This talk does not have an abstract.

This talk does not have an abstract.

We present a complete Visible Light Communications (VLC) transceiver system consisting of low-cost Commercial-Off-The-Shelf (COTS) components. In particular, we show that COTS IEEE 802.11n (WiFi) devices can be used so that the physical and data link layers of radio frequency (RF) WiFi, i.e. 2.4 GHz, are reused for VLC. Moreover, as WiFi is fully integrated with the Linux system, higher protocols from network to transport and application layer can be used and tested in VLC-related experiments. Our approach has the advantage that a VLC experimenter can fully focus on VLC-related low-level aspects like the design of novel VLC front-ends, e.g. LED drivers, lenses, and photodetectors and test their impact directly on the full network protocol stack in an end-to-end manner with real applications like adaptive video streaming. We present first results from experiments using our prototype showing the performance of unidirectional VLC transmission. Here we analyze the distortions introduced as well as the relationship between signal strength on frame error rate for different MCS and the maximum communication distance. Experimental results reveal that a data rate of up to 150 Mbps is possible over short ranges.

Recently, the expansion of wireless network deployments is resulting in increased scarcity of available licensed radio spectrum. As the domain of wireless communications is progressing rapidly, many industries are looking into wireless network solutions that can increase their productivity. Private LTE is a promising wireless network solution as it can be customised independently without the control of a mobile network operator while providing reliable and spectrum efficient services. For this reason, the deployment of Private LTE in the unlicensed spectrum and its coexistence with Wi-Fi is becoming a popular topic in research. In this paper, we propose a coexistence scheme for private LTE network in unlicensed spectrum that enables a fair spectrum sharing with co-located Wi-Fi networks. This is achieved by exploiting various LTE frame configurations consisting of different combinations of downlink, uplink, special subframe and muted subframes. The configuration of a single frame is decided based on a rule based algorithm that exploits Wi-Fi spectrum occupancy statistics that is obtained from a technology recognition system which is based on a Convolutional Neural Network. The performance of the proposed private LTE scheme and its coexistence with Wi-Fi is investigated for different traffic scenarios showcasing how the proposed scheme can lead to a harmless coexistence of LTE and Wi-Fi.

Applications that require extremely low latency are expected to be a major driver of 5G and WLAN networks that include millimeter wave (mmWave) links. However, mmWave links can experience frequent, sudden changes in link capacity due to obstructions in the signal path. These dramatic variations in link capacity cause a temporary "bufferbloat" condition during which delay may increase by a factor of 2-10. Low latency congestion control protocols, which manage bufferbloat by minimizing queue occupancy, represent a potential solution to this problem, however their behavior over links with dramatic variations in capacity is not well understood. In this paper, we explore the behavior of two major low latency congestion control protocols, TCP BBR and TCP Prague (as part of L4S), using link traces collected over mmWave links under various conditions. Our evaluation reveals potential problems associated with use of these congestion control protocols for low latency applications over mmWave links.

This talk does not have an abstract.

The Border Gateway Protocol (BGP) is the single routing protocol that glues the Internet together. Its performance, especially the convergence speed after path changes, is key to global efficiency, also in light of the fact that the number of Autonomous Systems (ASes) and Subnets has reached a level that makes path changes a frequent event. This work presents a testbed-based experimental analysis of BGP convergence time under different hypothesis of Minimum Route Advertisement Interval (MRAI) setting and a proposal to improve it by setting MRAI based on the topological position of the ASes. MRAI is a timer that regulates the frequency of successive UPDATE messages sent by a BGPs router for a given route and destination. The work is based on the modifications of the BIRD BGP daemon and shows that it is possible to execute experiments on testbeds with topologies that have Internet-like characteristics scaling up to thousands of ASes.

Research testbed fabrics have potential to support long-lived, evolving, interdomain experiments, including opt-in application traffic across multiple campuses and edge sites. We propose abstractions and security infrastructure to facilitate multi-domain networking, and a reusable controller toolkit (ExoPlex) for network service providers (NSPs) running in testbed-hosted virtual network slices. We demonstrate the idea on the ExoGENI testbed, which allows slices to interconnect and exchange traffic over peering links by mutual consent. Each ExoPlex NSP runs a peering controller that manages its interactions with its linked peers and controls the NSP's dataplane network via SDN. Our approach expresses policies for secure peering and routing in a declarative language---logical peering. The prototype uses logic rules to verify IP prefix ownership, filter and validate route advertisements, and implement user-specified policies for connectivity and path control in networks with multiple transit NSPs.

The Slice-based Federation Architecture (SFA) is the de facto standard framework for managing testbeds, federations of testbeds, and users access to these federations. It is the foundation of most major testbed federations in the world, including GENI and Fed4FIRE. However, there remain some testbeds that were designed and grew outside of this world, making different, interesting and sometimes better design choices. In this paper, we describe how we added support for the GENI Aggregate Manager to the Grid'5000 testbed, a major testbed focused on HPC and Cloud that was developed for the most part independently for the last 15 years. From this experience, we draw some lessons and recommendations that could help improve the testbed management ecosystem.

