| Author |
 Topic  |
|
steveheap
Average Member
220 Posts |
 Posted - 02/10/2004 : 18:13:50
|
As I study the industry and the developments underway - Auto-ID labs being bought (or absorbed) into the Uniform Code Council who own/operate the bar code mechanism, and the conflicts brewing over ISO standards versus EPC standards, it struck me that either my understanding is lacking, or something is missing in the overall vision.
The EPC tag is dumb - it holds no more than its serial number, and that serial number can be decoded by the network of ONS servers (similar to the DNS system in the Internet) to point to the database that holds the full data about that individual product - ie the manufacturer. All well and good - you can read the tag and realize that it is a cut of beef packed on a certain date in a certain factory (assuming the manufacturer makes that database open to the "public". However, the item will have passed through many other companies on its way to the retail store, and all those companies could have collected data about the product as it passed their readers. If you wanted to recontruct a history of that product, where would you look? The actual code seems to only link back to the manufacturer - where is the other data, or even pointers to the location of the other data. If the beef had been potentially contaminated along the way in a distribution depot, how would anyone in the supply chain know that apart from painstakingly going back step by step checking each companies internal databases.
Where is the smooth running end to end process here?
Anyone enlighten me?
Steve |
|
|
sapan.shah
Average Member
134 Posts |
Posted - 02/10/2004 : 21:07:02
|
I have the following questions / points in these aspects
1. Doesn't RFID mean a low cost of ownership?.
This would mean that some one (manufacturer) would be keeping the information about the product in his database and would be sending the data in PML format when required by the requestor.
2. PML Standards have a provision for history
Meaining, that the PML itself would be a recursive structure and it can point to other servers. However, programs would have to be written in such a way that one looking for the history will ahve to parse accordingly. THis requires audit trails enabled in such cases, but at the same time, companies need to maturely come out with these solutions. It for sure, will not be an easy job.
However, even if the cut of beef is lets say contaminated at the distribution depot, all RFID can help is know that the beef was present there. The question is, how would we know that it got contaminated at the depot and not before /after that?
3. Integration issues
This would require a lot of integration and seamless interactions with various owners of data. For sure, it would be an Internet of things.
Sapan Shah
Patni Computer Systems Ltd
http://www.patni.com
India |
 |
|
|
steveheap
Average Member
220 Posts |
Posted - 02/11/2004 : 06:12:55
|
I cant see what incentive the manufacturer has for keeping a database with more information than the data needed to manufacture and get it out of the factory. That implies that additional data is kept elsewhere by others in the supply chain, and then you need to have the naming/addressing structure to find it.
In my beef example, the issue is how do you find that a piece of beef has been in a depot at the time the contamination happened, when the only two databases that you are certain know something about the beef's history are the manufacturer (farmer) and the current owner of the item - perhaps the store. I cant see how you can track back through the lifecyle of that beef back through the chain, as no-one seems to have a history of it's progression through the chain. |
|
|
|
Tom Napier
Starting Member
6 Posts |
Posted - 02/11/2004 : 09:56:37
|
Hi Steve and Sapan,
The EPC tag (C1G2) that Wal*Mart wants will have either 64 or 96 bits of usable memory. Thus, the lot, package date, best before, sell by, weight in lbs or kg, etc can be written on the tag. This is similar to the UCC/EAN 14 with the post amble data on the case label. The unique id (UID) will be similar to the SSCC18 code. The whole idea of EPC is to merge the two into one super code and eventually get it on every product. The EPC will give a directional locator for every item manufactured from cradle to grave. This is something that the UPC or EAN item codes cannot give. The EPC was not designed to prevent bad or diseased products such as BSE in cattle. My concerns are the huge amount of data in the global database and privacy/security concerns.
We are currently working with the CCIA and different provincial governments in Canada with LF RFID ear tags. We get the longer range and we can capture and validate the where and when of the cattle. Best business practices will also associate vaccinations, feeding and exercise of the cattle. This can be done either with bar code or RFID and good software.
