Fingerprinting with Minimum Distance Decoding

This work adopts an information theoretic framework for the design of collusion-resistant coding/decoding schemes for digital fingerprinting. More specifically, the minimum distance decision rule is used to identify 1 out of t pirates. Achievable rates, under this detection rule, are characterized in two distinct scenarios. First, we consider the averaging attack where a random coding argument is used to show that the rate 1/2 is achievable with t=2 pirates. Our study is then extended to the general case of arbitrary $t$ highlighting the underlying complexity-performance tradeoff. Overall, these results establish the significant performance gains offered by minimum distance decoding as compared to other approaches based on orthogonal codes and correlation detectors. In the second scenario, we characterize the achievable rates, with minimum distance decoding, under any collusion attack that satisfies the marking assumption. For t=2 pirates, we show that the rate $1-H(0.25)\approx 0.188$ is achievable using an ensemble of random linear codes. For $t\geq 3$, the existence of a non-resolvable collusion attack, with minimum distance decoding, for any non-zero rate is established. Inspired by our theoretical analysis, we then construct coding/decoding schemes for fingerprinting based on the celebrated Belief-Propagation framework. Using an explicit repeat-accumulate code, we obtain a vanishingly small probability of misidentification at rate 1/3 under averaging attack with t=2. For collusion attacks which satisfy the marking assumption, we use a more sophisticated accumulate repeat accumulate code to obtain a vanishingly small misidentification probability at rate 1/9 with t=2. These results represent a marked improvement over the best available designs in the literature.

💡 Research Summary

The paper investigates collusion‑resistant fingerprinting from an information‑theoretic perspective, focusing on the use of a minimum‑distance (MD) decoding rule to identify a single traitor among a coalition of t pirates. Two distinct attack models are examined.

In the first model, the averaging attack, each of the t pirates contributes a binary fingerprint and the coalition outputs the arithmetic average of their symbols. For the case of two pirates (t = 2), the authors employ a random‑coding argument to show that a rate of ½ is achievable under MD decoding, meaning that half a bit of user information can be embedded per channel use while the probability of misidentification vanishes as the code length grows. This rate doubles the best known rates obtained with orthogonal codes or correlation detectors, which typically hover around 0.25. The analysis is then extended to arbitrary t, revealing a trade‑off between the required minimum Hamming distance (and thus code length) and the achievable rate; as t increases the code design becomes substantially more complex.

The second model assumes the marking assumption, which restricts the coalition to modify only those positions where at least one pirate’s symbol differs from another’s. Under this constraint, the authors prove that for t = 2 the ensemble of random linear codes can achieve a rate of 1 − H(0.25) ≈ 0.188, where H(·) denotes the binary entropy function. This result improves upon the rates achievable by correlation‑based detectors. Crucially, they also demonstrate that for any t ≥ 3 there exists a non‑resolvable collusion attack that defeats MD decoding for any positive rate, establishing a fundamental limitation of the MD approach in multi‑pirate settings.

Motivated by the theoretical findings, the paper proceeds to construct practical coding and decoding schemes using belief‑propagation (BP). For the averaging attack, a repeat‑accumulate (RA) code is designed, achieving an error probability that decays to zero at a transmission rate of 1/3. For attacks satisfying the marking assumption, a more sophisticated accumulate‑repeat‑accumulate (ARA) code is employed, delivering vanishing error probability at a rate of 1/9 when t = 2. Both constructions retain low decoding complexity comparable to traditional orthogonal‑code systems while delivering substantially higher rates and reliability.

Overall, the work establishes that MD decoding can provide significant performance gains over conventional orthogonal‑code and correlation‑detector designs in fingerprinting, especially for small coalitions. It also clarifies the inherent limits of MD decoding for larger coalitions, suggesting that additional mechanisms (e.g., multi‑level coding, joint detection) are required to maintain security when t grows. The combination of rigorous achievability proofs and concrete BP‑based code designs makes the paper a valuable contribution to both the theory and practice of digital fingerprinting.