Edge–cloud computing infrastructures are increasingly widespread as they combine the flexibility of cloud-native development tools with the performance and security of distributed computing environments. Function-as-a-Service has emerged as a powerful abstraction that overcomes the limitations of a micro-service architecture. However, it generally does not support stateful functions, making it unsuitable for many practical applications in, e.g., Internet of Things (IoT) and real-time analytics. In this paper, we explore a novel paradigm, based on stateful asynchronous agents, that goes beyond traditional serverless computing. We focus on several key technical aspects: programming model, deployment procedures, design of a flexible compute node, and state management. We illustrate our paradigm using the EDGELESS platform as a concrete implementation of this stateful agents’ pattern. We report proof-of-concept experiment results obtained in a testbed with heterogeneous resource-constrained edge nodes that showcase some distinguishing features of our platform: scalable management of lightweight function instances, the advantage of keeping the state local at function instances, and delegated orchestration to enable a third-party agent to make migration decisions in a group of local nodes.

Serverless computing is regarded as a good match for distributed edge infrastructures. However, bringing the function-as-a-service model to a highly dynamic, distributed, and heterogeneous pool of resources has its fair amount of challenges. Allocation of functions to the proper resource is an essential operation that avoids over-provisioning and the relative waste of computational resources. In this work, we propose Dike, a bi-level function scheduling and resource allocation framework designed to meet end-to-end latency SLOs (service-level objectives). Dike leverages deep reinforcement learning to balance resource provisioning and monetary cost by incorporating both composite cost and SLO violations into the reward. Extensive simulations with real-world production workloads demonstrate the superiority of Dike. Experimental results show that the proposed algorithms approximate the results of state-of-the-art ILP solver within a factor of 1.08 while dramatically reducing the scheduling time.

EDGELESS is a serverless platform targeting edge computing that supports widely distributed deployments using heterogeneous devices. We present its components, architecture, and programming model. Our working prototype of EDGELESS enables executing lightweight functions and is already available as open-source.

With the promise of federated learning (FL) to allow for geographically-distributed and highly personalized services, the efficient exchange of model updates between clients and servers becomes crucial. FL, though decentralized, often faces communication bottlenecks, especially in resource-constrained scenarios. Existing data compression techniques like gradient sparsification, quantization, and pruning offer some solutions, but may compromise model performance or necessitate expensive retraining. In this paper, we introduce FedSZ, a specialized lossy-compression algorithm designed to minimize the size of client model updates in FL. FedSZ incorporates a comprehensive compression pipeline featuring data partitioning, lossy and lossless compression of model parameters and metadata, and serialization. We evaluate FedSZ using a suite of error-bounded lossy compressors, ultimately finding SZ2 to be the most effective across various model architectures and datasets including AlexNet, MobileNetV2, ResNet50, CIFAR-10, Caltech101, and Fashion-MNIST. Our study reveals that a relative error bound${10}^{-2}$achieves an optimal tradeoff, compressing model states between 5.55-12.61× while maintaining inference accuracy within < 0.5 % of uncompressed results. Additionally, the runtime overhead of FedSZ is < 4.7% or between of the wall-clock communication-round time, a worthwhile trade-off for reducing network transfer times by an order of magnitude for networks bandwidths < 350Mbps. Intriguingly, we also find that the error introduced by FedSZ could potentially serve as a source of differentially private noise, opening up new avenues for privacy-preserving FL.

Increase in popularity of various deep learning methodologies has led to a huge increase in demand for training data. These training data often contain video featuring people but these videos often cannot be used due to privacy concerns as people are identifiable. We present an extension to the DeepDish object tracking system running on a Raspberry Pi. DeepDish performs object detection in the Raspberry Pi itself, avoiding the need to ship raw video data off the device. Our extension adds online face detection and blurring to preserve privacy. We implement and evaluate four different face detection algorithms, achieving 83% accuracy using YuNet, while increasing the latency by 75%. We then implement and evaluate three different object tracking algorithms to reduce the latency increase by running the simpler object tracking algorithm, reducing the number of calls to the expensive face detection algorithm. The result is a significant decrease in latency at the cost of a small decrease in accuracy. We also implemented and compared three different face obfuscation algorithms, one of which achieves differential privacy.

