Packet Aggregation May Harm Batched Network Coding

TL;DR

This work investigates how batched network coding (BNC) interacts with UDP-Lite when aggregating BNC packets to reduce protocol overhead. It introduces a frame-efficiency framework that models frame size, header integrity, and BNC batch dependencies, and analyzes how the aggregation level $N$ affects both per-hop efficiency and end-to-end throughput. The key finding is that larger aggregation is not universally better: local efficiency can improve at a hop-by-hop level, but end-to-end throughput may suffer in long networks, and naive cross-layer integration can be detrimental. The results call for carefully designed cross-layer strategies when combining BNC with UDP-Lite and payload aggregation, highlighting directions for robust optimization in multi-hop networks.

Abstract

Batched network coding (BNC) is a solution to multi-hop transmission on networks with packet loss. To be compatible with the existing infrastructure, BNC is usually implemented over UDP. A single error bit will probably result in discarding the packet. UDP-Lite is a variant of UDP that supports partial checksums. As long as the data covered by the checksum is correct, damaged payload will be delivered. With UDP-Lite, we can cope with other techniques such as payload aggregation of BNC packets to reduce the protocol overhead, and forward error correction to combat against bit errors. Unlike traditional transmissions, BNC has a loss resilience feature and there are dependencies between BNC packets. In this paper, we conduct a preliminary investigation on BNC over UDP-Lite. We show that aggregating as much as we can is not always the best strategy, and a hop-by-hop distributed efficiency optimization approach may lead to a worse throughput compared with the scheme without aggregation in a long network. These unnatural results caution that a casual integration of techniques with BNC can be harmful, and give us hints on future research directions.

Figures (11)

• Figure 1: An example of BNC header and BNC packet. The number of data fields in a BNC header varies from protocol to protocol.
• Figure 2: An illustration of sending batches sequentially. Each rectangle is a batch, and each square is a BNC packet.
• Figure 3: A frame with FCS and $2$ aggregated BNC packets over UDP-Lite.
• Figure 4: Frame efficiency when $\mathsf{PLR} = 10\%$ and $20\%$ respectively, with $M = 4$ and $K = 256$. Note that the frame efficiency is not monotonically increasing.
• Figure 5: Frame efficiency when $\mathsf{PLR} = 25\%$, $M = 4$ and $K = 110$. A sudden drop can be observed at $N = 76$, which makes the optimal $N = 75$.