Cost is relative. In Canada, we have an embargo of cattle because of two cows, one found with BSE in Canada and one we shipped to the USA and was found with the infection. This cost is huge to both the industry and consumer. This is why better tracking of beef is important. These two cows most likely contracted BSE through the feed which is traded between Canada and the USA. We must also remember that bar coding was also expensive and now it is part of our everyday existence in retail and business. Hopefully the initial costs will be offset by the efficiencies.
Cheers,
Tom Napier, R.Moroz Ltd.
www.RFIDCanada.com |
|
|
|
steveheap
Average Member
220 Posts |
Posted - 02/11/2004 : 13:46:20
|
Tom
Thanks for the extra insight. A couple of things are still confusing to me.
On the EPC tag - the number of bits is 96, and those are assigned as follows:
8 bits to the Universal Header, that then structures the rest of the bits. In the Universal scheme, the next 28 bits are assigned to a company - usually the manufacturer of the item, or at least the company putting the tag on the item. That company then assigns 24 bits to describe the product and then 36 bits for a unique serial number. Remember these are bits, not bytes, and any realistic text or other formatted characters would eat into this payload at a tremendous rate. The EPC tag therefore only gives this "dumb" number which can be linked via the EPC network to that original database where you can collect the meaningful information about the item - ie the EPC tag is an index key into the database.
That leaves me with my issue - does the network of ONS servers hold multiple locations for information about the tag - ie one reference for everytime it is spotted on it's journey, with the implicit assumption that each company with data about the tag is willing to share it along the path of the supply chain in order to reconstruct it's life.
There are two issues in there - the ability of the ONS system to handle all these pointers for all these tags in "real-time", and the willingness of companies to share the data that the pointers are pointing to.
Steve |
|
|
|
sapan.shah
Average Member
134 Posts |
Posted - 02/12/2004 : 03:27:17
|
Hi
Quoting from the ONS Specifications:"
Dynamic ONS services record a sequence of custodians as an object moves through a supply chain."
Which, I feel, again would be only pointers to the PML Document services and integrators would have to query them to go through the audit trail. Alternatively, as per PML Specs, there also are XML tags in the document format where one can specify the current owner and previous owner and the next owner. But again, the PML Services would have to query the ONS Servers only.
As I was discussing this same thing with some one else around here, something more intersting came up. Quoting that
So if I query at the retail warehouse level about the information of a product, will I be pointed to the data stored by the originator of this product (manufacturer) or to the data stored by the "distributor" who packaged the product. Or will every one down stream have to store back any relevant info into the originator's product database by design ?
Would this help in solving the contamination problem? How would I detect that contamination happened at this particular time?
This is really a good discussion and let us continue to clear more doubts.
Sapan Shah
Patni Computer Systems Ltd
http://www.patni.com
India |
|
|
|
sapan.shah
Average Member
134 Posts |
Posted - 02/12/2004 : 04:04:13
|
Tom,
Is it possible to get demo version of Savant Systems?. Do you know any one who can give me a copy?
Sapan Shah
Patni Computer Systems Ltd
http://www.patni.com
India |
|
|
|
steveheap
Average Member
220 Posts |
Posted - 02/12/2004 : 06:59:00
|
Sapan
Your comment on your discussion is the nub of the problem - each company in the distribution chain has a piece of the information about the history of that item, and I can see how each company can potentially make use of it's own information to make some internal improvements. However, if you are the ultimate receiving company and you want to know where the item is, you need two things - an ability to be given a pointer to the last known location (rapidly updating ONS servers may give that, but the amount of data they have to hold is potentially huge), and a willingness of all companies in the supply chain to open up their databases (or provide an external view of certain fields) to allow you to see what they know about the item in question. If a supplier serves both Wal-mart and Safeway, how does he permit Wal-mart to see their items and not Safeway's items? Interesting set of issues to be faced and resolved! |
|
|
|
Tom Napier
Starting Member
6 Posts |
Posted - 02/12/2004 : 13:09:24
|
Hi Steve,
I see a huge problem for the database. However, EPC Global is the new birth from the UPC/EAN. These companies are the blue-chip manufacturers and retailers. Do they know best? That will be for the historians to figure out. If you do a search on the efficiency consumer response (ECR), you will find a disaster (Jim Tompkins, Supply Chain Synthesis). The idea there was to reduce inventory by reducing lead time. Unfortunately in logistics things don't always work that way, so in the 5 years they wanted to get ready for this, inventory went up 28%. Ironing out the who’s and the where’s is their problem.
