect from label Receipts RN (or wait, but this does not obviously interfere with other readers).We are referred to as time slot this technology and divide (Slot Splitting).The tag recognition process for employing time slot partitioning technology includes two stages：RN collection phases and ID collect rank Section.In RN collection phases, all of reader sends QureyRep orders and receives RN (or wait)；In ID collection phases, institute Some readers send the ACK comprising the RN having successfully received in single time slot in the first stage and collect corresponding ID.Using this The technology of kind, all of reader is carried out same action at any time, or sending signal, or receiving signal, and not The problem of a reader sending signal and another reader reception occurs, so as to completely eliminate READER-TO-TAG punching Prominent.What time slot was divided comprises the following steps that：

Step 1：RN collection phases：RNs only collected by reader.When frame starts, tag broadcast one given first by reader Inquire about to select their answer time slot.Its RN is transferred to reader in its time slot for selecting by label.If reader success Receive RN, it records this RN.Then, without replying ACK, it by send QueryRep order beginning one new when Gap.

Step 2：ID collection phases：Reader collects tag ID according to the RNs that the first stage receives.In each time slot, Reader sends the ACK comprising previously received RN.The label with identical RN sends its ID to reader.Obviously, The number of labels of collection is solely dependent upon the quantity of the RNs for collecting in the first stage.

Time slot is divided and easily can be carried out rearranging realization by the sequential relationship of original EPC C1G2 agreements.To mark Sign and do not change：RNs replied by label when time slot counter is zero, and receive include before send RNs ACK when reply Tag ID.And for reader, unique modification is that they change the time for sending ACK.Fig. 3 gives a time slot The example of division.

1) the maximized reader-reader conflict avoidance of reading efficiency is based on

Time slot partitioning technology is based on, all of reader can be worked simultaneously without being subject to reader-mark in system Sign the interference that (RT) conflicts.But reader-reader (RR) conflict is still present so that collectively covered area by multiple readers Label in domain can not be successfully identified.Often increase by one and enliven reader, be also just likely to increase reader-reader (RR) region for conflicting, causes in some cases while the more many effects on the contrary of the reader of work are poorer.

In order to further improve reading efficiency, the present invention proposes a kind of reader dispatching algorithm to pick out one group of reading Device works and causes system reading efficiency highest, the algorithm to be referred to as Federal.For the clearer description for providing this agreement, We are given first and are defined as below：

1) subregion.One sub-regions are such a regions, and in region, all of point can be by identical reader collection Close and cover.As shown in figure 4, each reader has the reading area of oneself.The reading area of adjacent reader is overlapped, will The region division that whole system is located is many sub-regions.

2) associate reader/association subregion.If subregion s is in the read range of reader r, s is called the association of r Subregion, r are called the associate reader of s.In many reader systems, a reader generally can associate many sub-regions, And sub-regions can also associate multiple readers.All associate reader collection of subregion s are combined intoAs shown in figure 4, readding Read device r1 and be associated with { s1, s2, s8, s9 } 4 sub-regions.Likewise, subregion s2 is associated with { r1, r2 } two readers.

3) single subregion/cover subregion of covering more.If just there is one to be scheduled in the associate reader of sub-regions Work, then the region is called single covering subregion.If there is more than one scheduled work in the associate reader of sub-regions, Then the region is called many covering subregions.Label in the domain of multiple coverage areas can be because receive the information of multiple readers and send out simultaneously Raw reader-reader (RR) conflict.Obviously only single label covered in subregion can be successfully identified.

4) weights of the weights/reader of subregion.The weight w (s) of one sub-regions s is defined as in the subregion not yet Identified number of tags.The weight w (r) of one reader r is defined as the weights sum of single overlay area in the reader.

5) reading efficiency.Give a reader set A={ r1, r2 ..., rm }, and assume w=Max w (ri) | ri ∈A}.Then in the reader set, the reading efficiency of arbitrary reader ri is defined as eff (ri)=w (ri)/w.

