CN104657693A - RFID (radio frequency identification) anti-collision method based on GASS (grouped adaptive allocating slots) - Google Patents

Abstract

An RFID (radio frequency identification) anti-collision method based on GASS (grouped adaptive allocating slots) comprises steps as follows: firstly, a reader performs scanning statistics on slots randomly selected by tags and sends the slots to all the tags, then the tags perform corresponding slot adjustment, the reader skips free slots and collision slots and allocates effective slots adaptively, the tags are quickly identified, and when the number of unidentified tags is larger, the algorithm adopts strategies such as grouping, dynamic frame length adjustment and the like to shorten slot processing time. Simulation results indicate that with the adoption of the method, the identification efficiency and stability of a system are improved, the transmission cost is reduced, particularly when the number of the tags exceeds 1,000, the throughput of the algorithm is still kept higher than 71%, the efficiency of the system is 300% and 97.2% higher than that of a system adopting a traditional frame-slotted ALOHA-256 algorithm and that of a system adopting a grouped dynamic frame-slotted ALOHA algorithm respectively, and the method has practical application values for rapid identification of Internet-of-Things tags.

Description

A kind of RFID anti-collision method based on grouping self-adjusted block time slot

Technical field

The invention belongs to the many tag reader technology in technical field of RFID, relate to the method for multiple labels anti-collision.

Background technology

Radio-frequency (RF) identification (Radio Frequency Identification, RFID) be that one utilizes Electromagnetic Wave Propagation mode to carry out noncontact bidirectional data transfers between reader and label, and then obtain the recognition technology of identified object information, be acknowledged as the new and high technology of the most with prospects of 21 century and the power of change.This technology has that exchanges data is fast, tracking objects in time, without space constraint, penetration capacity by force, the advantage such as multi-targets recognition and antipollution, have a wide range of applications in industries such as logistics management, communications and transportation, automated production, public information service, significantly can improve management and operational paradigm, reduce costs.

Rfid system is made up of electronic tag (Tag), reader (Reader) and back-end data base (Database) three parts.In the radio-frequency recognition system of multiple reader and multiple label, there is the collision of two kinds of forms, i.e. reader collision and tag-collision.Probability due to reader collision generation is less and the processing power of reader itself is comparatively strong, and therefore reader collision problem is easier to solve.Scholar both domestic and external carries out with reader device at many labels having done large quantifier elimination in the collision problem that communicates simultaneously, and these methods are generally divided into four classes: space division multiplexing method, frequency multiplexing method, Code Division Multiplex method and time-division multiplex method.Due to features such as the low-power consumption of label, low storage capacity and limited computing powers, label anti-collision method mainly adopts time-division multiplex method.

In time-division multiplex method, multi-label anti-collision algorithm the most frequently used is at present divided into two kinds, be the deterministic type algorithm based on binary tree, its main representative algorithm has binary search algorithm, Dynamic binary searching algorithm, great-jump-forward dynamic tree algorithm, query tree algorithm etc.Such method selects label to communicate by reader according to the uniqueness of label ID, so the performance of searching algorithm sharply can worsen with the increase of ID figure place.Another kind is the statistical algorithm based on ALOHA.Its main algorithm comprises CDMA slotted ALOHA algorithm, Frame Slotted Aloha (Frame-slottedALOHA, FSA-256) algorithm, dynamic frame CDMA slotted ALOHA (Dynamic Frame-slotted ALOHA, DFSA) algorithm, also has EPC Class-1Generation-2 (EPC) standard that EPCglobal proposes in addition, it adopts a kind of random frame time slot algorithm based on Q value, and have document proposition in the recent period based on ALOHA anticollision (Collision Prediction-ALOHA, the CP-ALOAH) algorithm etc. of prediction of collision.These anti-collision algorithm implementation procedures are relatively simple, and inside tags does not need to design complicated circuit yet, and therefore the cost of label is also lower.But along with the number of label increases, time even thousands of, the probability collided also increases, and the performance of system identification sharply declines thereupon.Be directed to the identification problem of a large amount of label, researchers propose again the concept of dividing into groups, and have the dynamic frame time slot algorithm of enhancing, packet dynamic frame CDMA slotted ALOHA (Grouped Dynamic Frame-slotted ALOHA, GDFSA) algorithm etc.Compared with algorithm before, the anti-collision of classes of packets its gulp down Performance comparision stablize, but the same with other algorithms, its throughput is all lower, only has about 40% ~ 50%.

