More Capable Broadcast Channels

• More capable broadcast channels are defined by a condition where one receiver always has higher mutual information than the other, establishing a clear ordering across all input distributions.
• Optimal strategies such as superposition coding leverage a base layer for the weaker receiver and a refined satellite layer for the stronger one, ensuring capacity achievement even under cooperative settings.
• Extensions like essentially more capable channels and quantitative advantage variants provide practical measures to assess performance gaps, which are instrumental in designing robust multi-user networks.

A more capable broadcast channel is a two-receiver discrete memoryless broadcast channel in which one receiver is universally superior in mutual information for all input distributions: for every $p(x)$, $I(X; Y_1) \ge I(X; Y_2)$. This ordering induces a powerful structure on capacity regions, coding strategies, channel simulation, and channel comparison, and lies at the heart of several major advances in multi-user information theory. Recent developments have extended and refined the more capable notion, producing generalizations such as essentially more capable channels, reversely more capable product structures, and quantitative advantage variants.

1. Classical and Generalized Notions of More Capable Channels

The standard definition (Komlós–Csiszár–Körner) for a broadcast channel $p(y_1, y_2 \mid x)$ declares receiver $Y_1$ more capable than $Y_2$ if for all $p(x)$:

$I(X; Y_1) \ge I(X; Y_2).$

This global mutual-information criterion implies that any receiver-specific coding or decoding strategy targeting $Y_2$ can be matched or outperformed for $Y_1$. A key generalization is the essentially more capable channel: there exists a sufficient subset $\mathcal{P}$ of distributions such that, for all $p(x) \in \mathcal{P}$ and all $U \to X \to (Y_1, Y_2)$,

$I(X; Y_2 | U) \le I(X; Y_1 | U).$

This relaxes the universal requirement while preserving information-theoretic sufficiency for capacity characterization (0901.0595).

A further quantitative extension is the notion of "more capable with advantage": $W$ is more capable than $\widetilde W$ with advantage $\eta$ if

$I(X; Y) + \eta \ge I(X; Z) \quad \forall\, P_X,$

defining a scalar gap when strict comparability fails. The minimal $\eta$ for a pair $(W, \widetilde W)$ quantifies "how much" more capable one channel is over the other (Kougang-Yombi et al., 2024).

2. Capacity Regions: Characterizations and Unified Schemes

For more capable channels, single-letter characterizations of the capacity region are attainable. Without receiver message side information, the region consists of all rate pairs $(R_1, R_2)$ for which there exists some $U \to X \to (Y_1, Y_2)$ and $p(x)$ such that

$$\begin{align*} R_2 &\le I(U; Y_2), \ R_1 + R_2 &\le I(U; Y_2) + I(X; Y_1 | U), \ R_1 + R_2 &\le I(X; Y_1). \end{align*}$$

The superposition code is optimal: a "base" layer $U$ supports the weaker receiver, and a "satellite" layer $X|U$ carries refinement for the stronger receiver (0901.0595, Sutter et al., 2013, Asadi et al., 2015).

With receiver message side information, the region generalizes. For five messages $(M_1, \dots, M_5)$ and appropriate auxiliary $U$, the region is the set of non-negative $(R_1, \ldots, R_5)$ such that:

$$\begin{align*} R_1 + R_3 + R_5 &< I(U; Y_2), \ R_1 + R_2 + R_3 + R_5 &< I(U; Y_2) + I(X; Y_1|U), \ R_1 + R_2 + R_3 + R_4 &< I(X; Y_1). \end{align*}$$

The specialized Marton (multi-auxiliary) region collapses to this form under the more capable property (Asadi et al., 2015).

For product broadcast channels with reversely more-capable order, the capacity region is described by a set of six inequalities reflecting a product structure, involving minima over mutual informations for each component (Geng et al., 2011).

3. Structural Theory and Partial Orders

The more capable relation forms a partial order across the space of channels, especially well-behaved in the class of binary-input, symmetric-output (BISO) channels. In this setting, the preorder is governed by Lorenz-curve majorization: $Y \gg Z$ if and only if $F_Y(t) \le F_Z(t)$ for $t \in [0,1]$, where $F(\cdot)$ is the Lorenz curve associated to each channel (Geng et al., 2010). This provides a direct, computable criterion for comparability.

