5. HasData(M, x)
HasData( D, x )
Directory
1. WhoHasData(x)
2.1 getEvents (x)
2.2 getEvents (x)
Query Query Query
Manufacturer M Distributor D Retailer R
Repository Repository Repository
Capture Capture Capture
EPC
x
6.
7. Estimating remote procedure call cost
[Murthy and Robson, 2008]
Transfer
Send Receive
Lookup
Receive Send
Transfer
8. Decentralized Centralized
unstructured P2P metadata integration
Virtual data integration GS1 PoC [3] Verisign DS [4] IBM PoC [5]
ePedigree [1]
Theseos [2]
PTSP [6] EPCDS [7] ADS [8]
Theseos has several distributed data stores.
Queries are answered recursively.
Each company can enforce data ownership.
Afilias ESDS [9]
BRIDGE Directory [10] BRIDGE Query
Relay [11]
BRIDGE Directory relies on centralized
services to store data provider links.
There are scalability and single point-of-
failure issues to be considered. UniSalento DS
SLS [12]
[13]
UniPR DS [14]
IOTA [17]
EPCIS caching
[15] TraceSphere
LoTR [18] [16]
Materialized data integration
UniKoeln DS [19] OIDA [20] ID @URI [23]
OIDA relies on a Peer-to-Peer network with a ID@URI uses a product-agent architecture.
hashing algorithm for data placement in nodes. It All data concerning the item is forwarded to a
is fully decentralized and has potential for high central data store, managed by the product’s
scalability. manufacturer.
InnoSem [21] There are issues about response quality and
timeliness. Query capabilities are limited to
object ID matching.
WWAI [22]
structured P2P data integration
13. Decentralized Centralized
unstructured P2P metadata integration
Virtual data integration GS1 PoC [3] Verisign DS [4] IBM PoC [5]
ePedigree [1]
Theseos [2]
PTSP [6] EPCDS [7] ADS [8]
Theseos has several distributed data stores.
Queries are answered recursively.
Each company can enforce data ownership.
Afilias ESDS [9]
BRIDGE Directory [10] BRIDGE Query
Relay [11]
BRIDGE Directory relies on centralized
services to store data provider links.
There are scalability and single point-of-
failure issues to be considered. UniSalento DS
SLS [12]
[13]
UniPR DS [14]
IOTA [17]
EPCIS caching
[15] TraceSphere
LoTR [18] [16]
Materialized data integration
UniKoeln DS [19] OIDA [20] ID @URI [23]
OIDA relies on a Peer-to-Peer network with a ID@URI uses a product-agent architecture.
hashing algorithm for data placement in nodes. It All data concerning the item is forwarded to a
is fully decentralized and has potential for high central data store, managed by the product’s
scalability. manufacturer.
InnoSem [21] There are issues about response quality and
timeliness. Query capabilities are limited to
object ID matching.
WWAI [22]
structured P2P data integration
14. Chain parameters
Number of companies
Application parameters Average item records
Message size Average length
Item record size
System parameters
Bandwidth Product parameters
Processing speed Average sub -components
Seek time Average component depth
Traceability
cost model
Report
15. Auto supply chain
Short and broad chain – 700 companies, 6 levels deep, 3 components per level
16. Conclusion
• Developed a cost model to quantitatively
compare traceability systems
• Future work
– More detail
– Address scale and security concerns
– Validate model using actual systems
18. Bibliography
• [Do06]
– Hong-Hai Do and Jurgen Anke and Gregor Hackenbroich, Architecture
Evaluation for Distributed Auto-ID Systems, Proc. 17th International
Workshop on Database and Expert Systems Applications (DEXA), 2006
• [Evdokimov10]
– Sergei Evdokimov and Benjamin Fabian and Steffen Kunz and Nina
Schoenemann, Comparison of Discovery Service Architectures for the
Internet of Things, IEEE International Conference on Sensor Networks,
Ubiquitous, and Trustworthy Computing (SUTC), 2010
• [MurthyRobson08]
– Karin Murthy and Christine Robson, A model-based comparative study of
traceability systems, Proceedings of the International Conference on
Information Systems, Logistics and Supply Chain (ILS), 2008