Summary of the invention

The present invention is in order to solve the multiple labels anti-collision problem in radio-frequency recognition system, on the basis of analysis frame CDMA slotted ALOHA algorithm, a kind of RFID anti-collision algorithm (Grouped AdaptiveAllocating Slots, GAAS) based on grouping self-adjusted block time slot is proposed.First reader is allowed to carry out scan statistics to the random selected time slot of label, and sending it to each label, label carries out correspondingly time slot adjustment again, makes reader skip free timeslot and collision time slot, allocative efficiency time slot adaptively, and then label is identified fast.When unidentified number of tags is larger, algorithm adopts the strategies such as grouping and dynamic conditioning frame length, to reduce the time of time slot process.Simulation result shows: GAAS algorithm improves recognition efficiency and the stability of system, reduces transport overhead.Particularly when number of tags is more than 1000, the throughput of this algorithm still remains on more than 71%, improve 300% and 97.2% respectively than the system effectiveness of traditional Frame Slotted Aloha-256 algorithm and packet dynamic frame CDMA slotted ALOHA algorithm, there is certain Theory and applications and be worth.

1, basic definition

Definition 1: set the timeslot number in a frame, namely frame length is L, when identifying n label, each label can send oneself information of identification code by Stochastic choice time slot, according to binomial distribution theorem, the probability that certain label takies time slot arbitrarily in frame is p=1/L, then have the probability of r label to be expressed as in same time slot:

$P(L,n,r)=C_n^r\×{\left(\frac{1}{L}\right)}^r\×{(1-\frac{1}{L})}^{n-r}---\left(1\right)$

Work as r=0, namely do not ask identification label in a time slot, this time slot is called free time (idle) time slot, and its probability is:

$P_i=P(L,n,0)={(1-\frac{1}{L})}^n---\left(2\right)$

Work as r=1, namely only have a label request identification label in a time slot, this time slot is called successfully (success) time slot, and its probability is:

$P_s=P(L,n,1)=\frac{n}{L}\×{(1-\frac{1}{L})}^{n-1}---\left(3\right)$

When r >=2, namely have two and above label request identification label in a time slot, this time slot is called collision (collision) time slot, and its probability is:

P _c＝1-P _s-P _i(4)

Then after the recognition cycle of a frame, there is no the expectation value a of the timeslot number of label ₀ ^l,n, only have the expectation value a of the timeslot number of a label _s ^l,n, produce the expectation value a of the timeslot number of collision _c ^l,nbe respectively:

${a_0}^{L,n}=L\×P_i=L{(1-\frac{1}{L})}^n---\left(5\right)$

${a_s}^{L,n}=L\×P_s={n(1-\frac{1}{L})}^{n-1}---\left(6\right)$

a _c ^L,n＝L×P _c＝L-a ₀ ^L,n-a _s ^L,n(7)

The throughput S of definition 2:RFID system refers to that reader identifies the ratio shared by the number of tags of Successful transmissions information in the time of frame length at one, that is:

2, the inventive method describes

2.1 rfid system optimal frame length are analyzed

After the number of tags of reader in its readable range is estimated, need to adjust frame length dynamically according to number of tags, if frame length value will produce a large amount of free timeslots too greatly, otherwise collision time slot can be caused sharply to rise, finally all by influential system recognition efficiency.Therefore want to obtain higher throughput efficiency, corresponding relation between frame length and number of tags must be found out, namely determine optimal frame length.To (8) formula differentiate:

$\frac{\mathrm{dS}}{\mathrm{dn}}=\frac{1}{L}{(1-\frac{1}{L})}^{n-1}+\frac{n}{L}{(1-\frac{1}{L})}^{n-1}\mathrm{In}(1-\frac{1}{L})---\left(9\right)$

Make (9) formula equal 0, the number n reaching frame length L and label should meet:

$n=-\frac{1}{\mathrm{In}(1-1/L)}---\left(10\right)$

Obtained by Taylor series expansion again

$L\≈\frac{1+1/n}{1+1/n-1}=n+1---\left(11\right)$

According to (8) formula, adjacent fixing frame length L ₁and L ₂the number of label at throughput intersections of complex curve place, be the critical point of adjustment frame length.

