An integrated NTP-RTCP solution to audio skew detection and compensation for VoIP applications

Abstract

The circuit switched POTS (plain old telephone system) preserves the timing relationship between media samples from sender to receiver through use of a common clock. For PC-based Internet multimedia, the existence of separate audio and system clocks on either end-host can introduce significant complications. Much work has taken place in recent years that addresses the issue of system clock skew and its effect on precise delay measurement. In a voice over IP (VoIP) environment, where adaptive buffering techniques are employed, system and audio clock skew can distort both delay measurement and playout control as well as lead to poor buffer performance. This paper presents a high level mechanism to measure and compensate for the skew relationships between system and audio clocks at each end of a multimedia session. The mechanism utilises both the network time protocol (NTP) and the RTP (realtime transport protocol) control protocol or RTCP. Preliminary and positive results are presented from a testbed system and plans for further work are outlined.