The idea behind the EPC is to have everyone that touches the product and reads the tag is supposed to write to the database. Thus the need for EDI is eliminated. This can be driven in by the retailer or the manufacturer. I don't know which is going to responsible but if Wal-Mart gets their way, you won't get the product into the distribution centre unless you have the supply chain requirements complete.
All of this said, we have not done any programming with Savant. There are no ISO standard to follow so we will only be wasting time because the C1G2 tag is still vapourware until all of the intellectual property (IP) rights are solved. To sign on as a supplier to EPC Global, you must sign over the IP rights. To my knowledge (and if I am wrong, please let me know because things happen so quickly that its hard to keep up), Matrix and Alien are the only two that have so far and their signing is conditional upon other IP holders doing the same. Wal-Mart is now giving everyone until January 2006 (The RFID Power Play, APICS article by Robert B. Handfield) to comply. It won't be long before the DOD does the same. All of you programmers, you have more time.
We have more pressing issues as a distributor and VAR than the EPC. I need to engineer an autoclave metal tray and metal surgical instruments going into the autoclave machine. This is a real science project. I need to identify the parts going into and associate them to the tray. The tray needs to have a RFID tag that stores (write) all of the parts, time stamp and machine's data. My problem is with the metal and storage capacity of the tag (TI has only 2K bits, 256 ascii characters). Does anyone have any ideas?
Tom Napier, R.Moroz Ltd.
www.RFIDCanada.com |
|
|
|
sapan.shah
Average Member
134 Posts |
Posted - 02/12/2004 : 20:27:49
|
Steve mentioned [i]an ability to be given a pointer to the last known location (rapidly updating ONS servers may give that, but the amount of data they have to hold is potentially huge), and a willingness of all companies in the supply chain to open up their databases (or provide an external view of certain fields) to allow you to see what they know about the item in question[/]i
Would this mean that each level of the tag receiver would be opening up PML Services, when they are willing to share the data?
There ahs to be formal agreements for an e-business / data exchange like this and It will definitely involve more formalization..
Sapan Shah
Patni Computer Systems Ltd
http://www.patni.com
India |
|
|
|
MarkPalmer
New Member
43 Posts |
Posted - 04/30/2004 : 19:33:23
|
quote:
Originally posted by steveheap
Where is the smooth running end to end process here?
I guess I hate to be the guy raining on the ONS/EPC IS parade, but I just don't see this happening at all. I don't see any business driver for this solution - it's not a technical issue. These technical issues are not difficult to solve - versioning of data, security, etc. That's not the problem.
The problem is nobody cares, from a business perspective. In fact, I'll say it stronger - the big guys, who control the RFID mandates DO NOT WANT an open exchange of things - they want to control it themselves - the better they can control the supply chain, the better they can compete.
ONS/EPC IS goes AGAINST the vision of the RFID of all things.
- Mark
Mark Palmer
Director of Marketing/Product Management
ObjectStore (www.objectstore.net)
A member of EPCGlobal |
|
|
|
newton
Starting Member
7 Posts |
Posted - 05/01/2004 : 16:39:14
|
Interesting topics.
I just don't see that there will be some kind of "global database" to manage the supply chain. I would expect that every company in the chain will manage its own. When the item is received, its "last" tag will be scanned. The "last" tag and new tag will be linked and stored in the company's data base. Every company will manage its own data base. A query made to any company in the chain will eventually produce the history of the "last" tags.
What will be function of the global PML server ?
