Providing Media Control to SIP-based IVR Applications: The IVRObject Approach

Abstract

A popular application that runs on a Voice-over-IP network is the Interactive Voice Response (IVR), which provides a way for interacting with an end user's phone via a telephony interface by presenting a set of audible menu options, and collecting the user's Dual Tone Multi-Frequency responses as the user presses the telephone numeric keypad, and recording the user's voice. This thesis takes a closer look into IVR systems on an IP-based network, specifically in networks that support the Session Initiation Protocol (SIP) for controlling the telephony signalling, and more specifically on IVR systems that have their SIP signalling controlled by a SIP Application Server (SIPAS) implementing the SipServlet 1.0 or 1.1 specifications and where a SIP-based Media Server is used to stream IVR media to the end-user. We describe existing ways for supporting an IVR development in a SIP network, and then we propose an alternative way of accomplishing the same task: the IVRObject. With the help of prototypes, the IVRObject is compared with the existing state of the art against three criteria: a) how easy it is to develop, b) how portable the development solution is, c) and how scalable the solution is in order to sustain a high call volume. As a conclusion, it will be highlighted that the IVRObject provides an easy mechanism for development of IVR-based applications running on a SIP AS, that it is portable to different media server vendors, and that it supports a test strategy that can be leveraged to improve software development quality and faster development. This makes the IVRObject a good alternative especially for enterprise-based IVR applications where scalability is less of an issue than in carrier-space applications.