This experimental study presents a number of issues that pose a challenge for practical configuration tuning and its deployment in data analytics frameworks. These issues include: 1) the assumption of a static workload or environment, ignoring the dynamic characteristics of the analytics environment (e.g., increase in input data size, changes in allocation of resources). 2) the amortization of tuning costs and how this influences what workloads can be tuned in practice in a cost-effective manner. 3) the need for a comprehensive incremental tuning solution for a diverse set of workloads. We adapt different ML techniques in order to obtain efficient incremental tuning in our problem domain, and propose Tuneful, a configuration tuning framework. We show how it is designed to overcome the above issues and illustrate its applicability by running a wide array of experiments in cloud environments provided by two different service providers.

The execution of distributed data processing workloads (such as those running on top of Hadoop or Spark) in cloud environments presents a unique opportunity to explore multiple trade-offs between elasticity (and types of resources being allocated), overall runtime and total costs. However, beyond high-level constraints and objectives, it's not the end-users who should be mainly concerned with those optimizations, but the cloud providers. They have both the vantage point to collect actionable information, economies of scale and position to adjust parameters when dynamic conditions change, in order to fulfil SLOs that go beyond classic measures of latency and throughput. This is at odds with the existing approach of making software (including the interfaces to the cloud and the processing frameworks) as configurable as possible. We propose that rather than configurability, self-tunability (or the illusion of it as far as the end-user is concerned) is a better long-term goal.

The “Computing for the Future of the Planet” project is aimed at the intersection of computing and sustainability. An update on progress in the four main areas will be given: an optimal digital infrastructure, sensing and optimising with a global world model, reliably predicting and reacting to the environment, and digital alternatives to physical activities. Practical industrial examples will be presented in addition to longer term research goals.

Live migration of virtual machines is now commonplace but the issue of synchronising storage remains an obstacle to wide-area migration between datacentres. We discuss the range of possible designs for synchronising disk images and argue that we have to be adaptive at the sector level to efficiently transfer disk-state to the destination. We show, using a number of production workloads, that delaying the transmission of "hot"sectors can achieve considerable bandwidth reduction over a naive eager strategy. We introduce the Activity Based Sector Synchronisation (ABSS) algorithm and show by trace-driven simulations that ABSS is able to deliver a useful trade-off with a data transmission overhead between 0.3% and 2.4% of a gigabit link whilst incurring in the majority of cases a migration latency of less than 1.1 s.

With the ability to move virtual machines between physical hosts, live migration is a core feature of virtualisation. However for migration to be useful, deployable feature on a large (datacentre) scale, we need to predict migration times with accuracy. In this paper, we characterise the parameters affecting live migration with particular emphasis on the Xen virtualisation platform. We discuss the relationships between the important parameters that affect migration and highlight how migration performance can vary considerably depending on the workload. We further provide 2 simulation models that are able to predict migration times to within 90% accuracy for both synthetic and real-world benchmarks.

This paper explores the use of location-awareness to dynamically optimise the energy consumption of an office. It makes use of high-accuracy location data collected over 60 days randomly selected from a year in a commercial environment to evaluate the potential for energy savings and to motivate techniques that might be used.

Pervasive computing research has highlighted a need for location systems and input devices that are easy to deploy ubiquitously, are inexpensive, reliable, and practically invisible. This paper describes a novel tracking platform that attempts to meet all of these goals. The system exploits the static charge that naturally accumulates with movement through typical environments to create an inexpensive system that has potential both as a passive location system and as an input device. We demonstrate the physical principles, develop hardware and software to reliably detect interactions, and extensively evaluate a prototype implementation. We find the prototype to have a low false negative rate and a zero false positive rate. The power draw lies in the milliwatt-range and the cost per room is significantly lower than comparable systems

Ultrasonic location systems are a popular solution for the provision of fine-grained indoor positioning data. Applications include enhanced routing for wireless networks, computer-aided navigation, and location-sensitive device behavior. However, current ultrasonic location systems suffer from limitations due to their use of narrowband transducers, This paper investigates the use of broadband ultrasound for indoor positioning systems. Broadband ultrasonic transmitter and receiver units have been developed and characterized. The utilization of these units to construct two positioning systems with different architectures serves to highlight and affirm the concrete, practical benefits of broadband ultrasound for locating people and devices indoors.

