Search

QUIC is the new transport protocol for the Internet, designed for secure, reliable communication, especially with the web in mind. While multimedia streaming, allowing for some playout delay, has been widely run on top of reliable transport protocols, conversational multimedia usually requires unreliable ones, often using RTP over UDP. A recent extension to QUIC supports unreliable datagrams within QUIC connections so that also real-time media can be supported. In this paper, we investigate a recent design for RTP over QUIC with a focus on congestion control and the related signaling. We implement a strawman using Gstreamer and quic-go and evaluate different permutations of congestion control algorithms and signaling in a simple testbed.

Bluetooth Low Energy (BLE) is today's most popular low-power radio technology with compelling radio performance and battery-friendly characteristics, making it a promising deployment option for the Internet of Things (IoT). Little is known, however, about the performance and pitfalls when utilizing BLE as link layer in multi-hop IP over BLE scenarios, because of the lack of available software platforms and deployment experiences. In this work, we present both a fully open-source, configurable software platform and experiments to analyze multi-hop BLE network behavior. Our experiments, conducted in a larger testbed, reveal unexpected performance drawbacks. Even in scenarios with underutilized links, BLE connections break randomly. This results into large transmission delays on the network layer and thus hinders real-world deployments in the constrained IoT. As key reason for this behavior we identify the BLE connection interval. A deterministic interval leads to unpredictable link behavior and connection losses due to overlapping connection events. We propose randomizing connection intervals as mitigation strategy and demonstrate that this prevents connection losses and sporadic link degradation, improving the overall network behavior.

Inter-DataCenter Wide Area Network (Inter-DC WAN) that connects geographically distributed data centers is becoming one of the most critical network infrastructures. Due to limited bandwidth and inevitable link failures, it is highly challenging to guarantee network availability for services, especially those with stringent bandwidth demands, over inter-DC WAN. We present BATE, a novel Traffic Engineering (TE) framework for bandwidth availability (BA) provision, which aims to ensure that each bandwidth demand must be satisfied with a stipulated probability, when subjected to the network capacity and possible failures of the inter-DC WAN. The three core components of BATE, i.e., admission control, traffic scheduling and failure recovery, are formulated through different mathematical models and theoretically analyzed. They are also extensively compared against state-of-the-art TE schemes, using a testbed as well as real trace driven simulations across different topologies, traffic matrices and failure scenarios. Our evaluations show that, compared with the optimal admission strategy, BATE can speed up the online admission control by 30x at the expense of less than 4% false rejections. On the other hand, compared with the latest TE schemes like FFC and TEAVAR, BATE can meet the bandwidth availability targets for 23%~60% more demands under normal loads, and when network failure causes BA targets violations.

Stream processing applications are becoming increasingly important in areas such as IoT, video analytics and social media. As a result, developers and operators must meet stringent time-to-market and scale requirements before bringing them to production. Unfortunately, testing a networked stream processing system is currently a cumbersome process that usually requires an expensive testbed and deep expertise on both networking and distributed systems. In this poster, we present Raptor, a tool for the fast prototyping of large-scale networked stream processing applications. Raptor builds on Mininet and Apache Kafka, two widely adopted platforms, to enable stakeholders to easily test their solutions under various operational conditions. Through a reasonably large setup (20 nodes) running on a single server, we show how unbalanced Kafka's leader selection algorithm can be and its implications on the overall system's throughput. We envision this work can help paving the way for more reproducible research in the stream processing domain, currently a first-class network application.

In scientific research, the independent reproduction of experimental results is the source of trust. The release of experimental artifacts enables the reproduction of results; however, additional efforts of researchers are required to prepare and document their experiments accordingly. To honor this increased effort, multiple initiatives were implemented to incentivize the creation and release of experimental artifacts, e.g., awards for papers that provide experimental artifacts. In this work, we want to propose a novel approach toward a reproducible research---a structured experimental workflow that allows the creation of reproducible experiments without requiring additional efforts of the researcher. Moreover, we present our own testbed and toolchain, namely, plain orchestrating service (pos), which enables the creation of such experimental workflows. The experiment is documented by our proposed, fully scripted experiment structure. Further, we consider the entire experimental workflow from experiment orchestration, to data measurement, to result evaluation. In addition, pos provides scripts that enable the automation of the bundling and release of all the created experimental artifacts. In this case study, we release one of our own experiments together with the necessary tools so that others can reproduce our experiment. Additionally, we provide an interactive environment where pos experiments can be executed and reproduced, which is available at https://gallenmu.github.io/pos-artifacts.