The rapid technological advances in the Internet of Things (IoT) allows the blueprint of Smart Cities to become feasible by integrating heterogeneous cloud/fog/edge computing paradigms to collaboratively provide variant smart services in our cities and communities. Thanks to attractive features like fine granularity and loose coupling, the microservices architecture has been proposed to provide scalable and extensible services in large scale distributed IoT systems. Recent studies have evaluated and analyzed the performance interference between microservices based on scenarios on the cloud computing environment. However, they are not holistic for IoT applications given the restriction of the edge device like computation consumption and network capacity. This paper investigates multiple microservice deployment policies on edge computing platform. The microservices are developed as docker containers, and comprehensive experimental results demonstrate the performance and interference of microservices running on benchmark scenarios.

Live migration is a technology that seamlessly moves a virtualized running service between hosts, which allows services to rapidly adapt to changes. The principal of this technology is that the application downtime is small enough so the end user is not aware of the migration. The two most used off-the-shelf technologies for virtualization are docker for containers and KVM for virtual machines. However, there is still a lack of understanding of its performance in live migration. In this paper, we present a demonstrator that migrates different applications using docker and KVM under different conditions that can be configured manually. The audience will be able to select an application from a pool of applications and the conditions of the migration. Then, the migration will be triggered and the audience will observe different statistics such as the migration time, the downtime, and the resources consumed. Finally, they will be able to compare it to previous runs and observe the impact of the conditions selected.

High bandwidth, low latency modern-day applications are extremely sensitive to variations in network parameters. In-band Network Telemetry (INT) can help in extracting forwarding plane dynamics of the network element and characterize the service path to make preemptive service modifications. In this demonstration, we showcase a programmable telemetry framework via FPGA based NICs installed in e-science Data Transfer Nodes (DTN) operating over Ciena's Research Network Innovation Platform (CENI). We demonstrate the benefits of this framework in accelerating fault localization mechanisms and implementing corrective actions, in a closed loop with Software Defined Network (SDN) Orchestrators.

We present a complete Visible Light Communications (VLC) transceiver system consisting of low-cost Commercial-Off-The-Shelf (COTS) components. In particular, we show that COTS IEEE 802.11n (WiFi) devices can be used so that the physical and data link layers of radio frequency (RF) WiFi, i.e. 2.4 GHz, are reused for VLC. Moreover, as WiFi is fully integrated with the Linux system, higher protocols from network to transport and application layer can be used and tested in VLC-related experiments. Our approach has the advantage that a VLC experimenter can fully focus on VLC-related low-level aspects like the design of novel VLC front-ends, e.g. LED drivers, lenses, and photodetectors and test their impact directly on the full network protocol stack in an end-to-end manner with real applications like adaptive video streaming. We present first results from experiments using our prototype showing the performance of unidirectional VLC transmission. Here we analyze the distortions introduced as well as the relationship between signal strength on frame error rate for different MCS and the maximum communication distance. Experimental results reveal that a data rate of up to 150 Mbps is possible over short ranges.

The rapid technological advances in the Internet of Things (IoT) allows the blueprint of Smart Cities to become feasible by integrating heterogeneous cloud/fog/edge computing paradigms to collaboratively provide variant smart services in our cities and communities. Thanks to attractive features like fine granularity and loose coupling, the microservices architecture has been proposed to provide scalable and extensible services in large scale distributed IoT systems. Recent studies have evaluated and analyzed the performance interference between microservices based on scenarios on the cloud computing environment. However, they are not holistic for IoT applications given the restriction of the edge device like computation consumption and network capacity. This paper investigates multiple microservice deployment policies on edge computing platform. The microservices are developed as docker containers, and comprehensive experimental results demonstrate the performance and interference of microservices running on benchmark scenarios.

The Internet landscape is progressively transitioning towards a flat structure to prune multiple Internet Service Provider (ISP) layers. At the core of this transition are the Internet Exchange Points (IXPs) which play a critical role in mediating traffic exchange among ISPs. As such, it is pertinent to take a closer look at IXPs and accurately emulate their operative model into a computation model that enables a concrete characterization without losing generality. However, to the best of our knowledge, there is no existing model that is being used for understanding the behavior of IXPs, and as a result, various functional and operational properties of IXPs remain invisible to the research community. In this paper, we address this gap by proposing GENIX; a modular emulation framework that mimics the behavior of IXPs on a public test-bed. GENIX is implemented in Global Environment for Network Innovations (GENI) and is capable of enumerating various interactions among ISPs, which we expect in the real world. GENIX is the first effort towards alleviating the complexities of large-scale emulation to abstract a lightweight design where our observations of the IXPs can be usefully applied. We test GENIX for its ability to handle large volumes of traffic flow and its performance during network congestion.

This talk does not have an abstract.