Voice packet communication system for private networks

The authors describe the design methodologies and system configuration for a voice packet communication system (voice PAD) which is installed on a high-speed packet switching system for multimedia communications. The voice PAD is experimentally fabricated for voice communications in the packet multiplexing mode between PABXs via digital leased lines. Speech quality during network overload is protected from serious degradation by embedded ADPCM (adaptive differential pulse code modulation) speech coding and by most-significant bits (MSB) and least-significant bits (LSB) packet interleaving. This interleaving reduces the processing load on the switching system processors required to drop LSB speech sample packets following traffic congestion. The techniques presented are based on a fixed-length packet format and are well suited to asynchronous transfer mode environments.<>