[Paper Review] Network Coding as a WiMAX Link Reliability Mechanism

Surat Teerapittayanon, Kerim Fouli+5 more|arXiv (Cornell University)|Aug 23, 2012

Cooperative Communication and Network Coding49 references4 citations

TL;DR

This paper proposes an IP-layer network coding (NC) architecture using systematic intra-session random linear network coding as a packet erasure code to enhance WiMAX link reliability. Evaluated on GENI WiMAX platforms, the NC-based system reduces packet loss to nearly 0% and achieves up to 5.9× higher throughput and 5.5× lower file transfer delay compared to HARQ and ARQ mechanisms.

ABSTRACT

We design and implement a network-coding-enabled reliability architecture for next generation wireless networks. Our network coding (NC) architecture uses a flexible thread-based design, with each encoder-decoder instance applying systematic intra-session random linear network coding as a packet erasure code at the IP layer, to ensure the fast and reliable transfer of information between wireless nodes. Using Global Environment for Network Innovations (GENI) WiMAX platforms, a series of point-to-point transmission experiments were conducted to compare the performance of the NC architecture to that of the Automatic Repeated reQuest (ARQ) and Hybrid ARQ (HARQ) mechanisms. At the application layer, Iperf and UDP-based File Transfer Protocol (UFTP) are used to measure throughput, packet loss and file transfer delay. In our selected scenarios, the proposed architecture is able to decrease packet loss from around 11-32% to nearly 0%; compared to HARQ and joint HARQ/ARQ mechanisms, the NC architecture offers up to 5.9 times gain in throughput and 5.5 times reduction in end-to-end file transfer delay. Our experiments show that network coding as a packet erasure code in the upper layers of the protocol stack has the potential to reduce the need for joint HARQ/ARQ schemes in the PHY/MAC layers, thus offering insights into cross-layer designs of efficient next generation wireless networks.

Motivation & Objective

• Address the growing demand for reliable, high-throughput wireless transmission in next-generation networks, especially for mobile and broadband services.
• Overcome the limitations of traditional ARQ and HARQ mechanisms in wireless environments, which suffer from high retransmission overhead and sensitivity to packet loss and ACK loss.
• Explore the feasibility and performance benefits of deploying network coding as a cross-layer reliability mechanism at the IP layer in WiMAX networks.
• Demonstrate that NC can reduce dependency on joint HARQ/ARQ schemes in PHY/MAC layers, enabling more efficient and robust wireless communication.
• Provide a flexible, thread-based NC architecture that supports concurrent encoding and decoding for low-latency, high-reliability data transfer.

Proposed method

• Implement a thread-based, flexible network coding architecture at the IP layer using systematic intra-session random linear network coding (RLNC) over a Galois field.
• Apply network coding as a packet erasure code to protect data blocks, where each coded packet is a random linear combination of original packets.
• Use Iperf and UDP-based UFTP at the application layer to measure throughput, packet loss, and file transfer delay under controlled point-to-point WiMAX transmissions.
• Conduct experiments on GENI WiMAX platforms, comparing NC against default HARQ (chase combining) and ARQ mechanisms under identical channel conditions.
• Configure redundancy parameters such as $N_r$ (number of redundant packets) and $N_p$ (number of concurrent encoder-decoder threads) to evaluate performance trade-offs.
• Leverage ACK feedback to trigger decoding and ensure recovery of lost packets with minimal delay, using the property that any coded packet can recover any lost original packet in the block.

Experimental results

Research questions

• RQ1 Can network coding at the IP layer outperform traditional ARQ and HARQ mechanisms in terms of throughput and end-to-end delay in a WiMAX environment?
• RQ2 How does network coding reduce packet loss compared to ARQ and HARQ under varying wireless channel conditions?
• RQ3 To what extent does NC reduce retransmission overhead compared to HARQ, which may retransmit packets up to four times?
• RQ4 How does the NC architecture maintain high performance even when ACK packets are lost, given its block-level recovery mechanism?
• RQ5 What is the impact of different redundancy levels ($N_r$) and concurrent thread counts ($N_p$) on the performance of the NC-based system?

Key findings

• The NC architecture reduced average packet loss from 11–32% (under HARQ and ARQ) to nearly 0% across tested scenarios.
• Throughput gains of up to 5.9 times were achieved with NC compared to HARQ and joint HARQ/ARQ mechanisms.
• End-to-end file transfer delay was reduced by up to 5.5 times using NC relative to HARQ and ARQ.
• The NC system incurred significantly lower retransmission overhead—up to 100% for NC versus up to 400% for HARQ due to multiple retransmissions.
• NC demonstrated greater robustness to both data and ACK packet loss, as any coded packet can recover any lost original packet in a block.
• The architecture’s performance was not limited by RTT, and it maintained high efficiency even under high load, avoiding MAC-layer buffer overflows common in ARQ/HARQ.