With the massive amount of data generated from mobile devices and the increase of computing power of edge devices, the paradigm of Federated Learning has attracted great momentum. In federated learning, distributed and heterogeneous nodes collaborate to learn model parameters. However, while providing benefits such as privacy by design and reduced latency, the heterogeneous network present challenges to the synchronisation methods, or barrier control methods, used in training, regarding system progress and model convergence etc. The design of these barrier mechanisms is critical for the performance and scalability of federated learning systems. We propose a new barrier control technique called Probabilistic Synchronous Parallel (PSP). In contrast to existing mechanisms, it introduces a sampling primitive that composes with existing barrier control mechanisms to produce a family of mechanisms with improved convergence speed and scalability. Our proposal is supported with a convergence analysis of PSP-based SGD algorithm. In practice, we also propose heuristic techniques that further improve the efficiency of PSP. We evaluate the performance of proposed methods using the federated learning specific FEMNSIT dataset. The evaluation results show that PSP can effectively achieve good balance between system efficiency and model accuracy, mitigating the challenge of heterogeneity in federated learning.

As smart environments become laden with more and more sensors, there has been a need to develop systems that could derive useful information from these sensors and make the smart environments smarter. Complex Event Processing (CEP) has emerged as a popular strategy to identify crucial events from sensor data. However, the existing CEP strategies overlook the relationship with other sensors in the spatial vicinity and understate the temporal variation of sensor data. In this paper, we develop RACER, which is an end-to-end complex event processing system that takes into consideration both the spatial location of the sensor in observation and the varying impact of temporal changes in the sensor data. Experiments performed for a duration of five months over both collected and live streaming data shows that RACER fares well compared to the other state-of-the-art approaches.

IoT is developing rapidly with frequently appearing new wireless standards and applications. However, besides a large number of IoT benefits, its further development is now being slowed down due to the repetition of old Internet development flaws while dealing with IoT heterogeneity. The current misleading trend aims to solve all IoT interoperation problems by inserting IP Addresses into those wireless protocols where the IP stack clearly slows down application performance and drains the battery, e.g. LPWANs such as LoRaWAN and SigFox. This paper tackles IoT heterogeneity from a different perspective: it is the application interoperation which matters the most. The protocols beneath the application layer shall work for smooth upper-layer service provisioning where the IP shall be just one of the many underlying integration options instead of being the essential one. Inspired by previous proposals for a more flexible internetworking architecture, this paper applies those theoretical concepts in practice by proposing a protocol-independent distributed interoperation model for smooth service provisioning over heterogeneous IoT wireless contexts. The arguments pro the IP-agnostic IoT application interoperation are supported by the model's prototype which showed 1.6-2 times faster MQTT application operation over LoRa and WiFi compared to the legacy IP-based MQTT provisioning over that protocols.

As more and more devices are connected to the Internet-of-Things, often made by non-specialist companies or short-lived startups, the likelihood that these devices will be hacked and used for nefarious activity online increases. We seek to support non-expert users in managing the network behaviour of their IoT devices, and assisting them in handling the cases where those devices are hacked. To do so, we wish to enable anomaly detection at the network level, determining when a device starts behaving unusually. This requires capturing data about how devices behave in a diverse range of real deployments, not just lab environments.

Emerging Machine Learning (ML) techniques, such as Deep Neural Network, are widely used in today's applications and services. However, with social awareness of privacy and personal data rapidly rising, it becomes a pressing and challenging societal issue to both keep personal data private and benefit from the data analytics power of ML techniques at the same time. In this paper, we argue that to avoid those costs, reduce latency in data processing, and minimise the raw data revealed to service providers, many future AI and ML services could be deployed on users' devices at the Internet edge rather than putting everything on the cloud. Moving ML-based data analytics from cloud to edge devices brings a series of challenges. We make three contributions in this paper. First, besides the widely discussed resource limitation on edge devices, we further identify two other challenges that are not yet recognised in existing literature: lack of suitable models for users, and difficulties in deploying services for users. Second, we present preliminary work of the first systematic solution, i.e. Zoo, to fully support the construction, composing, and deployment of ML models on edge and local devices. Third, in the deployment example, ML service are proved to be easy to compose and deploy with Zoo. Evaluation shows its superior performance compared with state-of-art deep learning platforms and Google ML services.