$\frac{n}{L_1}\×{(1-\frac{1}{L_1})}^{n-1}=\frac{n}{L_2}\×{(1-\frac{1}{L_2})}^{n-1}---\left(13\right)$

Wherein represent downward rounding operation, thus obtain the corresponding relation of number of tags n and frame length L, as shown in table 1.Determined the size of frame length by the span of number of tags, when number of tags is greater than 354, still identify respectively with maximum frame length 256.

Table 1 frame length and number of tags corresponding relation

Frame length	8	16	32	64	128	256	256
								The number of minimum label	1	12	23	45	89	178	355
The number of maximum label	11	22	44	88	177	354	≥355

2.2 labeled packet principles

Owing to being subject to the restriction of label cost, frame slot number ad infinitum can not be increased along with the increase of number of tags, so be directed to the situation of a large amount of labels, only having the number of the label by limiting every secondary response, just can make the throughput that system keeps relatively high.According to (8) formula, choose the critical value that label is divided into groups, i.e. the performance curve intersection point P of two consecutive frames _a=P _bthe label numerical value at place.

$\frac{n}{256\×a}\×{(1-\frac{1}{256})}^{\frac{n}{a}-1}=\frac{n}{256\×b}\×{(1-\frac{1}{256})}^{\frac{n}{b}-1}---\left(15\right)$

Wherein, a, b are the adjacent packets number of label, and such as, when getting a=1, b=2 substitutes into (15) Shi Ke get:

$\frac{n}{256\×1}\×{(1-\frac{1}{256})}^{\frac{n}{1}-1}=\frac{n}{256\×2}\×{(1-\frac{1}{256})}^{\frac{n}{2}-1}---\left(16\right)$

$\⇒n=2\frac{\mathrm{In}\frac{1}{2}}{\mathrm{In}\frac{255}{256}}\≈354---\left(17\right)$

Namely n=354 is critical value label being divided into a group or two groups, and in order to the throughput efficiency making system keep higher, unidentified number of tags n should not be greater than 354, when n is greater than 354, needs to divide into groups to Unidentified label.The packet count of number of labels within 2283 can be calculated by (15) formula, as shown in table 2.

The corresponding relation of table 2 total number of labels and packet count

Packet count	1	2	4	6	8	……
							Minimum number of tags	1	355	710	1246	1767	……
Maximum number of tags	354	709	1245	1766	2283	……

Packet count 1,2,4,6 in table 2,8 is autonomous Design, and the minimum number of tags in each group and maximum number of tags are by a, b numerical evaluation threshold value out in (15) formula.Along with the increase of number of tags, packet count also constantly increases.

2.3 the inventive method are as follows concrete:

(S01) the number n of label to be identified: before carrying out digital independent, is first estimated with Vogt algorithm; When number of tags is less than 354, adopt dynamic frame time slot strategy, the length M of dynamic conditioning identification frame, the processing stage of directly entering time slot; When n is greater than 354, then needs to divide into groups to label, try to achieve packet count g by table 2;

(S02): label is Stochastic choice one number i between 1 to g, as the group # of oneself, the value of s [t] is increased by 1 simultaneously, record the number of tags of this group; When identifying first, the group # t=1 of the current identification of initialization, starts to identify t group;

(S03): then carry out time slot scanning, reader sends Query (M) order to each label with the form of broadcast; After label receives this order, then return the timeslot number preengage separately to reader.Reader, again according to received data message, judges whether each time slot can successfully identify; If the successful identification label of energy, just corresponding time slot sign position Flag is set to 0, Flag is just set to-1 by other situations;

(S04): reader, according to the situation of label institute reserving time slots, is passed through order Refresh_slot (Slots) and sends to each label, allow each label also can obtain the situation of other labels selection time slot, then adjust corresponding time slot; Element inside Slots is then the mark value recorded in the time slot scanning process above, and label adjusts corresponding selected timeslot number according to this array;