Monitoring WiFi networks is essential to gain insight into the network operation and develop methods capable of reacting to network dynamics. However, research and development in this field are hindered because there is a lack of a framework that can be easily extended to collect various types of monitoring data from the WiFi stack. In this paper, we propose FLIP, a framework for leveraging eBPF to augment WiFi access points and investigate the performance of WiFi networks. Using this framework, we focus on two important aspects of monitoring the WiFi stack. First, considering the high delay experienced by packets at access points, we show how switching packets from the wired interface to the wireless interface can be monitored and timestamped accurately at each step. We build a testbed using FLIP access points and investigate the factors affecting packet delay experienced in access points. Second, we present a novel approach that allows access points to track the duty-cycling pattern and energy consumption of their associated stations accurately and without the need for any external energy measurement tools. We validate the high energy measurement accuracy of FLIP by empirical experiments and comparisons against a commercial tool.

Testbeds remain indispensable instruments for experimentally evaluating IoT-oriented low-power wireless networking solutions. With the evolution of the field, they are increasingly expected to match envisioned deployment conditions of such solutions, notably in terms of scale. However, large-scale testbeds are scarce, likely because they have been believed to be expensive. This paper argues that this belief need no longer be justified by presenting the architecture and basic properties of 1KT, our new smart-building IoT testbed for solutions utilizing IEEE 802.15.4 and Bluetooth Low Energy. It comprises 1000 experimental devices deployed directly in human spaces of 168 rooms on all 5 floors of a sizable building. At the same time, its cost is relatively low considering the scale.

Convolutional neural network (CNN)-based object detection is a key technology to enable autonomous mobile vision applications on mobile end devices such as smart phones and drones. With the advance of edge computing technology, a prevalent solution is to offload the computation-intensive CNN inference tasks to edge networks for fast and accurate object detection. However, a single edge server may not be powerful enough to ensure fast and accurate object detection due to its resource constraints. In this paper, we propose a multi-edge assisted fast object detection framework, MASS, to further reduce the object detection latency while maintaining the detection accuracy. In MASS, the CNN model is divided into two parts, namely the Head part and the Tail part. The Head part is executed locally and the Tail part is further split into multiple subtasks, which can be offloaded to multiple edge servers for parallel execution. First, we propose a method to select the optimal parallel entry point to separate the Head part from the Tail part. Then, an adaptive subtask generation and offloading strategy is proposed to divide the Tail part into multiple subtasks and offload these subtasks to multiple heterogeneous edge servers. Besides, we also propose a uniformly sampled zero-padding scheme to reduce the communication cost among edge servers when executing these subtasks in parallel. We implement MASS in a testbed with four edge servers and evaluate its performance, where the experimental results show that MASS can reduce object detection latency by up to $64.83%$, while the detection accuracy degradation is less than $3%$.

Bump-in-the-wire (bump) devices can be used to protect critical endpoints in Industrial Control System (ICS) networks. However, bump devices cannot be used to authenticate incoming broadcast traffic, are complex to manage, and one bump is needed per host. In this work, we propose a virtual bump-like solution called vBump, which allows to insert virtual bumps in front of Ethernet-based legacy ICS devices. The vBumps can be used to limit traffic to whitelisted destinations, inspect all traffic on or above Link-layer like a centralized intrusion detection systems (or monitoring systems), or even police the traffic like a centralized intrusion prevention systems. In particular, this also allows the network to apply fine-grained control on traffic between nodes that need to be in the same Link-layer broadcast domain. Compared to traditional bumps, vBumps do not require any changes in physical network topology, and the central server's global view allows for more informed decision, with less computational constraints. We implement the system in a high-fidelity ICS testbed, and demonstrate its capabilities to support even time-critical protection control traffic in smart grids. Our system can handle traffic rates of 150Mbps with one-way delay of ~1ms.