No abstract available.

Machine Learning (ML) techniques have begun to dominate data analytics applications and services. Recommendation systems are the driving force of online service providers such as Amazon. Finance analytics has quickly adopted ML to harness large volume of data in such areas as fraud detection and risk-management. Deep Neural Network (DNN) is the technology behind voice-based personal assistance, self-driving cars [1], image processing [3], etc. Many popular data analytics are deployed on cloud computing infrastructures. However, they require aggregating users’ data at central server for processing. This architecture is prone to issues such as increased service response latency, communication cost, single point failure, and data privacy concerns.

As locative media and augmented reality spread into the everyday world so it becomes important to create aesthetic visual markers at scale. We explore a designer-centred approach in which skilled designers handcraft seed designs that are automatically recombined to create many markers as subtle variants of a common theme. First, we extend the d-touch topological approach to creating visual markers that has previously been shown to support creative design with two new techniques: area order codes and visual checksums. We then show how the topological structure of such markers provides the basis for recombining designs to generate many variations. We demonstrate our approach through the creation of beautiful, personalized and interactive wallpaper. We reflect on how technologies must enable designers to balance goals of scalability, aesthetics and reliability in creating beautiful interactive decoration.

This paper outlines the IoT Databox model as a means of making the Internet of Things (IoT) accountable to individuals. Accountability is a key to building consumer trust and mandated in data protection legislation. We briefly outline the `external' data subject accountability requirement specified in actual legislation in Europe and proposed legislation in the US, and how meeting requirement this turns on surfacing the invisible actions and interactions of connected devices and the social arrangements in which they are embedded. The IoT Databox model is proposed as an in principle means of enabling accountability and providing individuals with the mechanisms needed to build trust in the IoT.

The long-heralded Internet of Things (IoT) is finally becoming a reality. From factories and the ubiquitous Internet-connected fridge, we now see heating control systems, cars, dishwashers, and all manner of common-place devices being connected. While this has certainly realized new capabilities, such as the ability to control one’s domestic heating remotely, the benefits are perhaps more mixed: every device that we can remotely control is a device that someone else can remotely hack. And that is just the devices—the literal things; in tandem we also see increasing intrusion of Internet-connectivity into services, practices and everyday infrastructures such as transport and retail. Coupled with the increasingly invasive deployment of these devices into everyday lives, the result is a substantial increase in threats to privacy arising from the IoT.

We explore the crafting of interactive decoration for everyday artefacts. This involves adorning them with decorative patterns that enhance their beauty while triggering digital interactions when scanned with cameras. These are realized using an existing augmented reality technique that embeds computer readable codes into the topological structures of hand-drawn patterns. We describe a research through design process that engaged artisans to craft a portfolio of interactive artefacts, including ceramic bowls, embroidered gift cards, fabric souvenirs, and an acoustic guitar. We annotate this portfolio with reflections on the crafting process, revealing how artisans addressed pattern, materials, form and function, and digital mappings throughout their craft process. Further reflection on our portfolio reveals how they bridged between human and system perceptions of visual patterns and engaged in a deep embedding of digital interactions into physical materials. Our findings demonstrate the potential for interactive decoration, distilling the craft knowledge involved in creating aesthetic and functional decoration, highlight the need for transparent computer vision technologies, and raise wider issues for HCI’s growing engagement with craft.

We present a research-through-design exploration of transforming large-scale public illustrations into interactive media. We collaborated with creative practitioners to extend an existing visual marker technology to support spatial and layered interaction with wall-sized images. We document how these techniques were used to design interactive illustrations and how visitors to an exhibition engaged with these. We conclude that it is feasible to combine spatial and layered interactions to attach complex narratives to public illustrations, and highlight challenges around instruction giving, sociality and repeat experiences.