ONS can be used to retrieve the URI or some kind of address based on the "EPC Manager - company" section of the 96 bits. Or it can be a simple table lookup and the cache downloaded from a central site. Is there a need for instant ONS query ? I would expect that it is more like the iana.org type of addressing e.g. OID than the DNS type of setup.
Once the URI is located, B2B type of interface will be used to perform the necessary query if needed.
|
|
|
|
steveheap
Average Member
220 Posts |
Posted - 05/01/2004 : 17:05:50
|
Yes - I can follow that technical logic - although there is still a massive issue of trust. If I am Wal-Mart and I am waiting for my cans of beans, and I find that a trunking company is the one that currently has them, I may have no commercial relationship with that company, no record of trusted contact, and yet I want to be given access to their systems to see where my beans are. What if there are also cases of beans going to my competitors in the same pallet - would that be visible?
Steve |
|
|
|
sapan.shah
Average Member
134 Posts |
Posted - 05/02/2004 : 02:32:02
|
Am trying to put a few points forward - hope it is help ful.
Although, I am not debating on whether ONS would be there or not. It may or may not be. Companies right now, are exposing / planning to expose ONS services internal to their organizations - Atleast I, being into this, am helping companies do that.
The scenario will be that the Mandate Controllers will be using these services exposed by the manufacturers as the data transfer will be in EDI format.
The use of Global ONS Server, may not come true unless technologies like ebXML become omni-present in B2B scenario - Althought, its a very strong statement and the views are my own.
1. PML Server will not be global. The concept of PML Server is - The companies right now exchange information in EDI format. While RFID, as driven by the Auto ID Labs, is all about "Internet of things", if this dream is realized, then there has to be a common language / protocol for understanding and identifying each of these items. In this case, companies will be exposing their services which will give the product information in this PML Format. These services will be residing on an entity called PML Server.
The ONS Server, whether it is internal to the organization or global, will give the URI to this particular Service on the PML Server that will give information about the product.
quote:
originally posted by newton
Once the URI is located, B2B type of interface will be used to perform the necessary query if needed.
2. ONS- If and when it comes in place - will only be giving a URI to the EPC and then it would be a B2B agreement - using technologies like ebXML and Registeries like UDDI for EPC.
Sapan Shah
Patni Computer Systems Ltd
http://www.patni.com
India |
|
|
|
MarkPalmer
New Member
43 Posts |
Posted - 05/02/2004 : 16:16:26
|
Yeah, I think we're in violent agreement here!
No global ONS other than "lookup", and local, private, shared agreements based on extensions of existing EDI.
That's to the point I made (somewhere in here!) that I see the data sharing issue as a third order issue with RFID, at best; it's pretty much just doing better EDI.
Thoughts? |
|
|
|
CraigHatmaker
Starting Member
1 Posts |
Posted - 05/18/2004 : 11:21:52
|
The ONS with its security/visibility issues concerns me. Is the ONS required?
I have been told Wal*Mart will not be using the ONS and will rely on ASNs instead. Can anyone confirm?
I have been told Wal*Mart's objective for RFID in the near term is to enabling increasing case conveyor belt speeds to 600 feet per minute and still read case labels (to expand existing DCs capacity). Can anyone confirm? If so, would the latency of the global ONS registry make this goal unobtainable if Wal*Mart used the ONS?
I have been told Wal*Mart, the DOD, and UCC disagree on the format of the EPC. Can anyone confirm? If so, is the ONS still viable?
With the 96 bit tag, the ONS would seem to be required, but, if history predicts the future, it will be only a very short while before 128 bit tags cost the same or less. With a 128 bit tag, the EPC could contain all 14 digits of the GTIN (stored as a type of long integer), the manufacture's ID (which may not be the same as company prefix in the GTIN when the GTIN represents a "Private Label" product), and a serial number. At this point, the ONS does not seem to be required to "decode" the EPC. Does anyone have any knowledge about tags with more that 96 bits?
Has anyone estimated the cost of the Global ONS registry? Who will pay? Who will profit?
Craig Hatmaker
Perdue Farms, Inc. |
|
|
|
Topic |
|
|
|
How will RFID data really become visible to all