May certify that, it is assumed that A is one group and enlivens reader set, then in A reader reading efficiency and higher, system Tag recognition handling capacity bigger.The change for enlivening reader set also results in single change for covering subregion.Often increase by one Individual enliven reader, often increase single cover subregion, but be also possible to for existing single subregion that covers to become many coverings Subregion, so that the label in the region can not be identified, reduces the sum of reader reading efficiency on the contrary.Therefore, under us Face provides an algorithm and enlivens reader finding in system one group, is reduced between reader by maximizing reader efficiency RR conflict, so as to effectively improve system identification handling capacity.

The algorithm tests each reader by way of iteration successively, and satisfactory reader is added to active readding Read in device set A.Most start A for empty set.When needing to add new reader every time, the algorithm finds unlabelled maximum weight Subregion si, and this sub-regions is arranged to marked state.Then to associating all reading device wheels of this sub-regions Stream is tested, and is calculated and is assumed to be added to certain reader to enliven in reader collection A, enlivens the reading efficiency value of reader collection SumPick out and system effectiveness can be caused to increase maximum that reader rk so thatAnd add it to enliven reader collection.Repeat this flow process until not new Reader can add into.

The detailed process of algorithm is presented in Fig. 5.

Optimal frames length is arranged：

Due to the weights number of unidentified label in each reader coverage (be different) of each reader, because There is no a frame length that optimum is for all readers and arrange in this.Here we provide a setting optimal frame length The method of degree.Due to the RN numbers that the frame length of ID collection phases was collected depending on the first stage, so it is contemplated that how to exist First stage minimizes the acquisition time of each RN to arrange frame length.

Assume that the ID numbers that collects in second stage reader ri are n_ri.Assume the length of each time slot of RN collection phases simultaneously Spend for t_rn, and the length of ID collection phases each time slots is t_id.During so whole tag recognition needed for each tag recognition Time be

It is easy to draw

It is then maximization tag recognition handling capacity, then the average time of tag recognition should be minimized, so optimum Frame length should be

As above formula does not have analytic solutions, two methods are we have proposed：One kind is by f₁It is set in all w (ri) Between be worth, another kind is by f₁It is set to the mean value of all w (ri).Fig. 6 gives setting for the optimal frames in 100 simulated scenarios Result is put, shows f₁The median of w (ri) of the substantially all readers of optimal value or so.

1) performance of the time slot partitioning technology in single reader system

The average recognition time of label in single reader system derives as follows.Without loss of generality, it is considered to any one frame.With T_totalRepresent the total time (including stage 1 and stage 2) of this wheel.N ＇ are allowed to represent the number of labels determined in this wheel.Then know The average time of an other label is

The time-slot duration in two stages is different.As ID transmission needs to take more time than RN transmission, Longer than the time-slot duration trn in stage 1 in the time-slot duration tid in stage 2.F is used respectively₁And f₂Expression stage 1 and rank The frame sign of section 2.Then, total time Ttotal=f₁×trn+f₂×tid：Frame sign f₂It is equal to the RNs numbers that collects in the stage 1 Amount, i.e. f₂=n ＇=f₁×P1；P1 is the ratio of the single time slot in the stage 1.N represents the Unidentified label when this wheel starts Sum.Known as f1=n, a time slot is the maximum probability of single time slot, in this case

Wherein e is natural constant.The number of labels n ＇ of collection=f1 × P1 ≈ n × e^-1.So as to divide identification one with time slot The expeced time of individual label is

2) the identification handling capacity in many reader systems

We conducted substantial amounts of emulation and come comparison Federal methods proposed by the present invention and other two kinds of main readings Device dispatching algorithm：Season [28] and EGA [31].Index used in assessment includes that scheduling time (is defined as collection system In time needed for all of label), tag recognition handling capacity (is defined as the quantity of the label of collection per second), and tag recognition Postpone.It is used for calculating the scheme that clocks of scheduling time, it then follows EPC C1G2 standard criterions [5], it is assumed that data rate is 62.5Kbps.For the duration of Season and EGA, empty slot, single time slot and conflict time slot is respectively 0.184ms, 2.4ms And 0.44ms.For Federal, it is 0.44ms in the duration of one time slot of RN collection phases, and in ID collection phases one The duration of individual time slot is 1.96ms.

