無線射頻識別報告 授課 教師 : 黃秋煌

1. 無線射頻識別報告授課教師 : 黃秋煌 報告人:王重凱

2. Outline • ECA Rule-based RFID Data Management • An Efficient RFID Data Processing Scheme for Data Filtering and Recognition

3. ECA Rule-based RFID Data ManagementJie Wu, Dong Wang, Huanye Sheng

4. INTRODUCTION • We are faced with many challenges in the process: (1) Data Aggregation (2) State Transformation. (3)Business process • Due to these challenges, we argue that rule engine can be efficiently employed.

5. INTRODUCTION • We propose a way using ECA rules to process RFID data. • This paper puts forward a solution using ECA rules to process RFID data efficiently and flexibly.

6. RELATED WORK • The concept of the architecture of Savant[12] (ALE's predecessor) the RFID middleware component in the EPC Network was initially proposed by the Auto-ID Center, an industry-sponsored research program to foster RFID adoption.

7. RELATED WORK • The ALE specification is the application-level interface standard developed by EPCglobal to allow clients to obtain filtered and consolidated EPC observations from a variety of sources [6].

8. RELATED WORK • The requirements and the architecture of RFID middleware have been studied by [13, 14, 15, 16]. Our work differs from the above, because it exclusively focuses on introducing ECA rules to processing events in the RFID middleware, which makes RFID data management more flexible and efficient.

9. ECA RULES IN DATA PROCESSING • A. ECA Rules _ECA model was first used in active database systems to support reactive behavior by integrating rules into DBMSs (database management systems). _In ECA model, a rule is defined by the following event-condition- action pattern: WHEN E IF C DO A;

10. ECA RULES IN DATA PROCESSING • B. ECA Rule-based RFID data management _ Input Interface _ Rule Base _ Rule Manager _ Condition Checker _ Action Manager _ Action Set _ Event Manager _ Event Queue _ Event Forming _ RFID Deployment

11. ECA RULES IN DATA PROCESSING • B. ECA Rule-based RFID data management

12. DESIGN AND IMPLEMENTATION • A. Design Issues RFID Event Processing : There are 3 layers in event processing: (1) raw events process layer (2) intermediate events process layer (3) business events process layer

13. DESIGN AND IMPLEMENTATION

14. DESIGN AND IMPLEMENTATION • B. Implementation Issues Our RFID system is programmed in Java language. In our approach, we choose to use Drools to implement rule engine.

15. DESIGN AND IMPLEMENTATION • (1) Raw events process layer An example of raw tag event specification is given below : class RawTagEvent{ physicalReaderID; tagID; timestamp; }

16. DESIGN AND IMPLEMENTATION • (1) Raw events process layer An example primitive tag event specification is given below : class PrimitiveTagEvent{ logicalReaderID; // ID of the logical reader physicalReaderID; /* ID of the reader who observes the tag /null- if the tag disappears*/ tagID; // ID of the appeared/disappeared tag idType; // Encode Scheme status; // 1-Appear/0-Disappear timestamp; /*The timestamp of the tag's appearance or disappearance*/

17. DESIGN AND IMPLEMENTATION filter; /*additional data that is used for fast filtering and pre-selection of basic logistics types*/ company; //tag's company prefix index reference; /*tag's reference (Item Reference, Location Reference, Asset Type, Serial Reference Individual Asset Reference)*/ serialNum; //tag's serial number if exists, //otherwise, equals to 0 }

18. DESIGN AND IMPLEMENTATION • (2) Intermediate events process layer rule "TagFilter“ when event:PrimitiveTagEvent(logicalReaderlD== "LI", idType== "SGTIN-96", filter== "011", company== "0001289") then ...... //further process end

19. DESIGN AND IMPLEMENTATION • (3) Business events process layer rule "Smart shelf Shortage Report" when SSReportEvent(map:map) //smart shelf report event then //get shaver's set and report the shortage Set set= (Set)map.get("000128"); if(set! =null && set.size(O<10) ShortageReport.report( item, set);

20. DESIGN AND IMPLEMENTATION //get shampoo's set and report the shortage set= (Set)map.get("000130"); if(set!=null && set.size(<20) ShortageReport.report( item, set); …… end

21. CONCLUSIONS • ECA rules is able to reduce large sets of RFID data to the precise sets of valuable information efficiently and flexibly. • When we use drools, we just write down the rule and put the rule in the system. We don't need to compile the rules. This is done by the rule engine automatically.

22. An Efficient RFID Data Processing Scheme for Data Filtering and RecognitionSih-Ying Li1, Hsu-Yang Kung1, Chi-Hua Chen2,* and Wen-Hsi Lydia Hsu3

23. INTRODUCTION • This study expects to develop an effective method of RFID data processing that can help RFID middleware quickly and effectively deal with large numbers of RFID data. • This study uses the electronic product code (EPC) to construct the tree structure, and design a hash function to reduce the data length for hash value generation.