(S05): reader sends Collection () order, each label just sends data to reader according to the time slot after adjustment after receiving this order successively; Last reader sends S1eep () order, takes turns the label be correctly read in query script and will enter silent status, namely do not participate in ensuing inquiry at this; Meanwhile, the time slot counter of each label in this group is allowed to subtract 1;

(S06): after each takes turns end of identification, then the number of the unidentified label of estimation residue, if number of tags is not 0, then return (S01), until the label that this group is remaining has all been identified;

(S07): group # adds 1, i.e. t=t+1, then performs following one of two things:

1) if t≤g, then represent the group still existing and do not have to identify, proceed the identification of next group, return (S01);

2) if t > is g, namely all groups have all identified, end of identification.

The core of GAAS algorithm was exactly before carrying out tag recognition, first time slot scanning operation is carried out, record the situation of label institute reserving time slots, then in the tag recognition stage, reader is allowed to skip collision time slot and free timeslot, and direct allocative efficiency time slot, the label that success is preengage directly is identified, thus improves the utilization factor of time slot.Specific algorithm flow process as shown in Figure 1.

Feature of the present invention is:

(1) tag circuit analysis of complexity.

The random number supposing label random selecting time slot is R, then need position random number, because this algorithm takes the method for grouping, R maximal value is set to 256, only needs 8, so label only needs 8 random number generation circuits.And EPC anticollision protocol is except producing the random number choosing time slot and needing 4, also requires generation 16 RN16 random numbers, need altogether 16+4=20 position random number generation circuit.Therefore, this algorithm is starkly lower than EPC label to label design random number circuit complexity.In addition, also have the controller based on state machine to perform life in label, EPC label need perform primary commands 5 (Query, QueryAdjust, QueryRep, Ack, RN16), in agreement of the present invention, label also needs execution 5 (Query, Refresh-slot, Count, Collection, therefore the controller circuitry complicacy of GAAS protocol label is also suitable with EPC agreement Sleep).In sum, GAAS protocol label circuit design is simpler than EPC agreement, thus reduces the cost of label.

(2) transport overhead analysis.

Transport overhead is the important indicator of an assessment algorithm, and it comprises reader expense and label expense two parts.Respectively to self-adjusted block time slot (Adaptive Allocating Slots, AAS) transport overhead of algorithm in tag recognition process in algorithm, GAAS algorithm and EPC standard emulates, the GAAS order of using in emulation and parameter length as shown in table 3.

The order related in table 3 GAAS algorithm and parameter length

Order, parameter	Function	Length/bit
			Query	Adjustment frame length	8
Refresh-slot	Adjustment timeslot number	32
			Count	Slot count	8
Collection	Confirm communication	18
			Sleep	Enter silent status	2

1) expense of reader.

If number of labels is change in interval [0,2000], as shown in Figure 2, along with the increase of number of tags, the expense of reader also constantly increases the bit number of reader transmission.When the number of label is less than 1300, AAS algorithm is suitable with GAAS algorithm reader expense, and when the number of label is more than 1300, the advantage of GAAS algorithm starts to highlight.This is because label is not carried out packet transaction by AAS algorithm, along with the increase of number of tags, the probability of tag-collision sharply rises, and the instruction that reader sends also can increase thereupon.When number of tags is 2000, in AAS algorithm, reader bit number is that 66790bit, EPC canonical algorithm is respectively 60527bit, and GAAS algorithm is 54245bit, have dropped 18.8% than the expense of AAS algorithm, have dropped 10.4% than ECP canonical algorithm.

2) expense of label.

Fig. 3 shows the expense of three kinds of algorithm labels, and along with the increase of number of tags, in AAS algorithm, label transmits bit number close to exponential increase, and GAAS algorithm and EPC canonical algorithm approximately linear increase, and wherein GAAS increases the slowest.When the number of label is 2000, in AAS algorithm, reader bit number is that 422135bit, EPC canonical algorithm is respectively 121545bit, and GAAS algorithm is only 39601bit, have dropped 96.1% than the expense of AAS algorithm label, have dropped 67.4% than ECP canonical algorithm.

(3) total timeslot number analysis.

