27 January 2004
In this talk I am going to present an overview of some my recent research activities in wide-area wireless networks. Unlike in traditional wired and wireless IP-based networks, applications running over wide-area wireless (WWAN) environments are significantly affected by the vagaries of the cellular wireless link. Prior research has proposed and analyzed optimizations at individual layers of the protocol stack that improve application performance. In the first part of the talk I will present a detailed experiment-based evaluation and comparison of a wide selection of such optimization techniques in commercial WWAN environments. In the second part of this talk I will introduce a new architecture, called MAR, that leverages available diversity in the wireless environment to improve application performance. We have designed and implemented a prototype MAR system. Finally I will conclude by highlighting some of the ongoing research efforts in the context of this project.
24 February 2004
In this talk, we present an extensive passive measurement study of the messages exchanged in the Gnutella peer-to-peer (P2P) file sharing system. Based on the intuition gained by analyzing the measured data, we derive a detailed synthetic workload model for the search algorithm in P2P file sharing systems. Opposed to previous work, the workload model is based on the characterization of individual peers. It incorporates both, descriptions of a peer's active behavior (i.e., generating queries), and its passive behavior (i.e., responding to queries). As major contribution, we present an analytical formula characterizing the matching of files to queries, which closes the loop between a peer's active and passive behavior. The workload model consists of simple, easily computable probability distributions. Thus, it enables a straightforward generation of synthetic workloads with low computational effort. Furthermore, the workload model is independent of the Gnutella system and scalable in terms of the considered number of peers, files and queries. Thus, the workload model can be employed for the evaluation of arbitrary search algorithms for P2P file sharing systems.
This talk is based on joint work with Alex Klemm and Oliver Waldhorst.
17 February 2004
There is growing interest in increasing the spatial reuse of wireless medium through more elaborate physical layer techniques. In this talk, I will present our recent work on characterizing the effectiveness of these techniques in increasing the network level throughput. In particular, I will show that some of the assumptions commonly made in recent wireless networking papers are far from reality.
23 March 2004
A conventional video file contains a single temporally-ordered sequence of video frames. Clients requesting on-demand streaming of such a file receive (all or intervals of) the same content. For popular files that receive many requests during a file playback time, scalable streaming protocols based on multicast or broadcast have been devised. Such protocols require server and network bandwidth that grow much slower than linearly with the file request rate.
This paper considers "non-linear" video content in which there are parallel sequences of frames. Clients dynamically select which branch of the video they wish to follow, sufficiently ahead of each branch point so as to allow the video to be delivered without jitter. An example might be "choose-your-own-ending" movies. With traditional scalable delivery architectures such as movie theaters or TV broadcasting, such personalization of the delivered video content is very difficult or impossible. It becomes feasible, in principle at least, when the video is streamed to individual clients over a network. This paper analyzes the minimal server bandwidth requirements, and proposes and evaluates practical scalable delivery protocols, for on-demand streaming of nonlinear media.
6 April 2004
This paper considers the distribution of the rates at which flows transmit data, and the causes of these rates. First, using packet level traces from several Internet links, and summary flow statistics from an ISP backbone, we examine Internet flow rates and the relationship between the rate and other flow characteristics such as size and duration. We find, as have others, that while the distribution of flow rates is skewed, it is not as highly skewed as the distribution of flow sizes. We also find that for large flows the size and rate are highly correlated. Second, we attempt to determine the cause of the rates at which flows transmit data by developing a tool, T-RAT, to analyze packet-level TCP dynamics. In our traces, the most frequent causes appear to be network congestion and receiver window limits.