While cloud computing is transforming society, today's public clouds are black boxes, implemented and operated by a single provider that makes all business and technology decisions. In 2013 we launched the Mass Open Cloud (MOC) with the vision of creating a production cloud that would enable innovation by a broad industry and research community. This open cloud has become a laboratory for cloud research and innovation, resulting in hundreds of publications, contributions to open source software, and collaborations between researchers, open source developers, and production operations staff. Recently we launched the Open Research Cloud Initiative (ORCI) to provide a framework to coordinate the bazaar of interrelated projects and initiatives that have evolved since 2013, including the Red Hat Collaboratory@BU, Open Cloud Testbed (OCT), New England Research Cloud (NERC), Northeast Storage Exchange (NESE), Operate First, and OpenInfra Labs. With its launch, the MOC inspired and enabled research in cloud security. For example, the Modular Approach to Cloud Security (MACS) SaTC NSF frontier project, launched in 2014, brought together cryptographers, operating system, database and computer architecture researchers from BU, MIT, UConn and NEU. This security research resulted in new open-source software and products that are today enabling new services in the ORCI bazaar. This talk will discuss the ORCI cloud bazaar, some of the security research and projects it inspired, and some exciting new collaborations happening now to make the cloud both open and secure.

Person Re-ID is a challenging problem and is gaining more attention due to demands in security, intelligent system and other applications. Most person Re-ID works are vision-based, such as image, video, or broadly speaking, face recognition-based techniques. Recently, several multi-modal person Re-ID datasets were released, including RGB+IR, RGB+text, RGB+WiFi, which shows the potential of the multi-modal sensor-based person Re-ID approach. However, there are several common issues in public datasets, such as short time duration, lack of appearance change, and limited activities, resulting in un-robust models. For example, vision-based Re-ID models are sensitive to appearance change. In this work, a person Re-ID testbed with multi-modal sensors is created, allowing the collection of sensing modalities including RGB, IR, depth, WiFi, radar, and audio. This novel dataset will cover normal daily office activities with large time span over multi-seasons. Initial analytic results are obtained for evaluating different person Re-ID models, based on small datasets collected in this testbed.

Edge computing for the Internet of Things prescribes executing applications on server machines closer to devices rather than depending on the cloud. However, server machines are expensive, are not flexible to adapt to varying application requirements, require gateways to interact with IoT devices, and follow a centralized model which increases traffic and application latency. Special-purpose hardware for the edge is becoming increasingly sophisticated, with support for machine learning, secure enclaves etc., and this work is an attempt to leverage such hardware to cooperatively execute edge applications, rather than relying on expensive edge servers. To do so, our design relies on a distributed middleware which can seamlessly scale up with new hardware, and a task scheduler which best matches application requirements with the hardware capabilities available. We have built a prototype middleware that operates on multiple gateways in our testbed of 250 IoT devices, and we plan to further improve our platform to support more varying use cases.

Recent years have witnessed an emerging class of real-time applications, e.g., autonomous driving, in which resource-constrained edge platforms need to execute a set of real-time mixed Deep Learning (DL) tasks concurrently. Such an application paradigm poses major challenges due to the huge compute workload of deep neural network models, diverse performance requirements of different tasks, and the lack of real-time support from existing DL frameworks. In this paper, we present RT-mDL, a novel framework to support mixed real-time DL tasks on edge platform with heterogeneous CPU and GPU resource. RT-mDL aims to optimize the mixed DL task execution to meet their diverse real-time/accuracy requirements by exploiting unique compute characteristics of DL tasks. RT-mDL employs a novel storage-bounded model scaling method to generate a series of model variants, and systematically optimizes the DL task execution by joint model variants selection and task priority assignment. To improve the CPU/GPU utilization of mixed DL tasks, RT-mDL also includes a new priority-based scheduler which employs a GPU packing mechanism and executes the CPU/GPU tasks independently. Our implementation on an F1/10 autonomous driving testbed shows that, RT-mDL can enable multiple concurrent DL tasks to achieve satisfactory real-time performance in traffic light detection and sign recognition. Moreover, compared to state-of-the-art baselines, RT-mDL can reduce deadline missing rate by 40.12% while only sacrificing 1.7% model accuracy.