Total time slot is a key factor of decision systems efficiency, and total timeslot number is fewer, and the performance of system is better.From upper surface analysis, whole algorithm reads data period can be divided into two stages: time slot scanning stage and tag recognition stage, so inquiring about total time slot is also these two discrete consuming time slot sums.

Emulate total timeslot number of FSA-256 algorithm, DFSA algorithm, GDFSA algorithm and GAAS algorithm, simulation result as shown in Figure 4.Number of labels changes to the process of 1500 from 0, FSA-256 algorithm and the total timeslot number required for DFSA algorithm exponentially increase with the increase of number of labels, GDFSA algorithm and GAAS algorithm linearly increase, wherein FSA-256 algorithm growth rate is the fastest, and GAAS algorithm is the slowest, it is secondly GDFSA algorithm.Particularly when number of tags is larger, the advantage of GAAS algorithm is more obvious.When number of tags is 1000, GAAS algorithm only needs about 1400 time slots, reduces about 4165 than FSA-256, reduces about 3727 than DFSA, reduces about 1366 than GDFS.

(4) throughput analysis.

System throughput is also the important indicator weighing system performance.As can be seen from Figure 5, when label is less than 354, GDFSA and DFSA throughput is identical, and FSA-256 algorithm is minimum, only has about 0.2, and GAAS algorithm is the highest, can reach more than 0.7.GDFSA, DFSA, GAAS algorithm can according to physical tags number, and allocative efficiency time slot identifies adaptively, and FSA-256 algorithm adopts fixing frame length 256.When number of tags is greater than 354, the throughput of FSA-256, DFSA algorithm all sharply declines, and label is divided into many groups by GDFSA, GAAS algorithm, is identified each group of label by dynamic conditioning frame length, makes throughput stable within the specific limits.The system throughput of algorithm of the present invention is obviously more much bigger than other three kinds of algorithms, and the throughput of FSA-256 algorithm is between 0.1 ~ 0.25, and DFSA algorithm is between 0.2-0.36, and GDFSA algorithm only maintains about 0.36, and GAAS algorithm is all higher than them.When the number of label reaches 1000, compare with FSA-256 with GDFSA, the system effectiveness of GAAS algorithm improves 300% and 97.2% respectively.

The present invention proposes self-adjusted block time slot (AAS) method, along with the increase of number of tags to be identified, add again the concept of grouping, namely divide into groups self-adjusted block time slot (GAAS) algorithm.This algorithm combines the feature of traditional GDFSA and CP-ALOHA algorithm, their basis is optimized and improves.First estimate number of labels, then adopt the strategies such as grouping, dynamic conditioning frame length, time slot reservation and self-adjusted block time slot to identify fast label.

Accompanying drawing explanation

Fig. 1 is that the present invention proposes GAAS algorithm flow chart.

Fig. 2 is that the expense of the present invention and other algorithm readers contrasts.

Fig. 3 is that the expense of the present invention and other algorithm labels contrasts.

Fig. 4 the present invention and the total timeslot number of other algorithms are with the situation of change of number of tags.

Fig. 5 the present invention compares with the throughput of other algorithms.

Embodiment

The present invention includes is that number of tags is estimated and the grouping stage, the processing stage of time slot, and tag recognition stage three phases, detailed process is as follows:

(1) number of tags is estimated and the grouping stage

I, in the identification incipient stage, estimates the number n of label to be identified with Vogt algorithm;

Ii is when number of tags is less than 354, and adopt dynamic frame time slot strategy, the length M of dynamic conditioning identification frame, the processing stage of directly entering time slot; When n is greater than 354, then needs to divide into groups to label, try to achieve packet count g by table 2;

Iii label is Stochastic choice one number i between 1 to g, as the group # of oneself, the value of s [t] is increased by 1 simultaneously, records the number of tags of this group;

The group # t=1 of the current identification of iv initialization, starts to identify t group.

(2) processing stage of time slot

I, before carrying out digital independent, first carries out time slot scanning, and reader sends Query (M) order to each label with the form of broadcast; After label receives this order, then return the timeslot number preengage separately to reader;

Ii reader, again according to received data message, judges which time slot can successfully identify, which time slot will produce collision or free timeslot; If the successful identification label of energy, just corresponding time slot sign position Flag is set to 0, Flag is just set to-1 by other situations;

Iii reader, according to the situation of label institute reserving time slots, is passed through order Refresh_slot (Slots) and is sent to each label, and label also can be known, and the situation of time slot selected by other labels, then adjusts corresponding time slot; Element inside Slots is then the mark value recorded in the time slot scanning process above, and label adjusts corresponding selected timeslot number according to this array.