It is contemplated that the system of three kinds of different scales：Little, in, greatly.For each scale, it is contemplated that two kinds of readers Deployment strategy：Rule deployment, reader wherein are deployed to a mesh model, random placement, and reader wherein is Random distribution.In randomized policy, we delete unnecessary reader and keep their quantity minimum.For identical scale, Reader quantity in randomized policy is twice of reader quantity or so in rule strategy.Label is evenly distributed on system In, each reader covers about 500 labels.Interference radius is set to inquire 1.5 times of radius.Simulation parameter is listed in table 1.

1 simulation parameter of form is arranged

System size	Little	In	Greatly
				Reader number (rule deployment)	16	64	121
Reader number (random placement)	About 30	About 130	About 250
				Number of tags	4000	16000	36000

C. simulation result

Median and average：As it was previously stated, frame sign should be set to all " centre " values for enlivening reader weight.I Compare two kinds of heuristics：Median, uses the median of all reader weights as frame sign；Average, makes Average weight with all readers.Table 2 lists two kinds of tactful scheduling times and handling capacity in a minisystem.Flat Mean method is slightly better than median method.Therefore, our average counting methods used in following emulation.

2 median of form and the comparison of mean value

	Time (second)	Handling capacity (label is per second)
			Using median (random/rule)	1.627/2.092	2462/1917
Using mean value (random/rule)	1.576/1.958	2540/2045

2) scheduling time：As shown in Fig. 7 (a), when reader rule deployment, when the systems get large, all three algorithm Scheduling time only increases a bit.Federal and Season are substantially better than EGA, and Federal is further reduced about than Season 30% time.When reader is by random placement (Fig. 7 (b)), when system becomes big, the scheduling time of Season and EGA dramatically increases, Because they need more wheel numbers to arrange all of reader.By contrast, the scheduling time of Federal only slightly increases Plus.The time that all labels save 66% and 68% is collected compared to Season and EGA, Federal.Additionally, when system is advised When mould rises, improvement becomes readily apparent from.We can also be observed that Season is compared to EGA when reader is by random placement Improvement start to become less significant, Season is because most of label is located at overlapping region in this case, it is impossible to Its first stage effectively recognizes label.

Handling capacity：As shown in Fig. 8 (a) and Fig. 8 (b), when system scale increases, more readers can work parallel Make, so as to improve the handling capacity of all three algorithm.When reader rule deployment, Season and EGA executes good, such as Fig. 8 Shown in (a).However, when reader is by random placement, they execute variation, shown in such as Fig. 8 (b), because a reader May be than disturbing more readers under regular deployment mode.This causes Season and EGA using more wheels to dispatch There is reader.By contrast, Federal realizes similar handling capacity under both deployment scenarios.Compared to Season and EGA, in the case of rule deployment, Federal increases handling capacity up to 36% and 78%, is more up in the case of random placement 194% and 218%.Additionally, when system becomes much larger, Federal is become readily apparent to the improvement of Season and EGA.