In recent times, we have seen a proliferation of personal data. This can be attributed not just to a larger proportion of our lives moving online, but also through the rise of ubiquitous sensing through mobile and IoT devices. Alongside this surge, concerns over privacy, trust, and security are expressed more and more as different parties attempt to take advantage of this rich assortment of data.

This paper offers a sociological perspective on data protection regulation and its relevance to design. From this perspective, proposed regulation in Europe and the USA seeks to create a new economic actor—the consumer as personal data trader—through new legal frameworks that shift the locus of agency and control in data processing towards the individual consumer or “data subject”. The sociological perspective on proposed data regulation recognises the reflexive relationship between law and the social order, and the commensurate needs to balance the demand for compliance with the design of computational tools that enable this new economic actor. We present the Databox model as a means of providing data protection and allowing the individual to exploit personal data to become an active player in the emerging data economy.

This paper investigates the playful photo taking behaviour of teenagers during group visits to two touristic public events (an airshow and a guided tour of a museum). These studies provide the feedback for the iterative development of a smartphone based anonymous image annotation and sharing application. The resulting implications for the design of such photo systems are examined, specifically the appropriateness of opportunistic upload for social media. Playfulness in photography has many implications regarding wider social behaviours. Comic annotations provide the ability to create humorous reinterpretations of photos, and the presence of humour and in-jokes affect the makeup of the audience with whom a group would like to share. It is counter-productive that an application encouraging such limiting behaviours may conflict with the open nature of touristic events. In addition, the shared images have an ephemeral quality and are therefore of transient value (compared to more tangible souvenirs), and their production through the application can ultimately distract from the experience of the visit itself.

Journal Article Tailored Scenarios: A Low-Cost Online Method to Elicit Perceptions of Home Technologies Using Participant-Specific Contextual Information Get access Michael Brown, Michael Brown * 1Horizon Digital Economy Research, University of Nottingham, Nottingham, Nottinghamshire, UK2Human Factors Research Group, University of Nottingham, Nottingham, UK *Corresponding author: [email protected] Search for other works by this author on: Oxford Academic Google Scholar Tim Coughlan, Tim Coughlan 1Horizon Digital Economy Research, University of Nottingham, Nottingham, Nottinghamshire, UK3Computer Science, University of Nottingham, Nottingham, Nottinghamshire, UK Search for other works by this author on: Oxford Academic Google Scholar Jesse Blum, Jesse Blum 1Horizon Digital Economy Research, University of Nottingham, Nottingham, Nottinghamshire, UK Search for other works by this author on: Oxford Academic Google Scholar Glyn Lawson, Glyn Lawson 2Human Factors Research Group, University of Nottingham, Nottingham, UK Search for other works by this author on: Oxford Academic Google Scholar Robert Houghton, Robert Houghton 1Horizon Digital Economy Research, University of Nottingham, Nottingham, Nottinghamshire, UK2Human Factors Research Group, University of Nottingham, Nottingham, UK Search for other works by this author on: Oxford Academic Google Scholar Richard Mortier, Richard Mortier 1Horizon Digital Economy Research, University of Nottingham, Nottingham, Nottinghamshire, UK3Computer Science, University of Nottingham, Nottingham, Nottinghamshire, UK Search for other works by this author on: Oxford Academic Google Scholar Murray Goulden, Murray Goulden 1Horizon Digital Economy Research, University of Nottingham, Nottingham, Nottinghamshire, UK Search for other works by this author on: Oxford Academic Google Scholar Unna Arunachalam Unna Arunachalam 1Horizon Digital Economy Research, University of Nottingham, Nottingham, Nottinghamshire, UK Search for other works by this author on: Oxford Academic Google Scholar Interacting with Computers, Volume 27, Issue 1, January 2015, Pages 60–71, https://doi.org/10.1093/iwc/iwu028 Published: 25 July 2014 Article history Received: 02 January 2014 Revision received: 06 June 2014 Accepted: 19 June 2014 Published: 25 July 2014

