CN102508640B - Distributed radio frequency identification device (RFID) complex event detection method based on task decomposition - Google Patents

Abstract

The invention discloses a distributed radio frequency identification device (RFID) complex event detection method based on task decomposition. Firstly a complex pattern is decomposed to be a plurality of simple subtasks which are parallelly processed by a plurality of machines to reduce event rate of a single panel point and improve integral throughput; and secondly the distributed RFID complex event detection method provides the complex event detection method based on bitmap indexing when the subtasks are processed to improve detection efficiency.

Description

The distributed RFID complex events detecting methods that task based access control decomposes

Technical field

The present invention relates to a kind of distributed RFID complex events detecting methods.Be specially and complex task is decomposed into some simple subtasks, by multiple nodal parallel process, thus make subtask load lower than processing node upper loading limit, eliminate congested.

Background technology

RF identification (RFID, Radio Frequency Identification) be a kind of non-contact automatic identification technology (AIT, Auto Identification Technology), use radio frequency electromagnetic between reader and labeled mobile article, transmit data and reach the objects such as identification, tracking.

For tracking and the analysis of special article etc., need to integrate multiple data flow physically distributed.But the value that the limited initial data of these carry informations is too not large, what people paid close attention to is the high level data with semantic information.Complicated event detection is a kind of technology of special disposal interevent relation, and it performs data matching operation according to user-defined pattern in multiple data flow, according to rule, some simple atomic events can be aggregated into the complicated event of physical significance.Therefore RFID complicated event detection technique is a key technology of disposing rfid system.Pattern is the formalized description to complicated event, and define the atomic event and constraints that form complicated event, a pattern is a task.Macroscopically, complicated event detects using some atomic event stream as input, exports a complicated event stream by check processing; On microcosmic, once concrete complicated event detects and can be described as following process: the data 1) in system buffer one time window; 2) each is organized new data e and arrives and trigger one-time detection action, and system searches other related data in buffered data according to the constraints that pattern defines from e; 3) if find one group of complicated event, then export; 4) by new data data inserting buffering area.

Current, the method detected for RFID complicated event mainly contains: the complicated event 1) based on finite state machine detects; 2) complicated event based on Petri network detects; 3) complicated event based on Match Tree detects.The algorithm that above method detects at complicated event has their own characteristics each, but all adopts centralized solution in framework aspect.System cloud gray model is on a main frame, and same machine completes for the input of whole data flow, detection and output function.But the operational capability of individual machine is limited, will be absorbed in congested when high-speed data-flow.On the other hand, when centralized solution after congested generation is difficult to carry out system extension, can only upgrading hardware system, this just inevitably causes testing process to interrupt.

Along with the development of RFID technique, its application scenarios is just experiencing by the simple transformation to complexity; Globalization application is moved towards from topical application; Combine with backstage single application system from RFID, change into and realize data sharing etc. under isomerous environment, this large-scale RFID application will produce the real-time RFID data stream of flood tide.Under the data flow condition of this flood tide, above-mentioned centralized processing scheme can produce communication at central processing node and process congested, causes the sharply increase of processing delay.Complicated event detects a link as RFID application, and it receives the data of bottom input and provides testing result to upper strata.The process of this link is congested by the popularization of greatly limit RFI D application.

Summary of the invention

In order to overcome the congestion problems that prior art produces on reply high-speed data-flow, the invention provides a kind of distributed RFID complex events detecting methods decomposed based on pattern: first, complicated pattern is decomposed some simple subtasks, by the parallel processing of multiple stage machine, thus reduce the event rate of individual node, improve entire throughput; Secondly, the present invention proposes a kind of complex events detecting methods based on bitmap index when processing subtask, improve detection efficiency.

The technical solution adopted for the present invention to solve the technical problems comprises following content:

1. complex task is decomposed into the subtask that a group meets processing node load restriction, and all subtasks are organized into a Task Tree.

The present invention takes a kind of iterative strategy to decompose complex task and structure Task Tree, and concrete steps are as follows:

1) the upper loading limit rateT for given processing node selects several stream, together with the constraints between them as subtask from candidate events stream (being the atomic event stream complex task during first time iteration).