24. RELATED WORK • The middleware is divided into four methods which are proposed in previous work. _Device Adaption and Management _Data Processing _Process Design _Efficiency

25. SYSTEMATIC STRUCTURE

26. SYSTEMATIC STRUCTURE • RFID Device _ This part describes the RFID readers and tags which have been designed by the different equipment specifications. Middleware must connect various readers to achieve the aim of a single user interface.

27. SYSTEMATIC STRUCTURE • Middleware _ Middleware has two core components which are the data manager and reader controller. The data manager is responsible for data filtering, collection, and storage. The reader controller is responsible for connecting readers, and grouping and monitoring readers.

28. SYSTEMATIC STRUCTURE • Middleware _ Data Manager : 1. Collection Module : The reader effectively collects and transmits data to the corresponding filtration module for processing . 2. Filter Module : This module has multiple filter functions. This module uses the hash function to filter duplicate data, and lets users filter specific data. That function is also called data classification. 3. Output Module : A user can set a data flow as output from the filter module. Additionally, data can be saved to a database specified by a user or as output data in the XML format.

29. SYSTEMATIC STRUCTURE • Middleware _ Reader Controller : 1. Monitoring Module : This module can support users to manage readers. The users can monitor the working condition of readers for real-time monitoring and troubleshooting. 2. Grouping Module : This module manages many readers which are used for different application scenarios. Thus, this work designs a grouping function, and the readers which are in a group are used for the same application scenarios. A user can then control multiple readers in a group simultaneously. 3. Adaption Module : This module controls readers from different manufacturers for users.

30. SYSTEMATIC STRUCTURE • Application _The data was transformed to information by middleware. Then this module delivers information to other application system or database. Therefore, a user can refer to the designation database or obtain data in the XML format.

31. EVALUATION • Clean-data Filtering _ To increase the filtration rate, this work uses a hash function for repeat information filtering. We use the hash function to design filter made of two methods: (1) hash function and (2) tree structure with the hash function.

32. EVALUATION • Hash Function

33. EVALUATION

34. EVALUATION

35. EVALUATION • User-defined Filtering _The RFID data are used in different application situations repeatedly. Thus, different application situations require different filtering data. User-Defined Filtering is designed by users, which is mean the users can control the data of duplicate filter, and user also can transform data to information according to the application situation.

36. EVALUATION • Classified Information _ Information can be classified according to EPC fields (e.g., header, filter value, company, and item reference). _Filtering is designed according to user needs and information not required, or only retains the information needed (e.g., header, filter value, company, and item reference).

37. EVALUATION • Inquiries by the Respective Need _The queries are built according to the EPC field (e.g., header, filter value, company, and item reference).

38. SYSTEM IMPLEMENTATION • We provide the users an interface. The users can set Clean DataFiltering and UserDefineFiltering and depend on their application scenarios and conditions to set multiple filter sources for information reusing.

39. SYSTEM IMPLEMENTATION

40. EXPERIMENTAL ANALYSIS AND EXPLORATION • Filtering Performance Verification Method _In this paper, we use simulation to verify that the method is efficiency. In experiments, we design two data sets which include: (1) generating 96-bit random numbers (2) using SGTIN-96 specification.

41. Experimental Analysis and Exploration • Filtration Efficiency Simulations _We consider three cases which include: (1) Duplicate 1/4, (2) Duplicate 1/2, and (3) Duplicate 3/4 shown in table 1.

42. EXPERIMENTAL ANALYSIS AND EXPLORATION • Not Consider whether Analytical Code _SGTIN Data:

43. EXPERIMENTAL ANALYSIS AND EXPLORATION • Not Consider whether Analytical Code _SGTIN Data:

44. EXPERIMENTAL ANALYSIS AND EXPLORATION • Not Consider whether Analytical Code _Random Data:

45. EXPERIMENTAL ANALYSIS AND EXPLORATION • Not Consider whether Analytical Code _Random Data:

46. EXPERIMENTAL ANALYSIS AND EXPLORATION • Parse the Code and then Filtered _SGTIN Data:

47. EXPERIMENTAL ANALYSIS AND EXPLORATION • Parse the Code and then Filtered _SGTIN Data:

48. EXPERIMENTAL ANALYSIS AND EXPLORATION • Parse the Code and then Filtered _ Random Data :

49. EXPERIMENTAL ANALYSIS AND EXPLORATION • Parse the Code and then Filtered _ Random Data :

50. CONCLUSIONS • We consider two situations which include (1) non-considering whether analytical code and (2) parsing the code and then filtered are described as the following subsections. _In situation (1), the performance of using HashCode method is the fast in any cases. In situation (2), the performances of using Cut_Hash and HashCode methods are faster than other methods in any cases. Therefore, if the application doesn’t consider security issues, using HashCode method is better when. Otherwise, using Cut_Hash method is better.