We present unikernels, a new approach to deploying cloud services via applications written in high-level source code. Unikernels are single-purpose appliances that are compile-time specialised into standalone kernels, and sealed against modification when deployed to a cloud platform. In return they offer significant reduction in image sizes, improved efficiency and security, and should reduce operational costs. Our Mirage prototype compiles OCaml code into unikernels that run on commodity clouds and offer an order of magnitude reduction in code size without significant performance penalty. The architecture combines static type-safety with a single address-space layout that can be made immutable via a hypervisor extension. Mirage contributes a suite of type-safe protocol libraries, and our results demonstrate that the hypervisor is a platform that overcomes the hardware compatibility issues that have made past library operating systems impractical to deploy in the real-world.

We present MultiNet, a novel method for securely associating devices with a domestic wireless network. We show that MultiNet has usability benefits over currently deployed commercial solutions while being backwards compatible with existing devices. MultiNet reduces the interaction overhead of secure association by focusing on users' interactions rather than the network's requirements. This leads to a novel architectural arrangement of the home network infrastructure: the network is dynamically re-configured to accept each pre-configured device, rather than the current norm where each device is configured to be acceptable to the pre-configured network. Assuming devices are pre-configured for a unique, device-specific network name and passphrase, MultiNet constructs an out-of-band visual channel via an intermediary network controller device to convey the device's configuration to the network. This makes the interaction to join a device to the wireless network lightweight and identical across all devices, considerably reducing the interaction overheads for users.

We explore the idea of making aesthetic decorative patterns that contain multiple visual codes. We chart an iterative collaboration with ceramic designers and a restaurant to refine a recognition technology to work reliably on ceramics, produce a pattern book of designs, and prototype sets of tableware and a mobile app to enhance a dining experience. We document how the designers learned to work with and creatively exploit the technology, enriching their patterns with embellishments and backgrounds and developing strategies for embedding codes into complex designs. We discuss the potential and challenges of interacting with such patterns. We argue for a transition from designing 'codes to patterns' that reflects the skills of designers alongside the development of new technologies.

The ‘Internet of Things’ (IoT) can be expected to radically increase the amount of potentially sensitive data gathered in our homes. This study explores the social implications of the presentation of data that could be collected within the household. In particular, it focuses on how ambiguities in these data, combined with existing interpersonal relationships, could influence social dynamics. Thirty-five participants were each presented with three separate household scenarios, involving ambiguous data that were collected and presented via near-future IoT technologies. Each participant was asked to respond to a series of open and closed questions about how they would interpret the data, how they would react to it and their general opinions of the technologies presented. Through qualitative and quantitative analysis of their responses, we contribute an understanding of how people interpret information about those around them. We find a common willingness to make inferences based on ambiguities within the data, even when participants are aware of the limitations of their understanding. We also find that sharing data produced via tagging of everyday objects raises a high level of privacy concern, and that, in a somewhat incoherent stance, users are more comfortable in sharing data publicly than in a targeted fashion with commercial organizations. Our findings also suggest that the age of the target user group has a greater effect on ease of use judgements than the nature of the technology, and we find some evidence that user’s interpretations can be biased by an individual’s age.

This paper extends the focus of current research into home networks. It represents a shift in perspective from the home network as something that is essentially understood as a technological object by the inhabitants of the home, to something that is understood by household members as a sociological object wrapped up in the organisation of their everyday lives. This shift in perspective is significant. It moves the focus of design from developing home network technologies that better support users' management of the home network and the devices that hang off it, to developing home network technologies that support household members' management of everyday life and the social activities that compose it. Through a range of ongoing ethnographic studies we elaborate this turn to the social, and a number of sensitising concerns informing the continued development of home network technologies.

OS virtualization and cloud computing have radically changed the way Internet services are deployed: enterprises share third-party datacenters, deploying existing applications with minimal changes. Recent measurements reveal a lack of traffic isolation capabilities within the datacenter with network performance exhibiting high variability. We advocate addressing this problem by allowing applications to express their own forwarding logic using OpenFlow to achieve application specific optimal performance. We present an OpenFlow implementation within the Mirage application synthesis framework, in the form of library implementations of a modular controller and an extensible OpenFlow-enabled switch, able to expose the underlying network infrastructure to cloud applications. By linking into the application, this provides a safe yet highly extensible framework for programming network control that, although unoptimised, still provides reasonable performance when compared with existing controllers.