2) from Candidate Set, delete selected flow of event, and judge whether Candidate Set is empty.If be empty, then iteration terminates; Otherwise perform step 3).

3) intermediate object program exported is detected in subtask and be defined as empty flow of event, calculate subtask and detect in the speed of empty flow of event that exports and it and Candidate Set and remain matching rate between flow of event, and empty flow of event is added Candidate Set.Then step 1 is forwarded to), start for another processing node new round process.

Represent the number of candidate events stream with N, represent that the node load upper limit is for j with m (i, j), available maximum compression ratio when candidate events stream is [1-i]; N dimensional vector X represents selection scheme, and i-th of X is that 1 expression candidate events stream i is selected, and it is selected to be that 0 expression does not have, and X (i, j) represents that to set up corresponding selection with M (i, j) vectorial.Above-mentioned steps 1) in select the method for flow of event to be described below:

A) to the speed rate of rateT and each candidate events stream _i, carry out equal proportion scaling and the process that rounds up, they are become integer, and rounding error controls within 5%, and i represents that candidate events stream is numbered, rate _irepresent the speed being numbered the candidate data stream of i.

B) to the candidate events stream being numbered 1, represent the load capacity of processing node with j, investigate each j from 1 to rateT: as j < rate ₁time, (1, (1, j) the 1st is 0 in j)=0, X to put m; As j>=rate ₁time, (1, (1, j) the 1st is 1 in j)=1, X to put m.

C) to the candidate events stream being numbered i, if i > is N, then step f is forwarded to); Otherwise each j from 1 to rateT is investigated to i: as j < rate _itime, put m (i, j)=m (i-1, j), X (i, j)=X (i-1, j); As j>=rate _itime forward steps d to).

D) at selection scheme X (i, j-rate _i) in add i-th candidate events stream, and calculate its compression ratio compratio, then compare compratio and m (i-1, j) size: if compratio is large, then m [i] [j]=compratio, puts X (i, j)=X (i-1, j-rate _i), X (i, j) i-th is 1, otherwise puts m [i] [j]=m [i-1] [j], X (i, j)=X (i-1, j).

E) after all j having been investigated to flow of event i, investigate flow of event i+1, j is set to 0, forwards step c to).

F) selection course terminates, vectorial X (N, rateT) (this variable and X (i, j-rate _i) implication identical, bracket inside be all subscript, namely i value is N, j value is rateT; M (N, rateT) is hereafter in like manner) be selection scheme, corresponding compression ratio is m (N, rateT).

Above-mentioned steps d) in the ratio of the speed ratef of empty flow of event that exports with its detection of compression ratio compratio total speed of being defined as the incoming event stream of subtask, $\mathrm{ratef}={\mathrm{rate}}_1*\underset{k=1}{\overset{M-1}{\mathrm{\&Pi;}}}{\mathrm{\&eta;}}_{(k,k+1)},$ $\mathrm{compratio}=\frac{\underset{k=1}{\overset{M}{\mathrm{\&Sigma;}}}{\mathrm{rate}}_k}{\mathrm{ratef}},$ Wherein M represents the number of selected candidate events stream, rate ₁represent the speed of first selected flow of event, η represents matching rate, η _{(k, k+1)}represent the matching rate between adjacent two flows of event.

Above-mentioned steps 3) in remain the computational methods of matching rate between flow of event in empty flow of event and Candidate Set as follows: the numbering representing selected flow of event with s, the number of selected flow of event is M; The numbering of not selected flow of event is represented with h; Empty flow of event is represented with f; The then matching rate of h and f the matching rate of f and h ${\mathrm{\&eta;}}_{f,h}=\frac{{\mathrm{\&eta;}}_{h,f}*\mathrm{rateOut}}{{\mathrm{rate}}_h}.$

2. adopt the complicated event detection algorithm process subtask based on bitmap index.

Complicated event detection algorithm based on bitmap index on the basis of streamjoin algorithm, adds bitmap index to preserve intermediateness, improves detection efficiency.If subtask comprises P incoming event stream, be then specifically detected title and comprise following steps:

1) buffer container of P B+ tree as incoming event stream is created; The event entity being different event stream by matching rate ascending order specifies looked-up sequence.

