Your SlideShare is downloading. ×
0
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Slides
Upcoming SlideShare
Loading in...5
×

Thanks for flagging this SlideShare!

Oops! An error has occurred.

×
Saving this for later? Get the SlideShare app to save on your phone or tablet. Read anywhere, anytime – even offline.
Text the download link to your phone
Standard text messaging rates apply

Slides

112

Published on

Published in: Business, Technology
0 Comments
0 Likes
Statistics
Notes
  • Be the first to comment

  • Be the first to like this

No Downloads
Views
Total Views
112
On Slideshare
0
From Embeds
0
Number of Embeds
0
Actions
Shares
0
Downloads
0
Comments
0
Likes
0
Embeds 0
No embeds

Report content
Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel
No notes for slide
  • SCM&CRM /in ERP!
  • Transcript

    1. By Richard Cocci, Yanlei Diao, and Prashant Shenoy From University of Massachusetts, Amherst
    2. Basics on RFID <ul><li>Radio-frequency identification ( RFID ) is an automatic identification method, relying on storing and remotely retrieving data using devices called RFID tags. </li></ul><ul><li>Components: </li></ul><ul><li>- the tags, the readers and the antennas. </li></ul><ul><li>- application software for enterprise usage. </li></ul>
    3. Introduction This paper focus on transforming the raw data into “vanilla” data compatible enough for higher level managerial software to process. .
    4. Introduction <ul><li>Aims: </li></ul><ul><li>- Data management techniques that are capable of handling massive amounts of data generated by large RFID deployments. </li></ul><ul><li>- efficient data information transformation! </li></ul><ul><li>Techniques: </li></ul><ul><li>-stream processing, information integration; </li></ul><ul><li>-data cleansing, event processing, data compression; </li></ul>
    5. Introduction <ul><li>Problems and design requirements </li></ul><ul><li>Data Information mismatch for monitoring. </li></ul><ul><li>Incomplete, insufficient data for track-and-trace. </li></ul><ul><li>Scalability.. </li></ul><ul><li>Low-latency. </li></ul><ul><li>Techniques: </li></ul><ul><li>Data cleaning. </li></ul><ul><li>Data Compression </li></ul><ul><li>Event Processing </li></ul>
    6. Data Cleansing <ul><li>Filter out abnormal and corrupted data readings, remove duplicate readings, and smooth readings to assist in recreating missing data from imperfect readers. </li></ul><ul><li>After cleansing, a data set will be consistent with other similar data sets in the system. The inconsistencies should be detected and removed, deriving from tags’ reading errors. </li></ul>
    7. Data Compression <ul><li>Reduce the number of tag readings recorded, supporting precise location tracking and event detection. </li></ul>
    8. Event Processing <ul><li>an event processor operating over a stream of tag readings to search for user specified trends. </li></ul><ul><li>Distributed Event Processing: </li></ul><ul><li>-By compressing the local tag data and transforming it into a series of events, the data volume is reduced to a scale which can become manageable for a large central repository. </li></ul>
    9. Architecture of SPIRE
    10. Architecture of SPIRE <ul><li>Physical Device Layer. </li></ul><ul><li>Cleaning, Compression, and Association Layer </li></ul><ul><li>- RFID readings are known to be inaccurate and lossy_Data cleansing technique </li></ul><ul><li>- two compression techniques to effectively reduce data volume between the readers and the event processor. </li></ul><ul><li>- create events: using attributes such as product name, expiration date, and saleable state. </li></ul>
    11. Architecture of SPIRE <ul><li>Complex Event Processor </li></ul><ul><li>- the user writes a query and registers it as a continuous query with the complex event processor. </li></ul><ul><li>- remove duplicate data and transform data to the format required to archive. </li></ul><ul><li>- handle queries: detect the event; send a subquery to the database; combine information retrieved from the database with that obtained from the stream; return a complete result to the user. </li></ul>
    12. Filtering and Smoothing <ul><li>Anomaly filtering: </li></ul><ul><li>-remove spurious readings and truncated tagIDs. </li></ul><ul><li>Temporal Smoothing: </li></ul><ul><li>Time conversion </li></ul><ul><li>- apply a time-stamp to form the tuple </li></ul><ul><li><readerID, tagID, TimeStamp> </li></ul><ul><li>Duplication </li></ul><ul><li>- duplicates may originate from redundant setup, overlapping ranges, etc. </li></ul>
    13. Location (Data) Compression <ul><li>a cache is used to record RFID tags currently located at each reader according to most recent observations. </li></ul><ul><li>When new readings are performed, this cache is used to identify tags which might no longer be located at each reader. </li></ul><ul><li>If a tag is missed at a reader x times, an event is generated to indicate that the tag has left the reader. </li></ul><ul><li>X is a system parameter. </li></ul>
    14. Containment (Data) Compression <ul><li>Option: only one tag of a highest containment level will be present at a time. </li></ul><ul><li>Option: containment relationships will have to be manually entered as they are created. </li></ul><ul><li>Cache mechanism… </li></ul>
    15. An algorithm for data compression <ul><li>time-varying colored graph model. </li></ul>
    16. An algorithm for data compression
    17. Event Processing <ul><li>Event Generation: </li></ul><ul><li>- set up schema of events, e.g. </li></ul><ul><li>StartLocation(Tag A, Location B, Timestamp) </li></ul><ul><li>- events are also stored directly in the event database, after being transformed into the desired schema, to create a historical record for each tag of the movements and containment changes that it has experienced. </li></ul>
    18. Event Processing <ul><li>Complex Event Processing </li></ul><ul><li>- the complex event processor is used to allow a user to specify custom continuous queries over both the incoming event stream and historical data. </li></ul><ul><li>- similar to SQL query: </li></ul><ul><li>[FROM <stream name>] </li></ul><ul><li>EVENT <event pattern> </li></ul><ul><li>[WHERE <qualification>] </li></ul><ul><li>[WITHIN <window>] </li></ul><ul><li>[RETURN <return event pattern>] </li></ul>
    19. Event Processing <ul><li>Example: query a certain product, which has already been packaged, yet has not escaped from reading; also return the relevant ProductName and TimeStamp. </li></ul><ul><li>EVENT SEQ(PACKAGING_READING x, </li></ul><ul><li>!(EXIT_READING y)) </li></ul><ul><li>WHERE x.TagId = y.TagId </li></ul><ul><li>WITHIN 12 hours </li></ul><ul><li>RETURN x.TagId, x.ProductName, x.TimeStamp </li></ul><ul><li>Returned: potentially missed data items! </li></ul>
    20. Future Work <ul><li>This is an area of ongoing research, but their initial findings indicate that the compression techniques may be effective enough at reducing data volume at the local level to make feasible direct replication of the events from the local event databases to a global event database. </li></ul>

    ×