A Novel Method for Formally Detecting RFID Event Using Petri Nets

1. A Novel Method for Formally Detecting RFID Event Using Petri Nets SEKE 2011

2. Outline • Introduction • The Model of RFID Events • The Model of ED-net • ED-net models for complex events • Detecting Complex Events • Conclusion

3. Introduction – RFID data • An RFID system : RFID tags, readers, middleware and application software. • Middleware plays the primary role in RFID data management. • RFID observation is of some form (epc, reader, timestamp)

4. RFID data are temporal, dynamic and in large volume, must be processed in real time • RFID data are inaccurate and have implicit semantics

5. Introduction – Complex event • Originated from active database. • Detecting methods for RFID complex events RCEDA / SASE / QDDCattetc. • But lack formal semantics to descript complex events, which may bring ambiguity and confusion in expressing and understanding RFID complex events.

6. So, we propose a Petri net-based method named ED-net for the description of complex events in RFID • which is convenient for describing temporal and parameterized constraints with locality property, token-flow mechanism and combinability.

7. The Model of RFID Events • E: an event type • E: an event instance • t_begin(e): the start time of e • t_end(e): the end time of e • interval(e):t_end(e) − t_begin(e) • dist(e1,e2):t_end(e2) − t_ end (e1) • OR(∨), AND(∧), and NOT (¬)

8. The Model of ED-net • Definition (ED-net) An ED-net (Event Detection Net) is a tuple N = (Σ, P, T, A, C, G, B, E) • Σ is a finite set of non-empty types, called color sets; • P is a finite set of non-empty places; • T is a finite set of non-empty transitions; • A is an arc set, A ⊆ P × T ∪ T × P; • C is a color function, C : P→Σ; • G is a guard function, G : T→{Expr}, where Expr is a boolean expression;

9. The Model of ED-net B is a body function, B : T→{Stat}, where Stat is a group of assignment operations; E is an arc function, E : A→{AExp}, where AExp is an expression whose value is a multi-set. The form of AExp could be: AExp :＝ m′c|n′v|m′c + AExp|n′v + Aexp where m and n are positive integers, c is a constant value of a specific color, and v is a variable.. The sign ‘+’ means addition of two multi-sets

10. An example of ED-net model

11. Dynamic behavior of ED-net • Check if a transition could be fired under current marking according to the firing rules introduced later. • More than one transition may be enabled in one step, we randomly choose one to occur. • Do the assignment operations according to the body of the transition, calculating the start and end time of the complex event

12. ED-net models for complex events ED-net model for disjunction

13. ED-net model for aggregation

14. Detecting Complex Events Algorithm 1: Constructing ED-Net model Input: complex events set SE Output: ED-Net model N foreach complex event e in SE do if e has not been modeled in N then Get e’s sub-events set CE; foreach sub-event ce in CE do if ce has been modeled in N then Add a place p’ as a copy of place p; else Build ce’s ED-net model; Extend N with ce’s model; end end Build the ED-net model for event e (taking its sub-events as inputs); Extend N; end end return SN

15. Algorithm 2: Detect complex events based on ED-net Input: ED-net model N = (Σ,P,T,A,C,G,B,E) M:Current marking of N; Enabled transition set Te = ∅; Repeat foreach t in T are marked do repeat Find b for variables related to t until G(t) is true under binding b; Add transition t to Te; foreach t in Te do Fire t; Evaluate output arc expressions of t foreach place p in •t do M′(p)= M(p) − E(pt); end foreach place p in t• do M′(p)= M(p) + E(tp); if p is a complex event Send M to applications; end foreach place p in P − (•t ∪ t•) do M′(p)= M(p); end Change current marking M to M′; Remove transition t from Te; end until no RFID data are received;

16. Conclusion • ED-net offers unified ways for describing both temporal and parameterized constraints of events and defining rules for calculating attributes of complex events. • all complex events detected in one model; • common sub-events are detected only once; • improving efficiency

17. Thank You!