2) one-time detection action is triggered when each event entity e arrives:

A) for e sets up P position bitmap, by that position 1 of the numbering correspondence of flow of event belonging to it;

B) in buffer container, the event entity of coupling is searched by looked-up sequence.For arbitrary container, search the event entity mated with e in this embodiment, if found, perform step c), otherwise perform steps d);

C) merge the bitmap of e and the event entity that matches with it, and with their respective bitmaps of amalgamation result renewal, then between these two event entities, set up pointer.Judge after merging, whether all positions of bitmap are 1, if so, then represent and one group of result detected, Output rusults, perform steps d); Otherwise forward step b to), search in subsequent vessel.

D) event e is inserted corresponding container together with its bitmap, terminate this and detect, wait for new events entity.

3) Delete Expired event entity and bitmap thereof from buffer container, and by the relevant position 0 of the bitmap of the not out of date event entity with this event Entities Matching.

The invention has the beneficial effects as follows: the present invention considers the magnanimity of RFID data, concurrency and distributivity feature comparatively all sidedly, complex task is decomposed into some simple subtasks, by multiple processing node parallel detection, making subtask load lower than the upper loading limit of processing node, efficiently solving the congestion problems of centralized approach when processing high-speed data-flow; On the other hand, make the compression ratio of subtask maximum, thus controlled processing node less to greatest extent.Meanwhile, the present invention introduces bitmap index in streamjoin algorithm, improves the efficiency of subtask process.

Below in conjunction with accompanying drawing and example, the present invention is further described.

Accompanying drawing explanation

Fig. 1 is the schematic diagram of the process constructing tree-like task structure iteratively, and wherein solid line node represents the atomic event stream related in pattern, and broken circle indicates the subtask that child nodes polymerization generates and detects the empty flow of event exported;

Fig. 2 is the flow chart of the complicated event detection algorithm based on bitmap index.

Detailed description of the invention

The method that the complex task that the present invention proposes decomposes and Task Tree constructs is actually a kind of iteration aggregation strategy: the atomic event stream comprised using complex task is as initial Candidate Set, therefrom select some, aggregate into a subtask together with the constraints between them; Then this task is detected the empty flow of event of output (in order to distinguish with atomic event stream and complicated event stream, using subtask detect export intermediate object program be defined as empty flow of event) and remaining atomic event stream together as new Candidate Set, therefrom select some flows of event, aggregate into new subtask together with the restriction relation between them; Repeat above-mentioned polymerization process until only remain a stream in Candidate Set.Above-mentioned iterative process is logically formed as one tree, i.e. tree-like task structure.Wherein, leaf node represents the atomic event stream that complex task comprises; Non-leaf nodes has double implication: a) for its child nodes, it represents subtask, and the flow of event input subtask representated by child nodes processes; B) for its father node, it represents virtual events stream.Subtask representated by root node is detected and is exported target complicated event stream.For given RFID complicated event pattern (And<A, B, C, D, E>), the overall flow of the method for the invention is described.

1. complex task is decomposed into the subtask that a group meets processing node load restriction, and all subtasks are organized into a Task Tree.

Suppose: after standardization, the disposal ability of processing node is 1.8, and the speed of atomic event stream is followed successively by 1.0, and 0.7,0.8,0.6,0.6, matching rate between them is as shown in table 1, and in table, "/" represents that two streams do not have direct matching relationship, does not include calculating in.

Table 1 atomic event matching rate table

	A	B	C	D	E
						A	1	0.45	/	/	/
B	0.57	1	0.5	/	/
						C	/	0.44	1	0.6	/
D	/	/	0.8	/	0.82
						E	/	/	/	0.82	1

The process of grey iterative generation subtask is as follows:

Iteration for the first time: comprise flow of event <A in Candidate Set, B, C, D, E>, its numbering is followed successively by <1,2,3,4,5>, the speed of each flow of event is followed successively by <10,7 through scaling with after rounding off, 8,6,6, the value of >, rateT is the value of 18, N is 5.The implementation of flow of event selection algorithm is as follows: (shown below is the running of a selection algorithm, exemplarily)

1) for the flow of event being numbered 1, rate ₁=10, so m (1,1)=m (1,2)=...=m (1,9)=0, X (1,1)=X (1,2)=...=X (1,9)=[00000]; M (1,10)=m (1,11)=...=m (1,18)=1, X (1,10)=X (1,11)=...=X (1,18)=[10000].

