Peter Steenkiste

Project Aura: toward distraction-free pervasive computing

The most precious resource in a computer system is no longer its processor, memory, disk, or network, but rather human attention. Aura aims to minimize distractions on a users attention, creating an environment that adapts to the users context and needs. Aura is specifically intended for pervasive computing environments involving wireless communication, wearable or handheld computers, and smart spaces. Human attention is an especially scarce resource in such environments, because the user is often preoccupied with walking, driving, or other real-world interactions. In addition, mobile computing poses difficult challenges such as intermittent and variable-bandwidth connectivity, concern for battery life, and the client resource constraints that weight and size considerations impose. To accomplish its ambitious goals, research in Aura spans every system level: from the hardware, through the operating system, to applications and end users. Underlying this diversity of concerns, Aura applies two broad concepts. First, it uses proactivity, which is a system layers ability to anticipate requests from a higher layer. In todays systems, each layer merely reacts to the layer above it. Second, Aura is self-tuning: layers adapt by observing the demands made on them and adjusting their performance and resource usage characteristics accordingly. Currently, system-layer behavior is relatively static. Both of these techniques will help lower demand for human attention.

Rainbow: architecture-based self-adaptation with reusable infrastructure

While attractive in principle, architecture-based self-adaptation raises a number of research and engineering challenges. First, the ability to handle a wide variety of systems must be addressed. Second, the need to reduce costs in adding external control to a system must be addressed. Our rainbow framework attempts to address both problems. By adopting an architecture-based approach, it provides reusable infrastructure together with mechanisms for specializing that infrastructure to the needs of specific systems. The specialization mechanisms let the developer of self-adaptation capabilities choose what aspects of the system to model and monitor, what conditions should trigger adaptation, and how to adapt the system.

Evaluation and characterization of available bandwidth probing techniques

The packet pair mechanism has been shown to be a reliable method to measure the bottleneck link capacity on a network path, but its use for measuring available bandwidth is more challenging. In this paper, we use modeling, measurements, and simulations to better characterize the interaction between probing packets and the competing network traffic. We first construct a simple model to understand how competing traffic changes the probing packet gap for a single-hop network. The gap model shows that the initial probing gap is a critical parameter when using packet pairs to estimate available bandwidth. Based on this insight, we present two available bandwidth measurement techniques, the initial gap increasing (IGI) method and the packet transmission rate (PTR) method. We use extensive Internet measurements to show that these techniques estimate available bandwidth faster than existing techniques such as Pathload, with comparable accuracy. Finally, using both Internet measurements and ns simulations, we explore how the measurement accuracy of active probing is affected by factors such as the probing packet size, the length of probing packet train, and the competing traffic on links other than the tight link.

Self-management in chaotic wireless deployments

Over the past few years, wireless networking technologies have made vast forays into our daily lives. Today, one can find 802.11 hardware and other personal wireless technology employed at homes, shopping malls, coffee shops and airports. Present-day wireless network deployments bear two important properties: they are unplanned, with most access points (APs) deployed by users in a spontaneous manner, resulting in highly variable AP densities; and they are unmanaged, since manually configuring and managing a wireless network is very complicated. We refer to such wireless deployments as being chaotic.In this paper, we present a study of the impact of interference in chaotic 802.11 deployments on end-client performance. First, using large-scale measurement data from several cities, we show that it is not uncommon to have tens of APs deployed in close proximity of each other. Moreover, most APs are not configured to minimize interference with their neighbors. We then perform trace-driven simulations to show that the performance of end-clients could suffer significantly in chaotic deployments. We argue that end-client experience could be significantly improved by making chaotic wireless networks self-managing. We design and evaluate automated power control and rate adaptation algorithms to minimize interference among neighboring APs, while ensuring robust end-client performance.

On path selection for traffic with bandwidth guarantees

Transmission of multimedia streams imposes a minimum-bandwidth requirement on the path being used to ensure end-to-end Quality-of-Service (QoS) guarantees. While any shortest-path algorithm can be used to select a feasible path, additional constraints that limit resource consumption and balance the network load are needed to achieve efficient resource utilization. We present a systematic evaluation of four routing algorithms that offer different tradeoffs between limiting the path hop count and balancing the network load. Our evaluation considers not only the call blocking rate but also the fairness to requests for different bandwidths, robustness to inaccurate routing information, and sensitivity to the routing information update frequency. It evaluates not only the performance of these algorithms for the sessions with bandwidth guarantees, but also their impact on the lower priority best-effort sessions. Our results show that a routing algorithm that gives preference to limiting the hop count performs better when the network load is heavy, while an algorithm that gives preference to balancing the network load performs slightly better when the network load is light. We also show that the performance of using pre-computed paths with a few discrete bandwidth requests is comparable to that of computing paths on-demand, which implies feasibility of class-based routing. We observe that the routing information update interval can be set reasonably large to reduce routing overhead without sacrificing the overall performance, although an increased number of sessions can be misrouted.

