[Paper Review] Network Coding Capacity: A Functional Dependence Bound

Satyajit Thakor, Alex Grant+1 more|ArXiv.org|Jan 29, 2009

Cooperative Communication and Network Coding14 references4 citations

TL;DR

This paper proposes a new, easily computable outer bound for the multi-source network coding capacity region using functional dependence graphs to systematically identify information-theoretic cuts. By characterizing all functional dependencies in the network—especially cyclic ones—it derives a tighter bound than existing methods like network sharing or information dominance, leveraging polymatroidal pseudo-entropy functions and maximal irreducible sets to bound total source rates by edge capacities.

ABSTRACT

Explicit characterization and computation of the multi-source network coding capacity region (or even bounds) is long standing open problem. In fact, finding the capacity region requires determination of the set of all entropic vectors $Γ^{*}$, which is known to be an extremely hard problem. On the other hand, calculating the explicitly known linear programming bound is very hard in practice due to an exponential growth in complexity as a function of network size. We give a new, easily computable outer bound, based on characterization of all functional dependencies in networks. We also show that the proposed bound is tighter than some known bounds.

Motivation & Objective

• Address the long-standing challenge of computing the multi-source network coding capacity region, which is intractable due to the complexity of characterizing entropic vectors Γ*.
• Overcome the practical infeasibility of existing linear programming bounds that grow exponentially with network size.
• Develop a systematic method to identify all functional dependencies in networks, including cyclic dependencies, to derive tighter capacity bounds.
• Provide a computationally efficient alternative to the infeasible LP-based bounds while maintaining theoretical rigor.
• Demonstrate that the proposed bound is tighter than known bounds such as network sharing and information dominance.

Proposed method

• Define pseudo-variables and pseudo-entropy functions to generalize entropy functions, enabling representation of network coding constraints without requiring a joint probability distribution.
• Introduce a functional dependence graph (FDG) that models local dependencies between pseudo-variables due to encoding and decoding constraints, extending prior work to handle cycles.
• Develop an algorithm to compute all maximal irreducible sets in the FDG that do not contain source variables, identifying critical information blockers.
• Use subadditivity of polymatroidal pseudo-entropy functions to bound the sum of source entropies by the total capacity of edges in each maximal irreducible set.
• Formulate the functional dependence bound as the minimum over all such edge capacity sums, providing a computable outer bound on total source rates.
• Apply the bound to networks by identifying all maximal irreducible sets and computing the minimum capacity sum across them.

Experimental results

Research questions

• RQ1 Can a systematic method be developed to identify all functional dependencies in a network coding system, including those arising from cyclic dependencies?
• RQ2 Is it possible to derive a computable outer bound for the multi-source network coding capacity region that is tighter than existing bounds like network sharing or information dominance?
• RQ3 How can the structure of functional dependence graphs be leveraged to identify information-theoretic cuts that limit total source throughput?
• RQ4 Can the proposed bound be computed efficiently in practice, avoiding the exponential blowup of traditional linear programming approaches?
• RQ5 Does the functional dependence bound achieve tightness in known cases, such as single-source multicast networks?

Key findings

• The proposed functional dependence bound is tighter than the network sharing bound and bounds based on information dominance, as it considers all functional dependencies rather than subsets.
• For the butterfly network, the bound yields a minimum capacity sum across multiple edge sets, explicitly showing tighter rate constraints than simpler bounds.
• The bound reduces to the standard max-flow bound in single-source multicast networks, confirming its tightness in well-understood cases.
• The method efficiently computes all maximal irreducible sets in the functional dependence graph, enabling systematic identification of information blockers.
• The bound is computationally feasible, avoiding the exponential complexity of full LP-based bounds by focusing on structural dependencies.
• The approach provides a practical alternative to the infeasible LP formulation by replacing the need for full Γ* characterization with a tractable graph-based method.