[Paper Review] On the Packet Decoding Delay of Linear Network Coded Wireless Broadcast

Mingchao Yu, Alex Sprintson+1 more|arXiv (Cornell University)|Feb 8, 2018

Cooperative Communication and Network Coding25 references3 citations

TL;DR

This paper proposes Hyper-graphic Linear Network Coding (HLNC), a novel throughput-optimal and average packet decoding delay (APDD)-minimizing technique for wireless broadcast using linear network coding. By modeling receiver packet reception states as a hypergraph and applying hypergraph coloring, HLNC enables early decodings and achieves significantly lower APDD than RLNC and IDNC, even under intermittent feedback, with simulations showing consistent performance gains across settings.

ABSTRACT

We apply linear network coding (LNC) to broadcast a block of data packets from one sender to a set of receivers via lossy wireless channels, assuming each receiver already possesses a subset of these packets and wants the rest. We aim to characterize the average packet decoding delay (APDD), which reflects how soon each individual data packet can be decoded by each receiver on average, and to minimize it while achieving optimal throughput. To this end, we first derive closed-form lower bounds on the expected APDD of all LNC techniques under random packet erasures. We then prove that these bounds are NP-hard to achieve and, thus, that APDD minimization is an NP-hard problem. We then study the performance of some existing LNC techniques, including random linear network coding (RLNC) and instantly decodable network coding (IDNC). We proved that all throughput-optimal LNC techniques can approximate the minimum expected APDD with a ratio between 4/3 and 2. In particular, the ratio of RLNC is exactly 2. We then prove that all IDNC techniques are only heuristics in terms of throughput optimization and {cannot guarantee an APDD approximation ratio for at least a subset of the receivers}. Finally, we propose hyper-graphic linear network coding (HLNC), a novel throughput-optimal and APDD-approximating LNC technique based on a hypergraph model of receivers' packet reception state. We implement it under different availability of receiver feedback, and numerically compare its performance with RLNC and a heuristic general IDNC technique. The results show that the APDD performance of HLNC is better under all tested system settings, even if receiver feedback is only collected intermittently.

Motivation & Objective

• To characterize and minimize the average packet decoding delay (APDD) in linear network coded wireless broadcast under random packet erasures.
• To prove that APDD minimization is NP-hard, establishing theoretical limits on performance.
• To evaluate existing LNC techniques—RLNC and IDNC—on their APDD approximation ratios and throughput optimality.
• To design a novel LNC technique, HLNC, that achieves both throughput optimality and low APDD through hypergraph modeling of receiver state.
• To validate HLNC's performance under various feedback conditions, including intermittent feedback.

Proposed method

• Derives closed-form lower bounds on expected APDD for all LNC techniques under random erasures.
• Proves that minimizing APDD is NP-hard by reducing from the hypergraph coloring problem.
• Models receiver packet reception states as a hypergraph, where vertices represent receivers and hyperedges represent common missing packets.
• Applies hypergraph coloring to select coded packets that maximize the number of receivers able to decode immediately.
• Implements HLNC with both full and intermittent receiver feedback to assess robustness.
• Compares HLNC with RLNC and heuristic IDNC via numerical simulations across diverse system settings.

Experimental results

Research questions

• RQ1 What is the theoretical lower bound on average packet decoding delay (APDD) for any linear network coding technique under random erasures?
• RQ2 Can throughput-optimal LNC techniques approximate the minimum APDD within a bounded ratio, and what is that ratio?
• RQ3 Are instantly decodable network coding (IDNC) techniques capable of approximating optimal APDD, even when throughput is suboptimal?
• RQ4 Can a novel LNC scheme be designed to simultaneously achieve throughput optimality and low APDD through hypergraph modeling?
• RQ5 How does the performance of the proposed HLNC technique scale under limited or intermittent receiver feedback?

Key findings

• All throughput-optimal LNC techniques, including RLNC, can approximate the minimum expected APDD within a ratio between 4/3 and 2, with RLNC achieving exactly a 2-approximation.
• IDNC techniques are not APDD-approximation techniques, as they cannot guarantee bounded APDD performance for at least a subset of receivers.
• The proposed HLNC technique achieves significantly lower APDD than RLNC and heuristic IDNC across all tested system configurations, even with intermittent feedback.
• HLNC maintains throughput optimality while enabling early decodings through hypergraph-based coding decisions.
• Theoretical analysis confirms that APDD minimization is NP-hard, justifying the need for approximation techniques like HLNC.
• Numerical results show that HLNC's APDD performance is consistently better than RLNC and IDNC, with gains preserved under feedback constraints.