Smartphones have exploded in popularity in recent years, becoming ever more sophisticated and capable. As a result, developers worldwide are building increasingly complex applications that require ever increasing amounts of computational power and energy. In this paper we propose ThinkAir, a framework that makes it simple for developers to migrate their smartphone applications to the cloud. ThinkAir exploits the concept of smartphone virtualization in the cloud and provides method-level computation offloading. Advancing on previous work, it focuses on the elasticity and scalability of the cloud and enhances the power of mobile cloud computing by parallelizing method execution using multiple virtual machine (VM) images. We implement ThinkAir and evaluate it with a range of benchmarks starting from simple micro-benchmarks to more complex applications. First, we show that the execution time and energy consumption decrease two orders of magnitude for a N-queens puzzle application and one order of magnitude for a face detection and a virus scan application. We then show that a parallelizable application can invoke multiple VMs to execute in the cloud in a seamless and on-demand manner such as to achieve greater reduction on execution time and energy consumption. We finally use a memory-hungry image combiner tool to demonstrate that applications can dynamically request VMs with more computational power in order to meet their computational requirements.

We consider the mobile phone location tracking problem in indoor environments. We propose a hybrid approach for energy-efficient indoor mobile phone localization by combining two well-known techniques namely, pedestrian dead reckoning and WiFi fingerprinting. Our preliminary evaluation of the proposed approach shows that it achieves better location accuracy (median error around 2m) than either of the two techniques on which it is based.

This demo presents MultiNet, a novel method for joining devices to a domestic Wi-Fi network. MultiNet dynamically reconfigures the network to accept each device, rather than configuring each device to fit the network as is the norm. It does so by assuming that each device is pre-configured with a cryptographically generated WPA2 network SSID/passphrase pair, and then providing a lightweight interaction through which the user creates a new network for each device. This approach makes securely adding devices to a wireless network straightforward without compromising security or burdening the user, and maintaining backward compatibility with existing deployed standards and protocols.

This demo presents Signposts, a system to provide users with a secure, simple mechanism to establish and maintain communication channels between their personal cloud of named devices. Signpost names exist in the DNSSEC hierarchy, and resolve to secure end-points when accessed by existing DNS clients. Signpost clients intercept user connection intentions while adding privacy and multipath support. Signpost servers co-ordinate clients to dynamically discover routes and overcome the middleboxes that pervade modern edge networks. The demo will show a simple scenario where an individual's personal devices (phone, laptop) are interconnected via Signposts while sitting on different networks behind various middleboxes. As a result they will be able to fetch and push data between each other, demonstrated by, e.g., simple web browsing, even as the network configuration changes.

This paper presents a user driven redesign of the domestic network infrastructure that draws upon a series of ethnographic studies of home networks. We present an infrastructure based around a purpose built access point that has modified the handling of protocols and services to reflect the interactive needs of the home. The developed infrastructure offers a novel measurement framework that allows a broad range of infrastructure information to be easily captured and made available to interactive applications. This is complemented by a diverse set of novel interactive control mechanisms and interfaces for the underlying infrastructure. We also briefly reflect on the technical and user issues arising from deployments.

Home networks have evolved to become small-scale versions of enterprise networks. The tools for visualizing and managing such networks are primitive and continue to require networked systems expertise on the part of the home user. As a result, non-expert home users must manually manage non-obvious aspects of the network - e.g., MAC address filtering, network masks, and firewall rules, using these primitive tools. The Homework information plane architecture uses stream database concepts to generate derived events from streams of raw events. This supports a variety of visualization and monitoring techniques, and also enables construction of a closed-loop, policy-based management system. This paper describes the information plane architecture and its associated policy-based management infrastructure. Exemplar visualization and closed-loop management applications enabled by the resulting system (tuned to the skills of non-expert home users) are discussed.

