Randomization—An Interview with Ken Traub—Part 4: The Algorithmic Approach

This is the fourth installment of a five part interview with Ken Traub, GS1 standards expert and independent consultant, on GS1 serial number randomization.  The full series includes essays covering: GS1 Serial Number Considerations Properties of Randomization Threat Analysis  Algorithmic Approach (this essay) Other Approaches to Randomization In this installment, Ken explains the algorithmic approach to serial number randomization.  – Dirk.

Looking for Serialization & Tracing Insights?

To get full access to this and other great RXTRACE_LOGO_Trans_250x94 essays, you need to
either Login, or upgrade your current subscription!
 Unlimited access is just a few clicks away!
To learn why RXTRACE_LOGO_Trans_250x94 is such a great value, click here.

3 thoughts on “Randomization—An Interview with Ken Traub—Part 4: The Algorithmic Approach”

  1. I completely agree with Ken’s algorithmic approach. A set of sequential numbers can generate a set of corresponding random numbers in a very straight forward manner. This greatly simplifies the manufacturing serial number creation database.

    My US Patent 7,137,000 describes such a method, designed to generate random numbers to be stored in an HF RFID tag, based on the sequential tag id serial numbers. Here, these were stored in a tag on a ribbon spool. The card printer had a secure but undernourished $2 16-bit microprocessor which read the ribbon tag id, recreated the random number using the same algorithm, and compared it to the one stored on the tag to authenticate the ribbon’s manufacturer.

    As seen here, the amount of computation to generate the random number does not necessarily have excessive. The key is to invest the time in the elegance of the randomization algorithm to minimize the computation. That’s also part of what the patent is about. This can give every company a unique way of generating random numbers.

  2. Thank you both for an excellent and informative Q&A. I was wondering what your respective views might be on the potential benefits of a cipher(encrypted) serialization process, insofar as the avoidance of Big Data issues. In particular, indexing challenges and shear data infrastructure costs when dealing with 100’s of millions (or even billions) of product codes per annum. I understand these issues may be mitigated by leveraging cipher keys…thereby avoiding individual code storage after on-line application.

    1. Using a cipher to generate random serial numbers has the advantage that you can keep track of which numbers you have allocated through a simple counter, instead of keeping the complete list of random serials. In that sense, it reduces the stored data content.

      However, it is likely that you *also* want records of which serial numbers were issued on which date, with which batch/lot, which ones were discarded due to manufacturing errors, etc. So you will end up needing some sort of data record (a “commissioning event”) for each serial issued. Once you do that, the fact that the serial was generated by cipher doesn’t really help – you may as well store the actual serial with each commissioning event rather than the pre-ciphered index.

Comments are closed.