No abstract

Voice communication in IEEE802.15.4 networks is an attractive application for the emergency management support when a disaster occurs. Although this application can be of extreme utility in the immediate consequence of the disaster, it must take into account the bandwidth limitations of the IEEE802.15.4 standard, which allows the simultaneous transmission of a small number of voice streams. When more voice streams must be transmitted through the network, a bandwidth reduction of each speech flow must be performed. In this paper we first present an algorithm for the perceptual selection of voice data aiming at reducing the speech flow bandwidth while preserving as much as possible the end-to-end speech quality, then we propose a voice data protection technique based on speech perceptual importance and able to preserve speech quality against packet losses. The perceptual selection algorithm can reach a 30.8% reduction in bandwidth occupancy, with respect to a full rate voice communication, still maintaining an end-to-end speech quality between good and fair according to the Mean Opinion Score scale defined by the International Telecommunications Union. The protection technique, jointly adopted with the perceptual selection, can reach better end-to-end speech quality values with respect to a full rate voice communication while requiring a lower amount of transmission bandwidth.

No abstract