Quality-of-Service Routing for Traffic with Performance Guarantees

Quality-of-Service (QoS) routing tries to select a path that satisfies a set of QoS constraints, while also achieving overall network resource efficiency. We present initial results on QoS path selection for traffic requiring bandwidth and delay guarantees. For traffic with bandwidth guarantees, we found that several routing algorithms that favor paths with fewer hops perform well. For traffic with delay guarantees, we show that for a broad class of WFQ-like scheduling algorithms, the problem of finding a path satisfying bandwidth, delay, delay-jitter, and/or buffer space constraints while at the same time deriving the bandwidth that has to be reserved to meet these constraints, is solvable by a modified version of the Bellman-Ford shortest-path algorithm in polynomial time.

Measurement and analysis of the error characteristics of an in-building wireless network

There is general belief that networks based on wireless technologies have much higher error rates than those based on more traditional technologies such as optical fiber, coaxial cable, or twisted pair wiring. This difference has motivated research on new protocol suites specifically for wireless networks. While the error characteristics of wired networks have been well documented, less experimental data is available for wireless LANs.In this paper we report the results of a study characterizing the error environment provided by AT&T WaveLAN, a commercial product designed for constructing 2 Mb/s in-building wireless networks. We evaluated the effects of interfering radiation sources, and of attenuation due to distance and obstacles, on the packet loss rate and bit error rate. We found that under many conditions the error rate of this physical layer is comparable to that of wired links. We analyze the implications of our results on todays CSMA/CA based wireless LANs and on future pico-cellular shared-medium reservation-based wireless networks.

Efficient channel-aware rate adaptation in dynamic environments

Increasingly, 802.11 devices are being used by mobile users. This results in very dynamic wireless channels that are difficult to use efficiently. Current rate selection algorithms are dominated by probe-based approaches that search for the best transmission rate using trial-and-error. In mobile environments, probe-based techniques often perform poorly because they inefficiently search for the moving target presented by the constantly changing channel. We have developed a channel-aware rate adaptation algorithm CHARM - that uses signal strength measurements collected by the wireless cards to help select the transmission rate. Moreover, unlike previous approaches CHARM leverages channel reciprocity to obtain channel information, so the information is available to the transmitter without incurring RTS/CTS overhead. This combination of techniques allows CHARM to respond quickly to dynamic channel changes. We implemented CHARM in the Madwifi driver for wireless cards using the Atheros chipset. Our evaluation both in the real world and on a controlled testbed shows that channel-aware rate selection can significantly outperform probe-based rate adaptation, especially over dynamic channels.

Locating internet bottlenecks: algorithms, measurements, and implications

The ability to locate network bottlenecks along end-to-end paths on the Internet is of great interest to both network operators and researchers. For example, knowing where bottleneck links are, network operators can apply traffic engineering either at the interdomain or intradomain level to improve routing. Existing tools either fail to identify the location of bottlenecks, or generate a large amount of probing packets. In addition, they often require access to both end points. In this paper we present Pathneck, a tool that allows end users to efficiently and accurately locate the bottleneck link on an Internet path. Pathneck is based on a novel probing technique called Recursive Packet Train (RPT) and does not require access to the destination. We evaluate Pathneck using wide area Internet experiments and trace-driven emulation. In addition, we present the results of an extensive study on bottlenecks in the Internet using carefully selected, geographically diverse probing sources and destinations. We found that Pathneck can successfully detect bottlenecks for almost 80% of the Internet paths we probed. We also report our success in using the bottleneck location and bandwidth bounds provided by Pathneck to infer bottlenecks and to avoid bottlenecks in multihoming and overlay routing.

Providing contextual information to pervasive computing applications

Pervasive computing applications are increasingly leveraging contextual information from several sources to provide users with behavior appropriate to the environment in which they reside. If these sources of contextual information are used and deployed in an ad hoc manner however they may provide overlapping functionality, fail to provide needed functionality, and require the use of inconsistent interfaces by applications. To overcome these problems, we introduce a contextual information service that provides applications with contextual information via a virtual database. Unlike previous efforts, our service provides applications a consistent, lightweight, and powerful mechanism for obtaining contextual information, and includes explicit support for the on demand computation of contextual information. We show, via example applications and a contextual information service prototype that we have implemented, how this approach can be used to allow proactive applications to adapt their behavior to match a users current environment.