AT&T Labs

About: AT&T Labs is a based out in . It is known for research contribution in the topics: Network packet & The Internet. The organization has 1879 authors who have published 5595 publications receiving 483151 citations.

Characterizing the Rate Region of the (4,3,3) Exact-Repair Regenerating Codes

Abstract: Exact-repair regenerating codes are considered for the case (n,k,d)=(4,3,3), for which a complete characterization of the rate region is provided. This characterization answers in the affirmative the open question whether there exists a non-vanishing gap between the optimal bandwidth-storage tradeoff of the functional-repair regenerating codes (i.e., the cut-set bound) and that of the exact-repair regenerating codes. To obtain an explicit information theoretic converse, a computer-aided proof (CAP) approach based on primal and dual relation is developed. This CAP approach extends Yeung's linear programming (LP) method, which was previously only used on information theoretic problems with a few random variables due to the exponential growth of the number of variables in the corresponding LP problem. The symmetry in the exact-repair regenerating code problem allows an effective reduction of the number of variables, and together with several other problem-specific reductions, the LP problem is reduced to a manageable scale. For the achievability, only one non-trivial corner point of the rate region needs to be addressed in this case, for which an explicit binary code construction is given.

Scattercast: an adaptable broadcast distribution framework

Abstract: Internet broadcasting - the simultaneous distribution of live content streams to a large audience - has a number of interesting applications ranging from real-time broadcasts of audio/video streams for online concerts or sporting events to efficient and reliable large-scale software distribution. We identify three fundamental requirements for scalable broadcasting services: an efficient infrastructure for large-scale broadcasting, an ability to adapt the infrastructure to suit the requirements of a wide range of applications, and ease of deployment of the infrastructure. Although solutions such as the network-layer IP multicast approach and a slew of overlay distribution networks exist today, none of these technologies satisfactorily addresses all of the above concerns.In this paper, we argue that an application-customizable hybrid overlay is well suited to meet these challenges. To this end, we propose an architecture called scattercast that relies on a network of strategically located agents called Scatter-Cast proXies or SCXs. These agents collaboratively provide the broadcast service for a session. Clients locate a nearby SCX and tap into the session via that SCX. Scattercast constructs a hybrid overlay network composed of unicast links between SCXs that interconnect locally scoped multicast regions. Rather than define a single standardized service model for transmitting data on top of the overlay, scattercast builds a customizable transport framework that provides adaptability by leveraging application-defined semantics to drive the distribution of content. We demonstrate the ability of our architecture to provide efficient distribution via a set of simulation experiments. Finally, we present our experience with the adaptability of the framework by describing two applications, a real-time Internet radio and an online slide-presentation tool, both of which we have built on top of a prototype implementation of the architecture.

Methods and Apparatus for Individualized Content Delivery

Abstract: Systems and techniques for predicting customer response to content and selecting content for delivery to particular customers in accordance with the predictions. As information is delivered to and received from a plurality of customers over multiple communication channels, data streams representing communications between providers and customers are analyzed and selected data extracted therefrom. Linkages are created between data collected from the different channels and data are anonymized. The data are analyzed to create a customer response predictor for each customer that models customer behavior and predicts customer response to advertisements. As content, such as advertisements, are to be delivered to a destination, information from a predictor created using data collected from a customer associated with the destination is used to select appropriate content.

Early measurements of a cluster-based architecture for P2P systems

Abstract: Peer-to-peer applications such as Napster [4], Freenet [1], and Gnutella [2], [7] have gained much attention recently. These applications are mainly designed and used for largescale sharing of MP3 files. In such systems, end-hosts self-organize into an overlay network and share content with each other. Compared to the traditional client-server model, files are served in a distributed manner and replicated among the network on demand. Since hosts participating in peer-to-peer (P2P) networks also devote some computing resources, such systems scale with the number of hosts in terms of hardware, bandwidth, and disk space. With the wide deployment of P2P applications, the P2P traffic is becoming a growing portion of the Internet traffic. There has been very little examination of P2P traffic patterns and how they differ from traditional service models. Studying and understanding P2P traffic is thus important to provide efficient application-level content location and routing within the network. The existing applications use their own approach to do content location and routing and none of them are scalable. Napster uses a centralized server to locate content, while Gnutella clients broadcast queries to all their neighbors. [8] discusses the query locality observed in Gnutella traces and suggests caching as a short-term approach to increase Gnutella’s scalability. Recent designs such as CAN [5], Chord [9], Pastry [6], and Tapestry [10] propose distributed indexing schemes based on hashing to locate content. These systems assume a flat content delivery mesh. Each object’s location is stored at one or more nodes selected deterministically by a uniform hash function; queries for the object will be routed incrementally to the node. Although hash functions can help locate content deterministically, they lack the flexibility of keyword searching—a useful operation to find content without prior knowledge of exact object names. There is no real deployment at present and thus no measurement information is available for understanding the usability and scalability of

Voice signatures

Abstract: Most current spoken-dialog systems only extract sequences of words from a speaker's voice. This largely ignores other useful information that can be inferred from speech such as gender, age, dialect, or emotion. These characteristics of a speaker's voice, voice signatures, whether static or dynamic, can be useful for speech mining applications or for the design of a natural spoken-dialog system. This paper explores the problem of extracting automatically and accurately voice signatures from a speaker's voice. We investigate two approaches for extracting speaker traits: the first focuses on general acoustic and prosodic features, the second on the choice of words used by the speaker. In the first approach, we show that standard speech/nonspeech HMM, conditioned on speaker traits and evaluated on cepstral and pitch features, achieve accuracies well above chance for all examined traits. The second approach, using support vector machines with rational kernels applied to speech recognition lattices, attains an accuracy of about 8.1 % in the task of binary classification of emotion. Our results are based on a corpus of speech data collected from a deployed customer-care application (HMIHY 0300). While still preliminary, our results are significant and show that voice signatures are of practical interest in real-world applications.