2) for the flow of event being numbered 2, rate ₂=7, so m (2,1)=m (1,1), m (2,2)=m (1,2) ... m (2,6)=m (1,6), X (2,1)=X (1,1), X (2,2)=X (1,2) ... X (2,6)=X (1,6); When 10 > j>=7, the value of compratio be 1, m (1, value j) is 0, so m (2, j)=compratio, X (and 2, j)=[01000]; When 17 > j>=10, the value of compratio be 1, m (1, value j) is 1, so m (2, j)=m (and 1, j), X (2, j)=X (1, j); When j>=17, the value of compratio be 3.78, m (1, value j) is 1, so m (2, j)=compratio, X (and 2, j)=[11000].

3) for the flow of event being numbered 3, rate ₃=8, so m (3,1)=m (2,1), m (3,2)=m (2,2) ... m (3,7)=m (2,7), X (3,1)=X (2,1), X (3,2)=X (2,2) ... X (3,7)=X (2,7); When 15 > j>=8, the value of compratio be 1, m (2, value j) is 1, so m (3, j)=m (and 2, j), X (3, j)=X (2, j); When 17 > j>=15, the value of compratio be 4.28, m (2, value j) is 1, so m (3, j)=compratio, X (and 3, j)=[01100]; When j>=17, the value of compratio be 4.28, m (2, value j) is 3.78, so m (3, j)=compratio, X (and 3, j)=[01100].

4) for the flow of event being numbered 4, rate ₄=6, so m (4,1)=m (3,1), m (4,2)=m (3,2) ... m (4,5)=m (3,5), X (4,1)=X (3,1), X (4,2)=X (3,2) ... X (4,5)=X (3,5); When 14 > j>=6, the value of compratio be 1, m (3, value j) is 1, so m (4, j)=m (and 3, j), X (4, j)=X (3, j); When 15 > j>=14, the value of compratio be 2.92, m (3, value j) is 1, so m (4, j)=compratio, X (and 4, j)=[00110]; When j>=15, the value of compratio be 2.92, m (3, value j) is 4.28, so m (4, j)=m (and 3, j), X (4, j)=X (3, j).

5) for the flow of event being numbered 5, rate ₅=6, so m (5,1)=m (4,1), m (5,2)=m (4,2) ... m (5,5)=m (4,5), X (5,1)=X (4,1), X (5,2)=X (4,2) ... X (5,5)=X (4,5); When 12 > j>=6, the value of compratio be 1, m (4, value j) is 1, so m (5, j)=m (and 4, j), X (5, j)=X (4, j); When 14 > j>=12, the value of compratio be 2.44, m (4, value j) is 1, so m (5, j)=compratio, X (and 5, j)=[00011]; When 15 > j>=14, the value of compratio be 2.44, m (4, value j) is 2.92, so m (5, j)=m (and 4, j), X (5, j)=X (4, j); When j>=15, the value of compratio be 2.44, m (4, value j) is 4.28, so m (5, j)=m (and 4, j), X (5, j)=X (4, j).

Screen through selection algorithm, obtaining result is: X (5,18)=[01100], using B, C as the maximum compression ratio 4.28 obtained during subtask lower than upper loading limit, this task actual loading 15, its rate estimation value detecting the empty flow of event F exported is 3.5, and the matching rate between other flow of event and F is as shown in table 2.

Table 2 atomic event matching rate table

	A	F	D	E
					A	1	0.225	/	/
F	0.642	1	0.801	/
					D	/	0.467	1	0.82
E	/	/	0.82	1

Second time iteration: comprise flow of event <A in Candidate Set, F, D, E>, its numbering is followed successively by <1,2,3,4>, the speed of each flow of event is followed successively by <100,35 through scaling with after rounding off, 60, the value of 60>, rateT is the value of 180, N is 4.Screen through selection algorithm, obtaining result is: X (4,180)=[0111], using F, D, E as the maximum compression ratio 6.74 obtained during subtask lower than upper loading limit, this task actual loading 155, its rate estimation value detecting the empty flow of event G exported is 22.9, and the matching rate between other flow of event and G is as shown in table 3.