The Homework project has examined redesign of existing home network infrastructures to better support the needs and requirements of actual home users. Integrating results from several ethnographic studies, we have designed and built a home networking platform providing detailed per-flow measurement and management capabilities supporting several novel management interfaces. This demo specifically shows these new visualization and control interfaces (1), and describes the broader benefits of taking an integrated view of the networking infrastructure, realised through our router's augmented measurement and control APIs (2).

Despite the centrally-managed nature of and critical infrastructure provided by enterprise networks, analyses of their characteristics have been limited. In this paper we examine the dynamics of enterprise networks from two distinct perspectives, namely traffic and addressing. Using a large packet trace spanning approximately 3.5 weeks coupled with diverse other data sources, we pose and answer a series of questions pertinent to understanding the aforementioned aspects of todaypsilas enterprise networks. Our analysis results reveal characteristics regarding the geographical spread of traffic observed at a site in the enterprise network, the validity of the client-server model and mobility patterns within the Enterprise. Finally, we discuss the implications of our findings for tasks such as network management and dimensioning.

Enterprise network architecture and management have followed the Internet's design principles despite different requirements and characteristics: enterprise hosts are administered by a single authority, which intrinsically assigns different values to traffic from different business applications.

Routing loops are caused by inconsistencies in routing state among a set of routers. They occur in perfectly engineered networks, and have a detrimental effect on performance. They impact end-to-end performance through increased packet loss and delay for packets caught in the loop, and through increased link utilization and corresponding delay and jitter for packets that traverse the link but are not caught in the loop.Using packet traces from a tier-1 ISP backbone, we first explain how routing loops manifest in packet traces. We characterize routing loops in terms of the packet types caught in the loop, the loop sizes, and the loop durations. Finally, we analyze the impact of routing loops on network performance in terms of loss and delay.

Today's IP backbones are provisioned to provide excellent performance in terms of loss, delay and availability. However, performance degradation and service disruption are likely in the case of failure, such as fiber cuts, router crashes, etc. In this paper, we investigate the occurence of failures in Sprint's IP backbone and their potential impact on emerging services such as Voice-over-IP (VoIP). We first examine the frequency and duration of failure events derived from IS-IS routing updates collected from three different points in the Sprint IP backbone. We observe that link failures occur as part of everyday operation, and the majority of them are short-lived (less than 10 minutes). We also discuss various statistics such as the distribution of inter-failure time, distribution of link failure durations, etc. which are essential for constructing a realistic link failure model. Next, we present an analysis of routing and service reconvergence time during a controlled link failure scenario in our backbone. Our results indicate that disruption to packet forwarding after link failures depends not only on routing protocol dynamics, but also on the design of routers' architectures and control planes. Thus our results offer insights into two basic components for defining network-wide availability, which we consider a more appropriate metric for service-level agreements to support emerging applications.

Internet protocols currently use packet-level mechanisms to detect and react to congestion. Although these controls are essential to ensure fair sharing of the available resource between multiple flows, in some cases they are insufficient to ensure overall network stability. We believe that it is also necessary to take account of higher level concepts, such as connections, flows, and sessions when controlling network congestion. This becomes of increasing importance as more real-time traffic is carried on the Internet, since this traffic is less elastic in nature than traditional Web traffic. We argue that, in order to achieve better utility of the network as a whole, higher level congestion controls are required. By way of example, we present a simple connection admission control (CAC) scheme which can significantly improve the overall performance. This paper discusses our motivation for the use of admission control in the Internet, focusing specifically on control for TCP flows. The technique is not TCP specific, and can be applied to any type of flow in a modern IP infrastructure. Simulation results are used to show that it can drastically improve the performance of TCP over bottleneck links. We go on to describe an implementation of our algorithm for a router running the Linux 2.2.9 operating system. We show that by giving routers at bottlenecks the ability to intelligently deny admission to TCP connections, the goodput of existing connections can be significantly increased. Furthermore, the fairness of the resource allocation achieved by TCP is improved.