802.11 MAC Protocol with Selective Error Detection for Speech Transmission

The IEEE 802.11 standard currently does not offer support to exploit the unequal perceptual importance of multimedia bitstreams. All packets affected by channel errors, in fact, are simply discarded, irrespective of the position and percentage of corrupted bits. The objective of this paper is to investigate the effect of bit error tolerance in WLAN speech communications. More specifically, we introduce QoS support for sensitive multimedia transmissions by differentiating the scope of the standard MAC error detection step in order to discard multimedia packets only if errors are detected in the most perceptually sensitive bit class. Speech transmission using the GSM-AMR speech coding standard is simulated using a set of experimental bit-error traces collected in various channel conditions. Perceived speech quality, measured with the ITU-T P.862 (PESQ) algorithm, is consistently improved with respect to standard link layer technique. In other words, the results show that the negative effect of errors in the less perceptually important bits is clearly counterbalanced by the lower number of speech packets discarded because of retransmission limits. In fact, the number of received packets is consistently doubled throughout all the simulation conditions with quality gains that reach 0.4 points of the MOS scale in noisy scenarios.