As intelligence is moving from data centers to the edges, intelligent edge devices such as smartphones, drones, robots, and smart IoT devices are equipped with the capability to altogether train a deep learning model on the devices from the data collected by themselves. Despite its considerable value, the key bottleneck of making on-device distributed training practically useful in real-world deployments is that they consume a significant amount of training time under wireless networks with constrained bandwidth. To tackle this critical bottleneck, we present Mercury, an importance sampling-based framework that enhances the training efficiency of on-device distributed training without compromising the accuracies of the trained models. The key idea behind the design of Mercury is to focus on samples that provide more important information in each training iteration. In doing this, the training efficiency of each iteration is improved. As such, the total number of iterations can be considerably reduced so as to speed up the overall training process. We implemented Mercury and deployed it on a self-developed testbed. We demonstrate its effectiveness and show that Mercury consistently outperforms two status quo frameworks on six commonly used datasets across tasks in image classification, speech recognition, and natural language processing.

WiFi-based Human Gesture Recognition (HGR) becomes increasingly promising for device-free human-computer interaction. However, existing WiFi-based approaches have not been ready for real-world deployment due to the limited scalability, especially for unseen gestures. The reason behind is that when introducing unseen gestures, prior works have to collect a large number of samples and re-train the model. While the recent advance of few-shot learning has brought new opportunities to solve this problem, the overhead has not been effectively reduced. This is because these methods still require enormous data to learn adequate prior knowledge, and their complicated training process intensifies the regular training cost. In this paper, we propose a WiFi-based HGR system, namely OneFi, which can recognize unseen gestures with only one (or few) labeled samples. OneFi fundamentally addresses the challenge of high overhead. On the one hand, OneFi utilizes a virtual gesture generation mechanism such that the massive efforts in prior works can be significantly alleviated in the data collection process. On the other hand, OneFi employs a lightweight one-shot learning framework based on transductive fine-tuning to eliminate model re-training. We additionally design a self-attention based backbone, termed as WiFi Transformer, to minimize the training cost of the proposed framework. We establish a real-world testbed using commodity WiFi devices and perform extensive experiments over it. The evaluation results show that OneFi can recognize unseen gestures with the accuracy of 84.2, 94.2, 95.8, and 98.8% when 1, 3, 5, 7 labeled samples are available, respectively, while the overall training process takes less than two minutes.

The advent of Persistent Memory (PM) necessitates an evolution of Remote Direct Memory Access (RDMA) technologies for supporting remote data persistence. Previous software-based solutions require remote CPU intervention and postpone the visibility of remote persistence. In this paper, we design several hardware-supported RDMA primitives to flush data from the volatile cache of RDMA Network Interface Cards (RNICs) to the PM. We also propose durable RPCs based on the proposed RDMA Flush primitives to support remote data persistence and fast failure recovery. We emulate the performance of RDMA Flush primitives through other RDMA primitives, and compare our proposals with several state-of-the-art RPCs in a real testbed equipped with PM and InfiniBand networks. Experimental results show that our proposals can improve the throughput of RPCs by up to 90%, and reduce the 99th percentile latency by up to 49%. The experimental studies also provide instructive guidelines for designing RDMA-based distributed PM systems.

This paper takes a first look at the potential consequences of cyberattacks against structural control systems. We design algorithms and implement them in a testbed and on well-known benchmark models for buildings and bridges. Our results show that attacks to structures equipped with semi-active and active vibration control systems can let the attacker oscillate the building or bridge at the resonance frequency, effectively generating threats to the structure and the people using it. We also implement and test the effectiveness of attack-detection systems.

With recent advances in augmented reality (AR) and computer vision it has become possible to magnify objects in real time in a user’s field of view. AR object magnification can have different purposes, such as enhancing human visual capabilities with the BigHead technique, which works by up-scaling human heads to communicate important facial cues over longer distances. For this purpose, we created a prototype with a 4K camera mounted on a HoloLens 2. In this demo, we present the BigHead technique and proof of concept AR testbed to magnify heads in real-time. Further, we describe how hand gestures are detected to control the scale and position of the magnified head. We discuss the technique and implementation, and propose future research directions.

