Prediction-Based Link-Layer Dynamic Fragmentation and Aggregation for Wireless Mesh Networks

Abstract

Packet retransmissions due to packet losses reduce transmission performance in wireless networks obviously. Multi-hop wireless networks suffer from packet losses more than one-hop wireless networks. High bit-error rate of wireless channel and frequent channel collision due to shared channel are the most important ingredients of packet loss in multi-hop wireless networks. When bit error rate is between 10-3~10-4, fragment size control is efficient to enhance transmission performance. After analyzing the packet loss reason in wireless mesh networks, a prediction-based link-layer dynamic frame fragmentation and aggregation scheme called PDFA is proposed for 802.11-based wireless mesh networks. The size of sending fragment is optimized based on the prediction of channel BER. Simulation experiments validate that PDFA enhances transmission performance in wireless mesh networks efficiently.