D. identification postpones

Fig. 9 (a)～Fig. 9 (c) depicts the duration that the identification in reader random placement in algorithms of different postpones. As can be seen that can quickly recognize label (in the initial 500ms of mini system, in medium-and-large-sized system in three kinds of algorithms of scheduling initial stage Initial 1000ms).However, As time goes on, the identification of Season and EGA postpones substantially to increase, because only that one is little Part reader can be arranged work, and the label outside these reader interrogations is waited before recognition for a long time. By contrast, the most of readers in Federal can constantly recognize label.Season and EGA is compared, in Federal 49% (1831ms vs.3601ms) and 48% (1831ms vs.3574ms) is reduced respectively the time required to the label of identification 80%. We to can also be observed that and be close to EGA in random placement scheme Season performance.However, in regular deployment scheme, Season More far better than EGA, as shown in Figure 10.Under regular deployment scheme, incipient stage Season recognize label speed even Than Federal faster.This is because all of labeling requirement waits RN collection phases to terminate in Federal.Crosspoint occurs Label 70% is identified, and afterwards, Federal can faster recognize label than Season.

Platform validation：In USRP software definition platforms, we realize that time slot divides and uses programmable WISP label Verifications which can Row.We using have two can be in the USRP platforms of the RFX900 daughter boards of 900MHz band operations.USRP stage+modules two Individual antenna is simultaneously connected to a notebook computer as reader.Four WISP Programmables based on 4.1 firmwares of DL-WISP Part is used as label.Because most of common operation (such as QUERY, ACK) realizes in USRP and WISP firmwares that we only need Want slight modifications firmware code to realize the prototype of our time slot partitioning technologies.The experiment is can be with order to verify that time slot is divided The action of step reader, and meet existing tag recognition standard.

The present invention is used as reader and four programmable WISP labels by the USRP1 of two antennas to realize time slot stroke Point.

Figure 11 (a) and Figure 11 (b) show the duration of different time-gap in EPC C1G2 and time slot are divided.Such as Figure 11 Shown in (a), there are the different duration in EPC C1G2 different time-gaps.Figure 11 (b) shows the signal in time slot division.Same All time slots in stage all have the identical duration.Note, in empty slot, reader is sending next one QueryRep Wait a very short time before, make the duration of empty slot identical with other time slots.In order to preferably may be used in figure The property read we stretch the duration of time slot, but it is constant to be to maintain its relative scale, i.e. tE:tS:TC=1:13:2.4, tE, tS It is the duration of empty slot, single time slot and conflict time slot respectively with tC.

Figure 12 shows an example for enlivening reader selection.There are five reader R={ a, b, c, d, e } in system. In the weight that the numeral of every sub-regions represents it.According to reader selection algorithm, reader a, c, d, e, b are surveyed successively Examination.Because reader efficiency Eff (a, c, d, e, b)<Eff (a, c, d, e), last result are A={ a, c, d, e }.

Bibliography

[1]Zhen Bin,Mamoru Kobayashi,and Masashi Shimizu.Framed aloha for multiple RFID objects identification.IEICE Transactions on Communications,88 (3):991–999,2005.

[2]Kai Bu,Junze Bao,Minyu Weng,Jia Liu,Bin Xiao,Xuan Liu,and Shigeng Zhang.Who stole my cheese？:Verifying intactness of anonymous RFID systems.Ad Hoc Networks,36:111–126,2016.

[3]Kai Bu,Mingjie Xu,Xuan Liu,Jiaqing Luo,Shigeng Zhang,and Minyu Weng.Deterministic detection of cloning attacks for anonymous RFID systems.IEEE Transactions on Industrial Informatics,11(6):1255–1266,2015.

[4]Binbin Chen,Ziling Zhou,and Haifeng Yu.Understanding RFID counting protocols.In Proc.of Mobicom,pages 291–302,2013.

[5]EPCglobal.EPC Radio-Frequency Identity Protocols Class-1Gen-2UHF RFID Protocol for Communications at 860MHz-960MHz,2008.

[6]Wei Gong,Kebin Liu,Xin Miao,and Haoxiang Liu.Arbitrarily Accurate Approximation Scheme for Large-Scale RFID Cardinality Estimation.In Proc.of Infocom,pages 477–485,2014.

