Generalised tally-based decoders for traitor tracing and group testing

We propose a new type of score function for Tardos traitor tracing codes. It is related to the recently introduced tally-based score function, but it utilizes more of the information available to the decoder. It does this by keeping track of sequences of symbols in the distributed codewords instead of looking at columns of the code matrix individually. We derive our new class of score functions from a Neyman-Pearson hypothesis test and illustrate its performance with simulation results. Finally we derive a score function for (medical) group testing applications.