[Paper Review] On Deterministic Linear Network Coded Broadcast and Its Relation to Matroid Theory

Mingchao Yu, Parastoo Sadeghi+1 more|arXiv (Cornell University)|Jun 23, 2014

Cooperative Communication and Network Coding19 references3 citations

TL;DR

This paper establishes a direct equivalence between deterministic linear network coded (DLNC) broadcast solutions and matrix matroids, enabling the use of matroid theory to analyze and design optimal DLNC schemes. It proposes a heuristic algorithm based on graphic matroids that efficiently finds near-optimal DLNC solutions over any finite field, achieving performance close to the theoretical lower bound on transmission count, especially for moderate field sizes and receiver counts.

ABSTRACT

Deterministic linear network coding (DLNC) is an important family of network coding techniques for wireless packet broadcast. In this paper, we show that DLNC is strongly related to and can be effectively studied using matroid theory without bridging index coding. We prove the equivalence between the DLNC solution and matrix matroid. We use this equivalence to study the performance limits of DLNC in terms of the number of transmissions and its dependence on the finite field size. Specifically, we derive the sufficient and necessary condition for the existence of perfect DLNC solutions and prove that such solutions may not exist over certain finite fields. We then show that identifying perfect solutions over any finite field is still an open problem in general. To fill this gap, we develop a heuristic algorithm which employs graphic matroids to find perfect DLNC solutions over any finite field. Numerical results show that its performance in terms of minimum number of transmissions is close to the lower bound, and is better than random linear network coding when the field size is not so large.

Motivation & Objective

• To establish a direct theoretical link between deterministic linear network coding (DLNC) and matroid theory, bypassing indirect connections via index coding.
• To determine the minimum number of coded transmissions required for DLNC and its dependence on finite field size.
• To develop an efficient algorithm for generating near-optimal DLNC solutions over any finite field.
• To answer fundamental questions about the existence and feasibility of perfect DLNC solutions.
• To improve upon random linear network coding (RLNC) in terms of transmission efficiency and decoding delay, especially for small field sizes.

Proposed method

• Prove the equivalence between DLNC solutions and matrix matroids, showing that a valid DLNC solution corresponds to a representable matrix matroid over a finite field.
• Use the concept of matroid representability to derive sufficient and necessary conditions for the existence of perfect DLNC solutions (i.e., solutions achieving the theoretical lower bound on transmissions).
• Propose a heuristic algorithm that constructs a graphic matroid from the reception instance to generate a DLNC solution, prioritizing vertex and edge allocations to minimize transmission count.
• Apply the algorithm to the systematic feedback matrix (SFM) after the initial uncoded transmission phase to generate a U×K coding matrix C.
• Use the rank of the coding matrix C and the maximum weight w_max of the reception instance to evaluate solution quality, with U = w_max indicating a perfect solution.
• Evaluate performance through simulations comparing the proposed algorithm with random linear network coding (RLNC) over F₂ and F₈, measuring average U, U = w_max frequency, and U ≤ w_max + 1 frequency.

Experimental results

Research questions

• RQ1 What is the minimum number of coded transmissions required for a perfect DLNC solution in a wireless broadcast setting?
• RQ2 How does this minimum number depend on the choice of finite field size?
• RQ3 Under what conditions does a perfect DLNC solution exist over a given finite field?
• RQ4 Can a systematic algorithm be designed to generate near-optimal DLNC solutions over any finite field?
• RQ5 How does the performance of the proposed algorithm compare to random linear network coding in terms of transmission count and field size efficiency?

Key findings

• The proposed algorithm achieves a perfect DLNC solution (U = w_max) in over 85% of simulation cases under i.i.d. erasure channels with K=15 and Pe=0.2.
• The algorithm produces solutions with U ≤ w_max + 1 in nearly all cases, indicating strong performance close to the theoretical lower bound.
• The average number of transmissions U from the proposed algorithm is negligible compared to the lower bound w_max across all tested receiver counts (N ∈ [5,40]).
• The proposed algorithm outperforms RLNC over F₂ in terms of transmission efficiency, with the performance gap narrowing only when RLNC uses a larger field (F₈).
• The algorithm's performance is robust to increasing receiver count, with only a slight degradation possibly due to increased complexity in the reception instance W.
• The existence of perfect DLNC solutions is not guaranteed over all finite fields, and identifying such solutions remains an open problem in general.