The Capacity Region for Classes of Sum-Broadcast Channels

Abstract: We compute the capacity region of a sum of broadcast channels whose components are degraded, less-noisy, more-capable, deterministic, or semi-deterministic. We achieve this by showing that an auxiliary-receiver outer bound, previously introduced by some of the authors, matches Marton's inner bound. This result generalizes a previously known result for the sum of two reversely degraded broadcast channels due to El Gamal (1980). Moreover, we define a class of primary broadcast channels and show an analogous result for the sum of primary broadcast channels.

• The paper determines capacity regions for sum-broadcast channels by proving the tightness of Marton's inner bound when both component channels are primary.
• It introduces an auxiliary-receiver outer bound that overcomes limitations of classical UVW bounds, yielding explicit single-letter capacity expressions.
• It demonstrates with numerical examples that traditional bounds can be strictly loose, highlighting the need for novel converse techniques in broadcast networks.

Capacity Region Characterization for Classes of Sum-Broadcast Channels

Overview

The paper "The Capacity Region for Classes of Sum-Broadcast Channels" (2606.12839) addresses the long-standing open problem of characterizing capacity regions for non-trivial multi-receiver broadcast channels. Specifically, it rigorously determines the capacity region for sum-broadcast channels whose component channels belong to the degraded, less-noisy, more-capable, deterministic, or semi-deterministic classes. The methodology leverages an auxiliary-receiver outer bound, improving upon the classical UVW outer bound and Marton's inner bound, and resolves cases where these standard bounds do not coincide.

Background and Existing Results

A discrete-memoryless broadcast channel (DM-BC) with two receivers is modeled by a transition probability $T_{YZ|X}$. The sender transmits private messages $M_1$, $M_2$, and a common message $M_0$ to receivers $Y$ and $Z$. Marton's inner bound provides the largest known achievable rate region for such channels, but tight outer bounds remain elusive outside several special classes.

Standard results show that for degraded, less-noisy, more-capable, and semi/deterministic channels, Marton's inner bound matches the capacity region, coinciding with the UVW outer bound [nai11arx] and earlier results [bergmans73, gallager74]. However, for sum or product constructions of broadcast channels, these classical bounds do not necessarily coincide, and new analytical tools become necessary.

Sum-Broadcast Channels and Primary Classes

A sum-broadcast channel $T = T_a \oplus T_b$ is defined such that $T$ acts as $T_a$ or $T_b$ on disjoint portions of the input-output alphabet. El Gamal [g80] previously established the capacity for sums of reversely degraded channels, but this work deeply generalizes to broader classes, introducing two primary classes of broadcast channels: $M_1$0 and $M_1$1.

A channel is primary if for all convex combinations (parametrized by $M_1$2) and all non-negative $M_1$3, the carefully defined weighted-sum versions of Marton's and the outer bounds coincide. The paper rigorously proves that all less-noisy and semi-deterministic channels are primary; more-capable channels may or may not be, depending on the parameter regime. Also, MIMO Gaussian broadcast channels are shown to be primary by exploiting Gaussian extremality and Costa's DPC [gn14, costa1983writing], establishing the tightness of Marton's region in these cases.

Sharpness of UVW and Auxiliary-Receiver Outer Bounds

A central claim is the exhibition of sum-broadcast channels for which the UVW outer bound strictly exceeds Marton's inner bound, thus is strictly loose. For example, for a sum of two reversely semi-deterministic channels, the UVW outer bound allows a sum-rate of $M_1$4 while Marton's bound is $M_1$5—a clear strict gap. This demonstrates that traditional bounding approaches are fundamentally insufficient for the sum-channel problem, necessitating sharper techniques.

To overcome this, the paper invokes the auxiliary-receiver approach previously developed for the relay and interference channels [gn22], constructing an outer bound that, in several challenging cases, provably matches Marton's inner bound for the considered sum channels. This new outer bound incorporates additional auxiliary random variables, tailored to the sum-structure, and allows matching to the inner bound where classical arguments fail.

Technical Summary of Main Results

The main theorem establishes the following for sum-broadcast channels $M_1$6:

• If $M_1$7 and $M_1$8 are both primary channels,
  • The exact (convex) capacity region is determined:

  $M_1$9

  where $M_2$0 is Marton's region and $M_2$1 is the auxiliary-receiver outer bound.

• Implication: Marton's inner bound can be strictly loose for sum-channels if the components are not both primary. The auxiliary-receiver outer bound, however, is always tight for sums where the components are from the characterized classes.
• Closure properties: The class of primary channels is closed under taking sums for less-noisy, more-capable, and semi-deterministic components, but not in general for arbitrary more-capable channels.
• Explicit Characterization: The paper provides explicit formulas for the maximal weighted sum-rate over Marton's region and the auxiliary-receiver bound, via minimization over $M_2$2 and maximization over convex combinations, yielding closed-form single-letter expressions for practical computation.

Contradictory and Strong Claims

• Bold claim: The UVW outer bound is insufficient even for sum-broadcast channels whose components are semi-deterministic; only the auxiliary-receiver outer bound is able to match the achievable region in these cases.
• Sharp numerical example: For a sum-channel constructed from reversely semi-deterministic components, Marton's bound gives $M_2$3 sum-rate while the UVW bound gives $M_2$4—demonstrating a nontrivial and provable gap.

Implications and Future Directions

Theoretical: The methodology sharpens our understanding of the interplay between channel structure and achievable regions in network information theory. The auxiliary-receiver approach represents a robust new paradigm for converse proofs in multi-terminal settings, unifying and generalizing several previous results.

Practical: The explicit computability of capacity for sum-broadcasts with primary components is of practical significance. For wireless systems using time/band/domain-multiplexed resources (sum-channels), direct computation of capacity is now possible in more general scenarios.

Future Work: The success of auxiliary-receiver bounds motivates exploring higher-dimensional sum or composition operations, deeper nonprimary constructions, as well as capacity regions for further generalized broadcast or multiaccess networks. Extension to networks with confidential messages or feedback is a natural progression.

Conclusion

The paper rigorously delineates the capacity region for a wide range of sum-broadcast channels by establishing the tightness of Marton's inner bound combined with the auxiliary-receiver outer bound, generalizing all previously known results in this area. It also demonstrates, both analytically and numerically, explicit separation between classical outer bounds and the true capacity in new, nontrivial cases, showcasing the necessity and effectiveness of unconventional converse techniques for broadcast networks (2606.12839).