On the 3-Receiver Broadcast Channel with Degraded Message Sets and Confidential Messages

In this paper, bounds to the rate-equivocation region for the general 3-receiver broadcast channel (BC) with degraded message sets, are presented for confidential messages to be kept secret from one of the receivers. This model is more general than the 2-receiver BCs with confidential messages with an external wiretapper, and the recently studied 3-receiver degraded BCs with confidential messages, since in the model studied in this paper, the conditions on the receivers are general and the wiretapper receives the common message. Wyner’s code partitioning combined with double-binning is used to show the achievable rate tuples. Error probability analysis and equivocation calculation are also provided. The secure coding scheme is sufficient to provide security for the 3-receiver BC with 2 or 3 degraded message sets, for the scenarios: (i) 3 degraded message sets, where the first confidential message is sent to receivers 1 and 2 and the second confidential message is sent to receiver 1, (ii) 2 degraded message sets, where one confidential message is sent to receiver 1, and (iii) 2 degraded message sets, where one confidential message is sent to receivers 1 and 2. The proof for the outer bound is shown for the cases where receiver 1 is more capable than the wiretap receiver 3, for the first two scenarios. Under the condition that both receivers 1 and 2 are less noisy than the wiretap receiver 3, the inner and outer bounds coincide, giving the rate-equivocation region for (iii). In addition, a new outer bound for the general 3-receiver BC with 3 degraded messages is obtained.

💡 Research Summary

This paper studies the rate‑equivocation (rate‑security) region of a three‑receiver discrete memoryless broadcast channel (BC) with degraded message sets when one of the receivers acts as a wiretapper. The model extends the classical two‑receiver BC with confidential messages (Csiszár‑Körner) and the more recent K‑receiver BC with an external eavesdropper by allowing three receivers, hierarchical (degraded) messages, and by letting the eavesdropper (receiver 3) also decode the common message. Three message‑set configurations are considered: (i) three degraded sets (common W₀ to all, private W₁ to receivers 1 and 2, and a second private W₂ to receiver 1), (ii) two degraded sets of type 1 (common W₀ to all and a private W₁ to receiver 1), and (iii) two degraded sets of type 2 (common W₀ to all and a private W₁ to receivers 1 and 2).

The authors propose a secure coding scheme that combines Wyner’s code partitioning with the double‑binning technique of Liu et al. The scheme introduces three auxiliary random variables U₁, U₂, U₃. U₁ carries the common message, U₂ carries the first private message (W₁) and is decoded by receivers 1 and 2, while U₃ carries the second private message (W₂) and is decoded only by receiver 1. The encoder maps (W₀,W₁,W₂) → (U₁,U₂,U₃) → Xⁿ, and the three receivers decode the appropriate subsets of the auxiliary variables. By carefully selecting the rates of the binning partitions (denoted R₁′ and R₂′) the scheme guarantees both reliable decoding at the intended receivers and high equivocation at the wiretapper.

The inner (achievable) region is presented in Theorem 1 as a set of information‑theoretic inequalities (6a–6m). The key constraints are:

• The common‑message rate R₀ cannot exceed I(U₃;Y₃).
• The secrecy rate of W₁, R₁ᵉ, is bounded by the minimum of I(U₂;Y₂|U₁)−R₁′ and I(X;Y₁|U₃)−R₁′−R₂′.
• The secrecy rate of W₂, R₂ᵉ, is bounded by I(X;Y₁|U₂)−R₂′.
• The sum secrecy rate R₁ᵉ+R₂ᵉ is limited by I(X;Y₁|U₁)−R₁′−R₂′.

Additional constraints (6g–6m) involve combinations of R₀, R₁, and R₂ and ensure that each receiver’s decoding capability is respected. The joint distribution of (U₁,U₂,U₃,X) is specified in (7) and reflects the hierarchical code construction. A technical condition I(X;Y₃|U₂) ≤ I(X;Y₁|U₂,U₃) (equation 8) is required for the inner bound to hold.

For the converse (outer) bound the paper distinguishes two channel‑ordering conditions:

1. More‑Capable: Receiver 1 is more capable than the wiretapper, i.e., I(X;Y₁|U₂) ≥ I(X;Y₃|U₂). Under this condition the outer bound coincides with the inner bound, establishing optimality for scenarios (i) and (ii).

2. Less‑Noisy: Both receivers 1 and 2 are less noisy than the wiretapper, i.e., for any auxiliary U, I(U;Y₁) ≥ I(U;Y₃) and I(U;Y₂) ≥ I(U;Y₃). When this stronger condition holds, the inner and outer bounds for the three‑degraded‑set case (i) reduce to the known region for the two‑degraded‑set case (iii). Consequently, the rate‑equivocation region is completely characterized for type 2 configurations.

The authors also derive a new outer bound for the general three‑receiver BC with three degraded messages (without secrecy constraints), which was not available in the earlier works of Nair and El Gamal.

The paper provides detailed error‑probability analysis using typicality arguments and computes equivocation rates by bounding H(W₁|Y₃ⁿ) and H(W₂|Y₃ⁿ). The proofs rely on standard information‑theoretic tools such as the chain rule, Markov properties, and the Csiszár sum identity.

From an application perspective, the results are relevant to multimedia broadcasting where a base layer (W₀) is intended for all users, enhancement layers (W₁, W₂) are delivered to subsets of users, and the system must guarantee that an eavesdropping user (who can decode the base layer) learns nothing about the enhancement layers. The coding scheme shows that, even when the eavesdropper receives the common content, high secrecy rates for the private layers are achievable, provided the channel ordering conditions are satisfied.

In summary, the paper makes four principal contributions:

1. Introduces a novel secure coding scheme that merges Wyner’s code partitioning with double‑binning to handle two confidential messages in a three‑receiver degraded BC.
2. Derives an inner rate‑equivocation region (Theorem 1) expressed through a comprehensive set of mutual‑information inequalities.
3. Establishes outer bounds under “more‑capable” and “less‑noisy” channel conditions, showing optimality (inner = outer) for several important configurations.
4. Provides a new outer bound for the non‑secure three‑receiver BC with three degraded messages, extending the known capacity results for this class of channels.

These findings broaden the theoretical understanding of physical‑layer security in broadcast networks with hierarchical services and lay the groundwork for practical secure multicast and layered video streaming systems.