[7]Lei Kang,Kaishun Wu,Jin Zhang,Haoyu Tan,and Lionel M.Ni.DDC:A Novel Scheme to Directly Decode the Collisions in UHF RFID Systems.IEEE Transactions on Parallel and Distributed Systems,

23(2):263–270,2012.[8]Dheerag K.Klair,Kwan-Wu Chian,and Raad Raad.A Survey and Tutorial of RFID Anti-Collision Protocols.IEEE Communications Survery and Tutorials,2(3):400–421,2010.

[9]M.Kodialam and T.Nandagopal.Fast and reliable estimation schemes in RFID systems.In Proc.of ACM Mobicom,pages 322–333,2006.

[10]Linghe Kong,Liang He,Yu Gu,Min-You Wu,and Tian He.A Parallel Identification Protocol for RFID Systems.In Proc.of Infocom,pages154–162, 2014.

[11]Tao Li,Shigang Chen,and Yibei Ling.Efficient Protocols for Identifying the Missing Tags in a Large RFID System.IEEE/ACM Transactions on Networking,21(6):1974–1987,2013.

[12]Xuan Liu,Bin Xiao,Shigeng Zhang,and Kai Bu.Unknown Tag Identification in Large RFID Systems:An Efficient and Complete Solution.IEEE Transactions on Parallel and Distributed Systems,26(6):1775–1788,2015.

[13]Xuan Liu,Bin Xiao,Shigeng Zhang,Kai Bu,and Alvin Chan.STEP:A time-efficient tag searching protocol in large RFID systems.IEEE Transactions on Computers,64(11):3265–3277,2015.

[14]Xuan Liu,Shigeng Zhang,Bin Xiao,and Kai Bu.Flexible and timeefficient tag scanning with handheld readers.IEEE Transactions on Mobile Computing,15(4):840–852,2016.

[15]Wen Luo,Shigang Chen,Tao Li,and Yan Qiao.Probabilistic missingtag detection and energy-time tradeoff in large-scale RFID systems.In Proc.of Mobihoc,pages 95–104,2012.

[16]Vinod Namboodiri and Lixin Gao.Energy-aware tag anticollision protocols for RFID systems.IEEE Transactions on Mobile Computing,9(1):44–59, 2010.

[17]Ettus Research.Synchronization and mimo capability with usrp devices,2014.https://www.ettus.com/content/files/kb/mimo and sync with usrp updated.pdf.

[18]Muhammad Shahzad and Alex X.Liu.Probabilistic optimal tree hopping for RFID identification.In Proc.ofSIGMETRICS,pages 293–304,2013.

[19]Bo Sheng,Qun Li,and Weizhen Mao.Efficient Continuous Scanning in RFID Systems.In Proc.of Infocom,pages 1010–1018,2010.

[20]Shaojie Tang,Cheng Wang,Xiang-Yang Li,and Changjun Jiang.Reader activation scheduling in multi-reader RFID systems:A study of general case.In Proc.of IPDPS,pages1147–1155,2011.

[21]ShaoJie Tang,Jing Yuan,Xiang-Yang Li,Guihai Chen,Yunhao Liu,and JiZhong Zhao.Raspberry:A stable reader activation scheduling protocol in multi-reader RFID systems.In Proc.of ICNP,pages 304–313,2009.

[22]James Waldrop,Daniel W.Engels,and Sanjay E.Sarma.Colorwave:an anticollision algorithm for the reader collision problem.In Proc.Of ICC,pages 1206–1210,2003.

[23]Chuyu Wang,Lei Xie,Wei Wang,Tao Xue,and Sanglu Lu.Moving tag detection via physical layer analysis for large-scale RFID systems.In Proc.of Infocom,pages 1–9,2016.

[24]Qingjun Xiao,Min Chen,Shigang Chen,and Yian Zhou.Temporally or Spatially Dispersed Joint RFID Estimation Using Snapshots of Variable Lengths.In Proc.of Mobihoc,pages247–256,2015.