Evacuation planning is a critical task in disaster management. Especially in situations such as natural disasters or terrorist attacks, large crowds need to move away from danger and reach designated safe zones. For this purpose, various approaches that efficiently compute evacuation plans in urban areas have been proposed. To evaluate the computed plans, previous works employ heuristics that can only roughly estimate the egress time of each plan. Intuitively, a much better approach is to estimate the egress time via simulation. However, designing a simulation model is usually a time-consuming task and, what is more, this model can only be used to evaluate evacuation plans for a specific area. In this paper, we address these issues presenting EURASIM. Our system enables the automated generation of simulation models for urban areas. Furthermore, EURASIM is designed in a way that algorithms for evacuation planning can be easily integrated, thus functioning as a testbed for the development of even better solutions.

Recently, deep learning-based localization systems have become one of the most promising techniques due to their accuracy in complex environments. However, these techniques require large amounts of data for training. Obtaining such data is usually a tedious and time-consuming process, which hinders their practical deployment. In this paper, we propose a data augmentation framework for deep learning-based localization systems. The basic idea is to use a conditional Generative Adversarial Network that is able to learn the complex structures in the original training data and then generate high-quality synthetic data that matches the original data distribution. Evaluation of the proposed data augmentation framework in a real testbed shows that our technique can increase the average localization accuracy by 22.2% compared to the case of not using data augmentation. This demonstrates the promise of the proposed framework for enhancing deep learning-based localization systems.

Constructive multi-beam communication for mmWave is a new trend where the base station utilizes multiple paths to improve link throughput and reliability. This demo addresses a key challenge on maintaining a mmWave multi-beam link under user mobility. Traditional user-tracking algorithms for single-beam links do not directly apply to a multi-beam link due to large probing overhead. We demonstrate a low-overhead channel-based solution that only requires multi-beam channel estimates in the background to track multi-beams. We collect and release a one-of-its-kind channel dataset using our 28 GHz mMobile testbed.

As evolving wireless network architectures become more diverse, complex, and interdependent, and equipment costs prohibit broad access to such networks, remotely accessible experimental testbeds are gaining interest in recent years in wireless communication and networking research. This interest has exacerbated in 2020 and became a vital need during the current global pandemic. However, providing end-users of various educational backgrounds access to radio devices from a heterogeneous set of vendors is challenging. This paper introduces OneLNK, a remotely accessible testbed consisting of radio devices from three different vendors and developed using open source cloud-native technologies. End-users can access the functionalities of OneLNK from a single webpage without any local installations. Using the web URL, users can operate radio devices, set experiment parameters, observe results in real-time, and save generated experiment data for all radio devices. The interactive web UI and its working mechanism for supporting radio equipment are covered with specific experiment capabilities. A diverse set of radio equipment (mmWave, sub-GHz SDR, and sub-6GHz SDR) are facilitated to explain these capabilities. Moreover, measurements of path loss, Received Signal Strength (RSS), and Signal-to-Noise Ratio (SNR) using devices from three different vendors operating on a vast spectrum (568 MHz, 5.8 GHz, and 60 GHz) are reported. The majority of the remotely accessible OneLNK platform was developed remotely during the pandemic by a team of experts from five U.S. states.

While millimeter-wave (mmWave) wireless has recently gained tremendous attention with the transition to 5G, developing a broadly accessible experimental infrastructure will allow the research community to make significant progress in this area. Hence, in this paper, we present the design and implementation of various programmable and open-access 28/60 GHz software-defined radios (SDRs), deployed in the PAWR COSMOS advanced wireless testbed. These programmable mmWave radios are based on the IBM 28 GHz 64-element dual-polarized phased array antenna module (PAAM) subsystem board and the Sivers IMA 60 GHz WiGig transceiver. These front ends are integrated with USRP SDRs or Xilinx RF-SoC boards, which provide baseband signal processing capabilities. Moreover, we present measurements of the TX/RX beamforming performance and example experiments (e.g., real-time channel sounding and RFNoC-based 802.11ad preamble detection), using the mmWave radios. Finally, we discuss ongoing enhancement and development efforts focusing on these radios.

