CAREER: Internet Telephony Services and Mobile Multimedia Services

Columbia University, New York NY

Investigators

Abstract

The proposed research project is focused on Internet telephony and novel ad-hoc data distribution techniques. In the area of Internet telephony research, the research proposes to focus on the following topics: Programming Internet telephony services: The research proposes new mechanisms for programming In-ternet telephony services, usable by both end users and service providers. The mechanisms are designed to be robust, offer predictable performance and allow a large set of users to safely share server resources. The programming facilities are to be enhanced with a graphical service creation environment for Internet telephony. Feature interaction: The research plans to investigate the problem of feature interaction in Internet telephony, as the feature interaction problem differs significantly from that encountered in traditional telephony. Collaborative call filtering: The research plans to design, implement and analyze a system that allows Internet users to collaborate in filtering undesirable calls, based on group preferences. This avoids the current reliance on unlisted phone numbers and answering machines for call filtering. Scaling Resource Reservation: Particularly for the large number of small flows anticipated for Internet telephony, associated with their stringent delay and loss requirements, scaling resource reservations is of vital importance. The research proposes to investigate the performance of two new scaling mechanisms, sink-tree reservations as part of the Border Gateway Reservation Protocol (BGRP) and partial reservations. Performance prediction: The research plans to investigate simple statistical characterizations that predict the performance of a variety of standard forward error correction and playout delay algorithms for packet voice. 911 services: Internet telephony offers the opportunity to significantly enhance emergency call services compared to existing mechanisms. The research plans to investigate how to reliably identify users and grant temporary access to personal data to medical personnel, without having to store medical data in centralized locations. Mobility: The research proposes to investigate application-layer mobility mechanisms to complement traditional IP-layer mobility, potentially affording greater security and faster hand-off, with no network infrastructure modifications required. In addition, the research proposes a new data and media distribution system, e*meme, that allows data to propagate in ad-hoc networks where network nodes are only sporadically connected to the wide-area network, as is the case in areas where wireless connectivity is spotty. The research plans to investigate efficient, popularity-based cache replication algorithms, taking into account the particular considerations of power- and storage-limited mobile computers.