•  
  •  
 

Abstract

The capability of VoIP to provide internet telephony is limited by the lack of homogeneous quality of service (QoS) mechanisms in the Internet. Whereas approaches which reserve QoS resources will work well in an end-to-end managed environment, they are not automatically suited to the heterogeneous nature of the Internet. It may be possible to adopt the 'chirp-sounder' approach uses in establishing the optimal frequency channel for a high frequency (HF) radio transmission which dynamically samples a range of possible transmission channels and uses the echoing of a an established test pattern to ascertain the quality of the potential links. The optimal' channel' can then be selected for transmission. By repeating the process at intervals during the call, transparent handover can be achieved if the current channel deteriorates. This article asks if such an approach can be adapted to suit voice over IP telephony across the internet, specifically in relation to the Session Internet Protocol (SIP). SIP is an Internet-based protocol for establishing real-time end-toend conference calls between peers. It already includes a mechanism, through the Session Description Protocol (SDP), of establishing the lowest common media capability available on both peers, but currently has no mechanism for establishing if the proposed media connection has adequate latency or packet loss peiformance to support real-time voice packets. This article asks if SIP should be extended to include such functionality and proposes the adoption of a client/server based measurement-based approach to control call admission.

DOI

10.21427/D7R74Z

Share

COinS