Table 3 atomic event matching rate table

	A	G
			A	1	0.147
G	0.642	1

Iteration for the third time: comprise flow of event <A in Candidate Set, G>, its numbering is followed successively by <1,2>, the speed of each flow of event is followed successively by <100 through scaling with after rounding off, 22>, the upper loading limit of processing node is 180.Screen through selection algorithm, obtaining result is: X (2,180)=[11], using A, G as the maximum compression ratio 8.36 obtained during subtask lower than upper loading limit, this task actual loading 122, its rate estimation value detecting the empty flow of event H exported is 14.7.

When third time, iteration completed, Candidate Set is empty, and whole iterative process terminates.Fig. 1 demonstrates iterative process, and wherein B, C are at the 4th layer, composition subtask And<B, C>, and this subtask is detected and exported empty flow of event F; F, D and E belong to third layer, composition subtask And<F, D, E>, and this subtask is detected and exported empty flow of event G; G and A belongs to the second layer, composition subtask And<A, G>, detects and export target complicated event H behind this subtask.Three subtasks after decomposition are by three nodal parallel process.

2. adopt the complicated event detection algorithm based on bitmap index to carry out subtask process at processing node.

Processing node runs centralized detection algorithm process subtask.Can waste computational resource owing to carrying out expired events process frequently, thus the present invention adopts clocked flip mechanism to carry out expired process.With subtask And<F, D, E> are that example is described in detail:

(1) link is initialized: set up the buffer container of three B+ trees respectively as flow of event F, D and E; The bitmap of F, D, E is respectively [100], [010], [001], and the bitmap of son termination fruit is <111>; Be that three class atomic events set up detection sequence, be respectively: F → D → E, D → E → F and E → F → D; The clocked flip cycle is set; Start communication service and receive incoming event; Then detect thread and be absorbed in wait state.

(2) output element is detected: first it is also set up bitmap <010> for it stored in cell therefor when the entity of certain D class atomic event arrives.Then go to search the E class event matched according to detection sequence, suppose to find, then the bitmap of corresponding D, E event entity being merged (or operation) is <011>, and search corresponding match event along the buffer container that detection ordering proceeds to F, suppose not find, then this detects action termination; When the F event of mating finally arrives, detect sequence according to it to go to search the D event matched, this D bitmap is corrected for <011>, arrives dbjective state <111>, detect successfully after one query.

(3) expired processing links: after activation period arrives, the storage container that reverse scanning is all, Delete Expired atomic event, and the bitmap revising the atomic event matched.In the example of step (2), if E event is expired before F event arrives, then deletes E event, and the bitmap of D event is reduced to <010>.

Claims (1)

1. a distributed RFID complex events detecting methods for task based access control decomposition, is characterized in that comprising the steps:

The first step, is decomposed into the subtask that a group meets processing node load restriction, and all subtasks is organized into a Task Tree by complex task;

Take a kind of iterative strategy to decompose complex task and structure Task Tree, concrete steps are as follows:

1) the upper loading limit rateT for given processing node selects several to flow from candidate events stream, together with the constraints between them as subtask;

2) from Candidate Set, delete selected flow of event, and judge whether Candidate Set is empty, if be empty, then iteration terminates; Otherwise perform step 3);

3) flow of event exported is detected in subtask and be defined as empty flow of event, calculate subtask and detect in the speed of empty flow of event that exports and it and Candidate Set and remain matching rate between flow of event, and empty flow of event is added Candidate Set; Then the step 1 of the first step is forwarded to), start to carry out new round process for another processing node;

Second step, adopts the complicated event detection algorithm process subtask based on bitmap index, if subtask comprises P incoming event stream, then concrete testing process comprises following steps:

