The power of flexible slidingwindow: Adaptive network coding inwireless mesh networks

Existing works on wireless intra-session network coding can be classified into Segmented Network Coding (SNC) and Unsegmented Network Coding (UNC). However, neither of them exhibits fine adaptability to different network conditions. On the one hand, SNC waits for the ACK after transmitting a segment of packets. This will clearly delay the segment scheduling and degrade its performance, especially in large scale networks; On the other hand, UNC encodes packets in a sliding window and can be more resilient to feedback delay. However, UNC involves more feedback, and the costs can outweigh the benefits in small scale networks. In this work, we propose Adaptive Network Coding (ANC) to achieve high throughput under different network conditions. It incorporates an adaptive sliding window that enables dynamically adjusting the processing unit according to network conditions. Furthermore, it incorporates feedback reduction and decoding matrix restriction techniques to further improve the practicability. Simulations show that the throughput of ANC outperforms that of SNC and UNC in various scale networks, and validates better adaptability.