Authentication and Secrecy Codes for Equiprobable Source Probability Distributions

We give new combinatorial constructions for codes providing authentication and secrecy for equiprobable source probability distributions. In particular, we construct an infinite class of optimal authentication codes which are multiple-fold secure against spoofing and simultaneously achieve perfect secrecy. Several further new optimal codes satisfying these properties will also be constructed and presented in general tables. Almost all of these appear to be the first authentication codes with these properties.

💡 Research Summary

**
The paper investigates the simultaneous achievement of authentication and perfect secrecy for equiprobable source distributions by means of combinatorial constructions. After a brief historical overview of authentication codes—starting from the early Gilbert‑MacWilliams‑Sloane constructions based on finite projective planes and culminating in Stinson’s 1990 work that combined Steiner 2‑designs with perfect secrecy—the authors formalize the authentication‑secrecy model. Three parties (sender, receiver, opponent) are considered, with a source set (S) of size (k), a message set (M) of size (v), and a key set (E) of size (b). An encoding rule (e\in E) is an injective map (S\to M). Spoofing attacks of order (i) are defined, and Massey’s lower bound (P_{d_i}\ge (k-i)/(v-i)) is recalled. A code is called (t)-fold secure if equality holds for all (0\le i\le t). Perfect secrecy requires that the a‑posteriori probability of any source state given a message equals the a‑priori probability; under uniform key usage this translates to each message appearing equally often in every column of the encoding matrix (Lemma 1).

The authors then turn to combinatorial design theory. A (t)-(v,k,λ) design consists of a point set of size (v) and a block family of size‑(k) subsets such that each (t)-subset of points occurs in exactly λ blocks. Basic identities (Lemma 2, 3) relating the number of blocks (b), replication number (r), and the parameters are presented. Theorem 2 (Massey‑Schöbi) gives a general lower bound on the number of keys for a ((t-1))-fold secure code: (b\ge\lceil v^{t}/k^{t}\rceil). When equality holds the code is called optimal. Theorem 3 shows that any (t)-(v,k,λ) design yields an optimal authentication code with (\lambda\binom{v}{t}/\binom{k}{t}) keys, and conversely, an optimal code with exactly (\binom{v}{t}/\binom{k}{t}) keys implies the existence of a Steiner (t)-(v,k,1) design.

Stinson’s earlier results are revisited: a Steiner 2‑design with (v\mid b) gives an optimal code that is one‑fold secure and perfectly secret. The classic example is the Fano plane (a Steiner 2‑(7,3,1) design) yielding a code with (k=3), (v=7), (b=7). This construction is illustrated with an explicit encoding matrix.

The main contributions are Theorem 6 and Theorem 7. Theorem 6 generalizes Stinson’s construction: if a Steiner (t)-(v,k,1) design satisfies (v\mid b) (where (b=\binom{v}{t}/\binom{k}{t})), then one can order the blocks so that each point appears equally often in each position of the ordered blocks. Using these ordered blocks as encoding rules (each chosen with equal probability) yields an optimal code that is ((t-1))-fold secure and perfectly secret. The proof uses an edge‑coloring of the bipartite point‑block incidence graph to achieve a regular (k)-coloring, guaranteeing the required uniformity.

Theorem 7 provides an infinite family of such codes derived from spherical geometries, specifically from Möbius (or inversive) planes. For any prime power (q) and any even integer (d\ge2), the action of the projective linear group (PGL(2,q^{d})) on the projective line (GF(q^{d})\cup{\infty}) produces a 3‑((q^{d}+1,q+1,1)) design. Because (v=q^{d}+1) divides the number of blocks (b=\binom{v}{3}/\binom{q+1}{3}) when (d) is even, Theorem 6 applies, giving optimal codes with (k=q+1) source states, (v=q^{d}+1) messages, and (\binom{v}{3}/\binom{k}{3}) keys. These codes are two‑fold secure against spoofing and achieve perfect secrecy. An explicit smallest example (q=3, d=2) yields a code with (k=4), (v=10), (b=30), derived from the unique Steiner 3‑(10,4,1) Möbius plane of order 3.

Section VI expands the catalogue of optimal codes. Using known existence results for Steiner 2‑designs (triple systems) when (v\equiv1,3\pmod6) and for Steiner 2‑designs with block size 4 when (v\equiv1,4\pmod{12}), the authors list infinite families of one‑fold secure, perfectly secret codes. They also examine higher‑t designs (t≥3) where the divisibility condition (v\mid b) holds, presenting the first known examples of three‑fold, four‑fold, etc., secure codes with perfect secrecy. Tables summarizing parameters (k, v, b, t) are provided, demonstrating that many of these constructions are novel.

The paper concludes with open problems: extending the framework to non‑uniform source distributions, handling designs with λ>1, exploring algebraic (non‑linear) constructions, and investigating the existence of Steiner designs for t≥6, which would immediately yield higher‑fold secure authentication codes. The authors emphasize that the interplay between combinatorial design theory and cryptographic authentication offers a fertile ground for future research.

Overall, the work systematically builds a bridge between Steiner t‑designs and authentication codes, delivering new infinite families of optimal codes that simultaneously guarantee multi‑fold spoofing resistance and perfect secrecy, and thereby makes a substantial contribution to both combinatorial design theory and information‑theoretic cryptography.