5G networks are gaining pace in development and deployment in recent years. One of 5G's key objective is to support a variety of use cases with different Service Level Objectives (SLOs). Slicing is a key part of 5G that allows operators to provide a tailored set of resources to different use cases in order to meet their SLOs. Existing works focus on slicing in the frontend or the C-RAN. However, slicing is missing in the fronthaul network that connects the frontend to the C-RAN. This leads to over-provisioning in the fronthaul and the C-RAN, and also limits the scalability of the network. In this paper, we design and implement Fronthaul Slicing Architecture (FSA), which to the best of our knowledge, is the first slicing architecture for the fronthaul network. FSA runs in the switch dataplane and uses information from the wireless schedule to identify the slice of a fronthaul data packet at line-rate. It enables multipoint-to-multipoint routing as well as packet prioritization to provide multiplexing gains in the fronthaul and the C-RAN, making the system more scalable. Our testbed evaluation using scaled-up LTE traces shows that FSA can support accurate multipoint-to-multipoint routing for 80 Gbps of fronthaul traffic. Further, the slice-aware packet scheduling enabled by FSA's packet prioritization reduces the 95th percentile Flowlet Completion Times (FCT) of latency-sensitive traffic by up to 4 times.

Infrastructure-assisted autonomous driving is an emerging paradigm that aims to make affordable autonomous vehicles a reality. A key technology for realizing this vision is real-time point cloud registration which allows a vehicle to fuse the 3D point clouds generated by its own LiDAR and those on roadside infrastructures such as smart lampposts, which can deliver increased sensing range, more robust object detection, and centimeter-level navigation. Unfortunately, the existing methods for point cloud registration assume two clouds to share a similar perspective and large overlap, which result in significant delay and inaccuracy in real-world infrastructure-assisted driving settings. This paper proposes VI-Eye - the first system that can align vehicle-infrastructure point clouds at centimeter accuracy in real-time. Our key idea is to exploit traffic domain knowledge by detecting a set of key semantic objects including road, lane lines, curbs, and traffic signs. Based on the inherent regular geometries of such semantic objects, VI-Eye extracts a small number of saliency points and leverage them to achieve real-time registration of two point clouds. By allowing vehicles and infrastructures to extract the semantic information in parallel, VI-Eye leads to a highly scalable architecture for infrastructure-assisted autonomous driving. To evaluate the performance of VI-Eye, we collect two new multiview LiDAR point cloud datasets on an indoor autonomous driving testbed and a campus smart lamppost testbed, respectively. They contain total 915 point cloud pairs and cover three roads of 1.12km. Experiment results show that VI-Eye achieves centimeter-level accuracy within around 0.2s, and delivers a 5X improvement in accuracy and 2X speedup over state-of-the-art baselines.

Social distancing can reduce infection rates in respiratory pandemics such as COVID-19, especially in dense urban areas. To assess pedestrians' compliance with social distancing policies, we use the pilot site of the PAWR COSMOS wireless edge-cloud testbed in New York City to design and evaluate an Automated video-based Social Distancing Analyzer (Auto-SDA) pipeline. Auto-SDA derives pedestrians' trajectories and measures the duration of close proximity events. It relies on an object detector and a tracker, however, to achieve highly accurate social distancing analysis, we design and incorporate 3 modules into Auto-SDA: (i) a calibration module that converts 2D pixel distances to 3D on-ground distances with less than 10 cm error, (ii) a correction module that identifies pedestrians who were missed or assigned duplicate IDs by the object detection-tracker and rectifies their IDs, and (iii) a group detection module that identifies affiliated pedestrians (i.e., pedestrians who walk together as a social group) and excludes them from the social distancing violation analysis. We applied Auto-SDA to videos recorded at the COSMOS pilot site before the pandemic, soon after the lockdown, and after the vaccines became broadly available, and analyzed the impacts of the social distancing protocols on pedestrians' behaviors and their evolution. For example, the analysis shows that after the lockdown, less than 55% of the pedestrians violated the social distancing protocols, whereas this percentage increased to 65% after the vaccines became available. Moreover, after the lockdown, 0-20% of the pedestrians were affiliated with a social group, compared to 10-45% once the vaccines became available. Finally, following the lockdown, the density of the pedestrians at the intersection decreased by almost 50%.