Location-aware systems are typically deployed on a small scale and evaluated technically, in terms of absolute errors. In this paper, the authors present their experience of deploying an indoor location system (the Bat system) over a larger area and running it for a period exceeding two years.A number of technical considerations are highlighted: a need to consider aesthetics throughout deployment, the disadvantages of specialising sensors for location only, the need for autonomous maintenance of the computational world model, the dangers in coinciding physical and symbolic boundaries, the need to design for space usage rather than space and the need to incorporate feed-back mechanisms and power management. An evaluation of long term user experiences is presented, derived from a survey, logged usage data, and empirical observations. Statistically, it is found that 35% wear their Bat daily, 35% characterise their Bat as useful, privacy concerns are rare for almost 90% of users, and users cite the introduction of more applications and the adoption of the system by other users as their chief incentives to be tracked.Thia paper aims to highlight the need to evaluate large-scale deployments of such systems both technically and through user studies.

Individual wireless access networks show limitations that can be overcome through the integration of different technologies into a unified platform [i.e., fourth-generation (4G) system]. Nevertheless, the integration of heterogeneous networks poses many challenges such as adding complexity to the processes of deciding when to handoff, selecting the best network, and minimizing roaming effects using appropriate handover methods. This paper presents PROTON, a novel solution that supports decision-making processes related to roaming between heterogeneous technologies. PROTON deploys a formal policy representation model, based on finite-state transducers, that evaluates policies using information from the context to manage mobiles' behavior in a transparent manner, hiding 4G systems' complexities. We blend concepts of autonomic computing into the design of the solution and manage to improve user experience in typical 4G scenarios, while keeping transparency.

Context-aware computing systems demand an accurate and up-to-date world model which computationally represents the environment they oversee. Systems to date tend to have small-scale implementations with hand-programmed world models. In real environments, manual creation and maintenance of such models is infeasible. This work presents a method of using signals propagating in a multilateration positioning system to assist in creating and maintaining models of a dynamic world. It builds on a previous method for discovering objects in static environments. The methods are implemented and evaluated using a real positioning system. They are shown to build three-dimensional occupancy grids of indoor volumes, and have the capability of modifying those grids as time proceeds and the environment is reconfigured.

Article Share on Single Reflection Spatial Voting: A Novel Method for Discovering Reflective Surfaces Using Indoor Positioning Systems Authors: R. K. Harle View Profile , A. Ward View Profile , A. Hopper View Profile Authors Info & Claims MobiSys '03: Proceedings of the 1st international conference on Mobile systems, applications and servicesMay 2003 Pages 1–14https://doi.org/10.1145/1066116.1189035Published:05 May 2003Publication History 12citation168DownloadsMetricsTotal Citations12Total Downloads168Last 12 Months3Last 6 weeks1 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my Alerts New Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteGet Access

We present a novel method of extracting topological and metric geographical data using only positional data sensed from personnel movements. We extend research from the field of robotics to cope with the gross nonuniformity of sightings that is characteristic of real people in an indoor environment, and any unintentional obstruction of positioning by the user. We use real data collected using the Bat positioning system installed in the Laboratory for Communication Engineering to present the results of implementing the method. We successfully derive useful information from the data, and suggest further ways in which the techniques described are useful in a ubiquitous, sensor-driven computing environment.

The Walfische model predicts the average path loss from a conceptualised model of the propagation environment in which propagation is assumed to take place over rows of buildings of equal height and spacing. The propagation loss is then separated from the free space wave-front spreading, multiple forward-diffraction past the rows of buildings and diffraction over the final rooftop down to the receiver. In this paper we use numerical simulations to examine the effect of random building height variations on the multiple diffraction loss component of the model. In particular, our results show that the average multiple diffraction loss component is increased by any building height variations. A simple equation, which agrees within 1 dB of the simulation results, is presented that relates building height variations, wavelength and average building separation to the increase in average multiple diffraction loss.