[25]Kun Xie,Xin Wang,XueLi Liu,Jigang Wen,and Jiannong Cao.Interference-aware cooperative communication in multi-radio multichannel wireless networks.IEEE Transactions on Computers,

65(5):1528–1542,2016.[26]Kun Xie,Xin Wang,Jigang Wen,and Jiannong Cao.Cooperative routing with relay assignment in multiradio multihop wireless networks.IEEE/ACM Transactions on Networking,24(2):859–872,2016.

[27]Lei Xie,Jianqiang Sun,Qingliang Cai,Chuyu Wang,Jie Wu,and Sanglu Lu.Tell me what I see:Recognize RFID tagged objects in augmented reality systems.In Proc.of UbiComp,pages1–9,2016.

[28]Lei Yang,Yong Qi,Jinsong Han,Cheng Wang,and Yunhao Liu.Shelving interference and joint identification in large-scale RFID systems.IEEE Transactions on Parallel and Distributed Systems,26(11):3149–3159,2015.

[29]Shigeng Zhang,Xuan Liu,Jianxin Wang,Jiannong Cao,and Geyong Min.Energy-efficient active tag searching in large scale RFID systems.Information Sciences,317:143–156,2015.

[30]Yuanqing Zheng and Mo Li.P-MTI:Physical-layer Missing Tag Identification via compressive sensing.In Proc.of Infocom,pages 917–925,2013.

[31]Zongheng Zhou,Himanshu Gupta,Samir R.Das,and Xianjin Zhu.Slotted scheduled tag access in multi-reader RFID systems.In Proc.of ICNP,pages 61– 70,2007.

[32]Yanmin Zhu,Wenchao Jiang,Qian Zhang,and Haibing Guan.Energy- Efficient Identification in Large-Scale RFID Systems with Handheld Reader.IEEE Transactions on Parallel and Distributed Systens,25(5):1211–1222, 2014.

Claims (3)

1. a kind of many reader RFID system fast tag recognition methods divided based on time slot, it is characterised in that including following Step：

Then etc. 1) first stage, when frame starts, reader group simultaneously gives one querying command of tag broadcast, and label to be received is returned Multiple, the label of certain reader inquiry is successfully received according to the parameter included in querying command, select to answer using hash function Time slot is answered, then the RN for carrying is returned to reader in its time slot for selecting by the label, in a slot, if reader success RN is received, the reader records this RN, other are confiscated the reader of RN or detect the reader unification of signal conflict Wait for a period of time, then one QueryRep order of reader group synchronized transmission starts a new time slot；

2) second stage, in each time slot, in reader group, all readers are sent comprising the RN being previously respectively received ACK, the label with identical RN sends the ID of the label to reader.

2. many reader RFID system fast tag recognition methods divided based on time slot according to claim 1, which is special Levy and be, the reader group selection process is：

1) initialization enlivens reader set A for sky；

2) when new reader is added, the subregion si of unlabelled maximum weight is found, and this sub-regions is arranged to Marked state；

3) all readers to associating above-mentioned subregion are tested in turn, if certain reader is added to enlivening reader In collection A, the reading efficiency value of reader set is enlivened and for E (rj)=eff ({ rj } ∪ A)；Eff ({ rj }) is The reading efficiency of reader rj；For the reader group of subregion si can be covered；

4) pick out and system effectiveness can be caused to increase maximum reader rk so that enliven the reader efficiency of reader set A Increase, and the reader is added to A；

5) all subregions that rk is covered are arranged to flag state；

6) repeat step 1)～5), until adding without new reader.

3. reader-tag recognition methods according to claim 1, it is characterised in that the length of the frame is：

$f_1^{*}=\arg {\min }_{f_1}T=\arg {\min }_{f_1}\frac{f_1\&times;t_{rn}+Max\left\{n_{r_i}\right\}\&times;t_{id}}{\&Sigma;n_{r_i}}.;$

Wherein, f₁Frame length for the first stage；n_riFor the ID numbers that reader ri is collected；t_rnFor RN collection phases each time slots Length；t_idLength for each time slot of ID second stage.