Priority-Based Voice Segmentation and Transmission in Quality-Driven Wireless Audio Sensor Networks
Okan Turkes and Sebnem Baydere
Ubiquitous sensing and transmission of audio signals is a recent challenging research area in Wireless Multimedia Sensor Networks (WMSNs). Validity of gathered signals at any point of the network strictly relates with Quality of Service (QoS) characteristics. For streaming an audio type, considerable amount of information must be preserved. We investigate the impact of error concealment and threshold-based prioritization for preserving quality in voice transmission over Wireless Audio Sensor Networks (WASNs). We propose a lightweight voice encoding scheme, called PrSERVE, to provide robust delivery, error resilience, and data quality conservation. Considering an activity monitoring scenario, we find out reasonable parameters for the presented schemes through a real homogeneously constructed test-bed. We extend our real experiments with sensitivity and random sampling simulation tests. We verify the soundness of our work with an objective voice quality assessment and analysis. Our results indicate that PrSERVE provides a significant voice quality enhancement with a suitable network design in WASNs.
Keywords: Audio sensor networks, wireless multimedia sensor networks, voice quality assessment, prioritization, error correction