(3) the tag recognition stage

I to be aware of the selection situation of each time slot, at the direct allocative efficiency time slot of cognitive phase due to reader in a upper stage; Following reader sends Collection () order, and label just sends data to reader according to the time slot after adjustment after receiving this order successively;

Ii reader sends Count () order, allows the time slot counter of each label in this group subtract 1;

The last reader of iii sends S1eep () order, takes turns the label be correctly read in query script and will enter silent status, namely do not participate in ensuing inquiry at this;

Iv is after each takes turns end of identification, then the number of the unidentified label of estimation residue, if number of tags is not 0, then above repeating (2), (3) two stages, until all identified the label that this group is remaining;

V group # adds 1, i.e. t=t+1, then performs following one of two things:

1) if t≤g, then represent the group still existing and do not have to identify, proceed the identification of next group;

2) if t > is g, namely all groups have all identified, end of identification.

The present invention proposes a kind of grouping self-adjusted block time slot RFID anticollision (GAAS) algorithm, by identifying fast label strategies such as the estimation of number of labels and grouping, time slot reservation and self-adjusted block time slots.Simulation result shows, along with the continuous increase of number of labels, particularly when the number of label is more than 1000, the throughput of GAAS algorithm maintains more than 0.71, than the throughput raising all by a relatively large margin based on the traditional algorithm of ALOHA, the total number of timeslots required for whole identifying and transport overhead almost maintain linear increase.Effectively can improve the work efficiency of rfid system, add the stability of system throughput, reduce the cost of label.Be directed to the identification of a large amount of labels, the advantage of GAAS algorithm is particularly remarkable, has broad application prospects.

Claims (2)

1., based on a RFID anti-collision method for grouping self-adjusted block time slot, it is characterized in that according to the following steps:

(S01): before carrying out digital independent, first estimate the number n of label to be identified with Vogt algorithm and carry out packet transaction; When identifying first, the group # t=1 of the current identification of initialization, starts to identify t group;

(S02): then carry out time slot scanning, reader sends Query (M) order to each label with the form of broadcast; After label receives this order, then return the timeslot number preengage separately to reader; Reader, according to received data message, judges whether each time slot can successfully identify; If the successful identification label of energy, just corresponding time slot sign position Flag is set to 0, Flag is just set to-1 by other situations;

(S03): reader, according to the situation of label institute reserving time slots, is passed through order Refresh_slot (Slots) and sends to each label, allow each label also can obtain the situation of other labels selection time slot, then adjust corresponding time slot; Element inside Slots is then the mark value recorded in the time slot scanning process above, and label adjusts corresponding selected timeslot number according to this array;

(S04): reader sends Collection () order, each label just sends data to reader according to the time slot after adjustment after receiving this order successively; Last reader sends S1eep () order, takes turns the label be correctly read in query script and will enter silent status, namely do not participate in ensuing inquiry at this; Meanwhile, the time slot counter of each label in this group is allowed to subtract 1;

(S05): after each takes turns end of identification, then the number of the unidentified label of estimation residue, if number of tags is not 0, then return (S01), until the label that this group is remaining has all been identified;

(S06): group # adds 1, i.e. t=t+1, then performs following one of two things:

(1) if t≤g, then represent the group still existing and do not have to identify, proceed the identification of next group, return (S01);

(2) if t > is g, namely all groups have all identified, end of identification.

2. the RFID anti-collision algorithm based on grouping self-adjusted block time slot according to claim 1, is characterized in that the packet transaction of step (S01), according to the following steps:

(1) when number of tags is less than 354, then adopt dynamic frame time slot strategy, the length M of dynamic conditioning identification frame, the processing stage of directly entering time slot; When n is greater than 354, then needs to divide into groups to label, try to achieve packet count g;

(2) label Stochastic choice one number i between 1 to g, as the group # of oneself, increases by 1 the value of s [t] simultaneously, records the number of tags of this group.