Key consequences:

• Every BISO channel $F$ with capacity $C$ sits between the binary erasure channel (BEC) and the binary symmetric channel (BSC): $\mathrm{BEC}(C) \gg F \gg \mathrm{BSC}(C)$.
• For BISO channels with outputs of size $\le 3$, comparability is always guaranteed.
• In BISO, more capable coincides (in the opposite direction) with the essentially less-noisy ordering.

A table summarizing relationships for BISO channels:

Condition	Implication	References
$F_Y(t) \le F_Z(t)$ for all $t$	$Y \gg Z$	(Geng et al., 2010)
BEC$(C)\gg F \gg$ BSC$(C)$	Channel hierarchy	(Geng et al., 2010)
More capable in BISO	Inverse of essentially less-noisy	(Geng et al., 2010)

4. Coding Schemes and Universality

Superposition coding is the fundamental and capacity-achieving strategy for more capable broadcast channels: a cloud center $U$ is chosen for the weak receiver, and a satellite code $X|U$ is superimposed for the strong receiver. In the presence of cooperation (e.g., rate-limited link from strong to weak), optimal capacity points are achievable by augmenting the superposition code with decode-and-forward at the strong receiver (Gouic et al., 13 Jan 2026).

Polar codes exhibit universality under the more capable relation. A polar code designed for the weaker receiver (e.g., channel $Z$) can—with only $o(n)$ changes to its frozen set—be reliably used for any stronger channel $Y$ with $P_{Y|X} \gg P_{Z|X}$, preserving rate and complexity (Sutter et al., 2013).

Cooperative settings admit explicit capacity-benefit expressions:

• In the Gaussian case, the region is parameterized by a convex combination matching power or rate split, with cooperation extending the boundary up to a precise threshold (Gouic et al., 13 Jan 2026).
• For BEC–BSC, the analogous parametric region uses binary entropy and mixture identities.

Strategies involving compress-and-forward or partial decode-forward provide no additional benefit in this class; decode-and-forward (bin notification) suffices (Gouic et al., 13 Jan 2026).

5. Quantitative and Product-Channel Generalizations

The "more capable with advantage" concept delivers a one-parameter family of orderings. The minimal advantage parameter $\eta_{\rm mc}(W, \widetilde W)$ required for $W + \eta \succ_{\rm mc} \widetilde W$ is additive:

$$\eta_{\rm mc}(W^n, \widetilde W^n) = n\,\eta_{\rm mc}(W, \widetilde W).$$

This enables operational finite-blocklength trade-offs, supporting, for example, list decoding on $\mathrm{BSC}_p$ using codes for $\mathrm{BEC}_q$ at an explicit mutual information gap cost (Kougang-Yombi et al., 2024).

Product channels with mixed (reversely more-capable) components enable new matching inner and outer bounds. These match Marton's region exactly and are not captured by standard single-letter UV outer bounds, which are proven strictly suboptimal for certain examples (Geng et al., 2011).

6. Connections to Other Orderings and Gaps in Bounds

More capable is strictly weaker than degradedness but stronger than the less-noisy relation. In many cases (BISO, degraded, essentially more capable), superposition coding achieves the capacity. For general broadcast channels not satisfying these relations, Marton's region is not tight, and there is a strict separation between known inner and outer bounds (Geng et al., 2010, Geng et al., 2011).

A salient feature: when two BISO receivers are not more-capable comparable, both Marton's region and the best outer bounds are non-tight—demonstrated by explicit Lorenz curve analysis and auxiliary-variable optimizations.

7. Operational Significance and Applications

The more capable structure simplifies both theoretical and practical aspects of broadcast channel communication:

• Any code designed for the weaker receiver suffices (with possibly negligible modifications) for the stronger—enabling universal designs (Sutter et al., 2013).
• In multi-user and compound settings or where channel realization varies within a more capable class, robust capacity-achieving schemes can be constructed without continual re-optimization.
• In cooperative communication, the addition of a single unidirectional link is efficiently exploited via superposition and bin notification (Gouic et al., 13 Jan 2026).
• The "advantage" generalization provides a continuous measure for how far a code construction designed for one channel can exceed or fall short for another, directly informing code design and blocklength-resource allocation (Kougang-Yombi et al., 2024).

The modern theory thus offers a complete taxonomy, explicit capacity and code constructions, rigorous orderings, and practical design prescriptions for classes and generalizations of more capable broadcast channels.