A ray-tracing algorithm is used to simulate broadband radio propagation at 3.5 GHz for a FWA network in a high-rise city environment. Both coverage and uplink interference are investigated, with linear antenna polarisation. It is shown that some rooftop locations are unsuitable positions for SU's, due to the existence of single or multiple-bounce reflected rays causing uplink interference. It is usually possible to find a SU position just a few metres away on the same rooftop which gives much lower uplink interference.

Networked surfaces are surfaces such as desks which provide network connectivity to specially augmented devices, for example handheld computers. When the devices are physically placed on top of the surface, they can connect to different kinds of services-mainly, but not exclusively to send and receive data. This paper discusses challenges in implementing networked surfaces, paying particular attention to data flow issues, focusing on how the various software and hardware entities comprising the surface interact to transport data to and from objects.

We describe a platform for context-aware computing which enables applications to follow mobile users as they move around a building.The platform is particularly suitable for richly equipped, networked environments.The only item a user is required to carry is a small sensor tag, which identifies them to the system and locates them accurately in three dimensions.The platform builds a dynamic model of the environment using these location sensors and resource information gathered by telemetry software, and presents it in a form suitable for application programmers.Use of the platform is illustrated through a practical example, which allows a user's current working desktop to follow them as they move around the environment.

Article Free Access Share on Preparing for the digital media monsoons Author: Andy Hopper Olivetti & Oracle Research Laboratoy, 24a Trumpington Street, Cambridge UK and University of Cambridge, Department of Engineering, Cambridge UK Olivetti & Oracle Research Laboratoy, 24a Trumpington Street, Cambridge UK and University of Cambridge, Department of Engineering, Cambridge UKView Profile Authors Info & Claims MULTIMEDIA '98: Proceedings of the sixth ACM international conference on MultimediaSeptember 1998 Pages 465–466https://doi.org/10.1145/290747.290824Online:01 September 1998Publication History 0citation189DownloadsMetricsTotal Citations0Total Downloads189Last 12 Months3Last 6 weeks0 Get Citation AlertsNew Citation Alert added!This alert has been successfully added and will be sent to:You will be notified whenever a record that you have chosen has been cited.To manage your alert preferences, click on the button below.Manage my AlertsNew Citation Alert!Please log in to your account Save to BinderSave to BinderCreate a New BinderNameCancelCreateExport CitationPublisher SiteeReaderPDF

The micro- and picocellular architectures proposed for wireless ATM LANs lead to wireless terminals frequently changing their point of attachment to the network. Because ATM connections have quality of service (QoS) guarantees which must be maintained, handover must be as seamless as possible. We present a novel architecture and protocol which primarily aims to keep the interruption period due to handover low, rather than seeking to keep the process entirely lossless. We compare the tradeoffs made, with those in other schemes from the literature, and give quantitative results from an implementation of our scheme on a 10 Mbits/s/sup -1/ prototype wireless ATM LAN.

This paper summarises our experience of using data analysers to enhance network multimedia services. By analysers we mean simple processing modules that extract information contained in various multimedia streams. They are categorised based on their location with respect to the network, the location being determined by balancing bandwidth requirements and computational complexity. Various applications are described where analysers are used to enhance aspects of the service provided. Details of the multimedia environment are given, followed by an overview of the analyser architecture and examples of analyser-enhanced applications. The paper is concluded by indicating directions of future development. KEYWORDS Data analysers, distributed multimedia, ATM network services. INTRODUCTION Advances in network technology permit sophisticated configurations in the local area, which have resulted in a wealth of multimedia data being exchanged on the network. It has therefore become necessary ...

No abstract available.

