Bit error correction without redundant data: a MAC layer technique for 802.11 networks

In wireless network monitoring, received packets such as MAC frames may contain bit errors. Such errors either makes the monitoring data less accurate, or forces the monitoring software to drop errored frames to avoid such inaccuracy. We present a MAC layer based measurement technique for monitoring 802.11 networks. It can correct bit errors in certain MAC headers without the use of FEC or redundant data. This allows greater accuracy and/or extended range in network monitoring even when signal reception is less than ideal. This technique has several other applications. First, it provides better support for cross-layer protocols such as UDP-Lite, because a MAC frame with a corrupted destination MAC address would have been discarded in the OS stack even with UDP-Lite enabled. Second, the technique allows blind estimation of bit error rate in the received MAC frame, so that a multimedia decoder can decide to whether perform soft-decoding on the corrupted media payload, or to rely on previous media packet for error concealment. Finally, it facilitates the use of NAK frames, which could improve 802.11 MAC layer performance if such modifications are adopted. We evaluate our technique in an 802.11 testbed, quantifying its performance gain over a normal 802.11 monitoring application.