1) buffer container of P B+ tree as incoming event stream is created; The event entity being different event stream by matching rate ascending order specifies looked-up sequence;

2) trigger one-time detection action when each event entity e arrives, comprise following content:

A) for e sets up P position bitmap, by that position 1 of the numbering correspondence of flow of event belonging to it;

B) in buffer container, search the event entity of coupling by looked-up sequence, for arbitrary container, search the event entity mated with e in this embodiment, if found, perform step c), otherwise perform steps d);

C) bitmap of e and the event entity that matches with it is merged, and upgrade their respective bitmaps with amalgamation result, then between these two event entities, pointer is set up, judge after merging, whether all positions of bitmap are 1, if, then represent and one group of result detected, Output rusults, perform steps d); Otherwise forward step b to), search in subsequent vessel;

D) event entity e is inserted corresponding container together with its bitmap, terminate this and detect, wait for new events entity;

3) Delete Expired event entity and bitmap thereof from buffer container, and by the relevant position 0 of the bitmap of the not out of date event entity with this expired events Entities Matching;

Described first step step 2) first time iteration time candidate flow of event be atomic event stream in complex task;

Described first step step 1) number of candidate events stream is represented with N, represent that the node load upper limit is that j, candidate events stream are for available maximum compression ratio time [i-1] with m (i, j); N dimensional vector X represents selection scheme, and i-th of X is that 1 expression candidate events stream i is selected, and it is selected to be that 0 expression does not have, and X (i, j) represents that to set up corresponding selection with m (i, j) vectorial, and the method for selection flow of event comprises the following steps:

A) to the speed rate of rateT and each candidate events stream _icarry out equal proportion scaling and become integer with the process that rounds up, rounding error controls within 5%, and i represents that candidate events stream is numbered, rate _irepresent the speed being numbered the candidate events stream of i;

B) to the candidate events stream being numbered 1, represent the load capacity of processing node with j, investigate each j from 1 to rateT: as j < rate ₁time, (1, (1, j) the 1st is 0 in j)=0, X to put m; As j>=rate ₁time, (1, (1, j) the 1st is 1 in j)=1, X to put m;

C) to the candidate events stream being numbered i, if i > is N, then step f is forwarded to); Otherwise each j from 1 to rateT is investigated to i: as j < rate _itime, put m (i, j)=m (i-1, j), X (i, j)=X (i-1, j); As j>=rate _itime forward steps d to);

D) at selection scheme X (i, j-rate _i) in add i-th candidate events stream, and calculate its compression ratio compratio, then compare compratio and m (i-1, j) size: if compratio is large, then m [i] [j]=compratio, puts X (i, j)=X (i-1, j-rate _i), X (i, j) i-th is 1, otherwise puts m [i] [j]=m [i-1] [j], X (i, j)=X (i-1, j);

E) after all j having been investigated to flow of event i, investigate flow of event i+1, j is set to 0, forwards step c to);

F) selection course terminates, and vectorial X (N, rateT) is selection scheme, and corresponding compression ratio is m (N, rateT); Steps d in the method for described selection flow of event) in the ratio of the speed ratef of empty flow of event that exports with its detection of compression ratio compratio total speed of being defined as the incoming event stream of subtask, wherein M represents the number of selected candidate events stream, rate ₁represent the speed of first selected flow of event, η represents matching rate, η _{(k, k+1)}represent the matching rate between adjacent two flows of event;

Described first step step 3) in remain the computational methods of matching rate between flow of event in empty flow of event and Candidate Set as follows: the numbering representing selected flow of event with s, the number of selected flow of event is M; The numbering of not selected flow of event is represented with h; Empty flow of event is represented with f; The then matching rate of h and f ${\mathrm{\&eta;}}_{h,f}=\underset{s=1}{\overset{M}{\mathrm{\&Pi;}}}({\mathrm{\&eta;}}_{h,s}*\frac{{\mathrm{rate}}_f}{{\mathrm{rate}}_s}),$ The matching rate of f and h ${\mathrm{\&eta;}}_{f,h}=\frac{{\mathrm{\&eta;}}_{h,f}*\mathrm{rateOut}}{{\mathrm{rate}}_h}.$