Handling audio and video in a digital environment requires timely delivery of data. This paper describes the principles adopted in the design of the Pandora networked multi-media system. They attempt to give the user the best possible service while dealing with error and overload conditions. Pandora uses a sub-system to handle the multi-media peripherals. It uses transputers and associated Occam code to implement the time critical functions. Stream implementation is based on self-contained segments of data containing information for delivery, synchronisation and error recovery. Decoupling buffers are used to allow concurrent operation of multiple processing elements. Clawback buffers are used to resynchronise streams at their destinations with minimum latency. The system has proved robust in normal use, under overload, and in the presence of errors. It has been in use for a number of years. The principles involved in this design are now being used in the development of two complemen...

In recent years considerable interest has been shown in the development of packet-based in-building radio LANs. Most of these networks use more than one frequency together with slot reservation strategies to handle mobile traffic in a radio environment. This paper proposes an architecture and a related media access protocol, using a single frequency, to support ATM (asynchronous transfer mode) traffic in multihop wireless LANs. Fixed size packets (known as ATM cells) are used to permit statistical muitiplexing, which enables the support of multimedia traffic in radio LANs. CSMA/AED (carrier sense multiple access with acknowledgement error detection) is the protocol proposed and analysed for ATM services. In conjunction with carrier sensing, the protocol uses a per cell basis acknowledgement packet to reduce the ATM cell loss rate by indirectly detecting errors, caused either by collision or by reception of a faulty cell at the receiver. The results presented in this paper are obtained from a simulation model, developed to investigate the protocol performance at the media access layer and its effects on overall network performance.>

A novel system for the location of people in an office environment is described. Members of staff wear badges that transmit signals providing information about their location to a centralized location service, through a network of sensors. The paper also examines alternative location techniques, system design issues and applications, particularly relating to telephone call routing. Location systems raise concerns about the privacy of an individual and these issues are also addressed.

An experimental ring communication network, the Cambridge Backbone Ring, for integrated computer data, voice, and real-time video applications is described. The network is aimed as a backbone facility for interconnecting groups of Cambridge Fast Ring (CFR) networks in the Cambridge area. Media access control is through the empty slot technique, with transmitting stations filling multiple slots each ring revolution. The network operates on monomode optical fiber and is designed for an eventual line rate of 1000 Mb/s. In the backbone ring architecture, the bandwidth of the fiber-optic channel can be partitioned into a number (currently four) of time division multiplexing (TDM) channels. This enables stations of varying cost and bandwidth to be attached to one network, parameterized by the number of channels a station can use concurrently. The backbone ring architecture achieves full connectivity, even with the simplest type of station. Some brief traffic simulation results are presented.>

In this paper we compare the simulated performance of a family of multiprocessor architectures based on a global shared memory. The processors are connected to the memory through caches that snoop one or more shared buses in crossbar arrangement.

A discussion is presented of the choices facing a designer as faster speeds for local area networks are contemplated. The 100-Mb/s Cambridge fast ring (CFR) is described. The ring protocol allows one of a number of fixed-size slots to be used once or repeatedly. The network design allows sets of rings to be constructed by pushing the bridge function to the lowest hardware level. Low cost and ease of use is normally achieved by design of special chips; two-chip VLSI implementation is described. This VLSI hardware forms the basis of a kit of parts from which many different network components can be constructed.>

The Cambridge ring is a local area network used both in universities and industry. It is based on the empty slot principle and data are transmitted using minipackets containing two bytes of data. This paper describes the design process, decisions, and tradeoffs in implementing an integrated system which incorporates both analog and digital components. The technology chosen is a bipolar gate array. A number of options are provided for the implementor who can optimize network parameters such as minipacket size and transmission speed to his needs. He can also choose the style of interface between the communicating device and the network. An important option provided by the integrated Cambridge ring is the ability to simultaneously transmit short control minipackets and long data packets. A system exploiting this feature has been built and is described. Its proposed uses are to interconnect telephones and other real-time systems as well as computers where the partitioning of bandwidth and precise performance specification are important.

The design of a communications network based on simple routing of packets through the network and suitable for use in local systems is presented. Several network topologies are considered using differing node structures. Generation of routing information is discussed together with flow control. It is shown that high performance is easily obtained and the system drawbacks do not affect the design for local networks.

This note presents a simple way of localizing errors and failures in ring communication systems. The method is also applicable to some other types of networks.