Patexia Contest




Patexia seeks prior art for US Patent 6,170,073 (‘073), with a focus on claims 35 and 43. The patent allegedly describes a method of arranging encoded data signals into two classifications in an overlapping manner, then creating error detection codes for both classifications so that parts of the digital signal will have redundant error detection codes.

In speech signal transmission, a bad frame of data (whether due to poor signal or hardware failure) can easily propagate through the decoding of a digital speech signal to make the end result unintelligible.

The patent allegedly describes a method of minimizing the propagation of bad frames or errors by:

  • classifying the encoded digital signals into a 1st and 2nd class according to its influence on data quality (e.g. parts crucial to the signal versus parts that could be safely neglected) in an overlapping manner such that some data is included in both
  • generating at least 2 error detection codes with at least one for each of the 1st and 2nd classes
  • decoding the encoded digital signals for error detection in each classification
  • and finally estimating the quality of the digital signal based on the result of decoding

Figure 1. Venn diagram showing a first way the classifications may overlap

Figure 2. Venn diagram showing a second way the classifications may overlap

Submission Boundaries

  • references filed or published before 03/29/1996
  • patents or non-patent literature references
  • no geographic restrictions for references
  • no references allowed from the known references list

Contest clarification (01/31/2015):

For this contest, convolution coding algorithms, unfortunately, do not count as a method for overlapping data between two classes.

Have a question about this contest? Ask a Question


1Is the reference either a US patent filed, a foreign patent published, or a non-patent document published before March 29th, 1996? T/F
2Does the reference describe a system for encoding and decoding digital communications? 5
3Does the system first encode the data into digital signals, and then classify the digital signals into a 1st and a 2nd class according to influence on data quality, making sure that the classes overlap so that some data is included in both? 35
4Does the system generate at least 2 error detection codes with at least one error detection code for each class? 15
5Does the system then receive the encoded digital signals, classified into their 1st and 2nd classes, and decode the signals for error detection for each classification? 15
6Does the system finally estimate the quality of the received digital signals based on the results of the decoding of the received encoded digital signals and their 2 classifications? 30

Additional Notes

Prior Art Search

This is a Prior Art Search contest aimed at determining if a patent idea was known and publicly available before a patent was filed.

+5 bonus points will be awarded for non-patent literature and for foreign language references.

This contest will close on Sunday, February 8th, 2015 at 11:59 PM PST and is subject to modified prize distribution awards:

  • the top submission will be designated as a winner and receive $3,000
  • the next scored 2 submissions will be designated as winners and receive $500 each
  • the next scored submissions (to a maximum of 10) will be designated as runners up and receive a minimum of $100 each

Please review the Contest Rules. For more information on how to submit to this Contest type, please read the Intro to Prior Art Search page.