[Paper Review] Practical Inner Codes for Batched Sparse Codes in Wireless Multihop Networks

Zhiheng Zhou, Congduan Li+2 more|arXiv (Cornell University)|Sep 2, 2017

Cooperative Communication and Network Coding15 references3 citations

TL;DR

This paper proposes a practical optimization framework for inner codes in batched sparse (BATS) codes within multi-hop wireless networks, minimizing total transmitted packets via centralized and decentralized real-time strategies. By formulating the problem as a mixed-integer nonlinear program (MINLP) and deriving a tight upper bound using the incomplete beta function, the authors achieve near-optimal performance with minimal gap to the theoretical bound, significantly improving transmission efficiency and energy savings.

ABSTRACT

Batched sparse (BATS) code is a promising technology for reliable data transmission in multi-hop wireless networks. As a BATS code consists of an outer code and an inner code that typically is a random linear network code, one main research topic for BATS codes is to design an inner code with good performance in transmission efficiency and complexity. In this paper, this issue is addressed with a focus on the problem of minimizing the total number of packets transmitted by the source and intermediate nodes. Subsequently, the problem is formulated as a mixed integer nonlinear programming (MINLP) problem that is NP-hard in general. By exploiting the properties of inner codes and the incomplete beta function, we construct a nonlinear programming (NLP) problem that gives a valid upper bound on the best performance that can be achieved by any feasible solutions. Moreover, both centralized and decentralized real-time optimization strategies are developed. In particular, the decentralized approach is performed independently by each node to find a feasible solution in linear time with the use of look-up tables. Numerical results show that the gap in performance between our proposed approaches and the upper bound is very small, which demonstrates that all feasible solutions developed in the paper are near-optimal with a guaranteed performance bound.

Motivation & Objective

• Address the challenge of minimizing total packet transmissions in multi-hop wireless networks using BATS codes, which are critical for energy efficiency and network lifetime.
• Formulate the inner code optimization problem as a mixed-integer nonlinear programming (MINLP) problem, which is NP-hard and difficult to solve directly.
• Develop a nonlinear programming (NLP) relaxation using properties of inner codes and the incomplete beta function to compute a tight upper bound on optimal performance.
• Design both centralized and decentralized real-time optimization strategies to enable scalable, low-complexity implementation at each node.
• Demonstrate that the proposed methods achieve near-optimal performance with a very small gap to the theoretical upper bound, ensuring practical reliability and efficiency.

Proposed method

• Model the inner code optimization as a mixed-integer nonlinear programming (MINLP) problem to minimize the expected number of transmissions from source to destination.
• Leverage the properties of random linear network coding and the incomplete beta function to derive a nonlinear programming (NLP) relaxation that provides a valid upper bound on the optimal solution.
• Propose a centralized optimization strategy that solves the NLP relaxation globally using the derived upper bound as a performance benchmark.
• Design a decentralized real-time algorithm where each node independently computes its coding strategy using precomputed look-up tables, achieving linear-time complexity per node.
• Use belief propagation decoding at the destination to recover source messages efficiently, relying on the rank distribution of received batches.
• Validate the performance of the proposed algorithms through numerical experiments under varying batch sizes and hop counts, comparing transmission efficiency and average rank.

Experimental results

Research questions

• RQ1 How can the total number of transmitted packets in BATS-coded multi-hop wireless networks be minimized to improve energy efficiency and network lifetime?
• RQ2 What is the theoretical upper bound on the minimum number of transmissions, and how can it be computed efficiently using the incomplete beta function?
• RQ3 Can a decentralized, real-time optimization strategy be designed that enables each node to independently compute its coding parameters with low complexity?
• RQ4 How close do the proposed centralized and decentralized strategies perform relative to the theoretical upper bound in terms of transmission efficiency?
• RQ5 How does the average rank of the transfer matrix at the destination vary with hop count and batch size under the proposed algorithms, and what does this imply for batch generation control?

Key findings

• The proposed decentralized real-time algorithm achieves near-optimal performance with a very small gap to the theoretical upper bound, demonstrating strong practical feasibility.
• Numerical results show that the relative gap between the proposed approaches and the upper bound is consistently small across different batch sizes (M=12,16,20,24) and hop counts, confirming near-optimality.
• The average rank of the transfer matrix at the destination remains close to the batch size M under the proposed RLT-based algorithm, indicating stable and predictable decoding performance.
• The average rank fluctuation is significantly smoother under the proposed method compared to the original BATS code, which improves predictability in batch generation control.
• Despite requiring slightly more transmissions than the original BATS code for the same M, the proposed method achieves higher transmission efficiency overall, especially when M is small.
• The results suggest that real-time, locally adaptive coding decisions based on current network state lead to better system-wide performance than static or global-optimization-only approaches.