Low-complexity perceptual packet marking for speech transmission over tiny mote device

Multimedia applications inWireless Sensor Networks (WSNs) require different approaches with respect to traditional networks due to the adopted devices' limitations in energy consumption and computational capabilities. This paper proposes a low-complexity algorithm for selecting speech packets according to their perceptual importance for voice transmission over WSNs. The proposed algorithm allows wireless sensor nodes to select which packets to protect during transmission in order to increase speech quality while at the same time minimizing the necessary energy. Experimental results based on cooperative transmission among devices show that the proposed algorithm achieves a good speech quality level while reducing the need to protect packets by 40% when compared to random selection.