This paper describes a wireless experimentation framework for studying dynamic spectrum access mechanisms and an experiment that showcases its capabilities. The framework was built on COSMOS, an advanced wireless testbed designed to support real-world experimentation of next generation wireless technologies and applications. Our deployed framework supports experimentation over a large number of wireless networks, with a PUB-SUB based network interaction structure, based on the Collaborative Intelligent Radio Networks (CIRN) Interaction Language (CIL) developed by DARPA for the Spectrum Collaboration Challenge (SC2). As such, it enables interaction and message exchanges between the networks for the purposes of coordinating spectrum use. For our experiment, the message exchanges are aimed primarily for, but not limited to, Spectrum Consumption Model (SCM) messages. RF devices/systems use SCM messages which contain detailed information about their wireless transmission characteristics (i.e., spectrum mask, frequency, bandwidth, power and location) to determine their operational compatibility (non-interference) with prior transmitters and receivers, and to dynamically determine spectrum use characteristics for their own transmissions.

In this paper, we devise new attacks exploiting the unprotected data-plane signaling in cellular IoT networks (aka both NB-IoT and Cat-M). We show that, despite the deployed security mechanisms on both control-plane signaling and data-plane packet forwarding, novel data-plane signaling attacks are still feasible. Such attacks exhibit a variety of attack forms beyond simplistic packet-blasting, denial-of-service (DoS) threats, including location privacy breach, packet delivery loop, prolonged data delivery, throughput limiting, radio resource draining, and connection reset. Our testbed evaluation and operational network validation have confirmed the viability. We further propose a new defense solution within the 3GPP C-IoT standard framework.

Practical experimentation and prototyping are core steps in the development of any wireless technology. Often times, however, this crucial step is confined to small laboratory setups that do not capture the scale of commercial deployments and do not ensure result reproducibility and replicability, or it is skipped altogether for lack of suitable hardware and testing facilities. Recent years have seen the development of publicly-available testing platforms for wireless experimentation at scale. Examples include the testbeds of the PAWR program and Colosseum, the world's largest wireless network emulator. With its 256 software-defined radios, 24 racks of powerful compute servers and first-of-its-kind channel emulator, Colosseum allows users to prototype wireless solutions at scale, and guarantees reproducibility and replicability of results. This tutorial provides an overview of the Colosseum platform. We describe the architecture and components of the testbed as a whole, and we then showcase how to run practical experiments in diverse scenarios with heterogeneous wireless technologies (e.g., Wi-Fi and cellular). We also emphasize how Colosseum experiments can be ported to different testing platforms, facilitating full-cycle experimental wireless research: design, experiments and tests at scale in a fully controlled and observable environment and testing in the field. The tutorial concludes with considerations on the flexible future of Colosseum, focusing on its planned extension to emulate larger scenarios and channels at higher frequency bands (mmWave).

Video conferencing applications have seen explosive growth both in the number of available applications and their use. However, there have been few studies on the detailed analysis of video conferencing applications with respect to network dynamics, yet understanding these dynamics is essential for network design and improving these applications. In this paper, we carry out an in-depth measurement and modeling study on the rate control algorithms used in six popular commercial video conferencing applications. Based on macroscopic behaviors commonly observed across these applications in our extensive measurements, we construct a unified architecture to model the rate control mechanisms of individual applications. We then reconstruct each application's rate control by inferring key parameters that closely follow its rate control and quality adaptation behaviors. To our knowledge, this is the first work that reverse-engineers rate control algorithms of popular video conferencing applications, which are often unknown or hidden as they are proprietary software. We confirm our analysis and models using an end-to-end testbed that can capture the dynamics of each application under a variety of network conditions. We also show how we can use these models to gain insights into the particular behaviors of an application in two practical scenarios.

To protect an Industrial Control System (ICS), defenders need to identify potential attacks on the system and then design mechanisms to prevent them. Unfortunately, identifying potential attack conditions is a time-consuming and error-prone process. In this work, we propose and evaluate a set of tools to symbolically analyse the software of Programmable Logic Controllers (PLCs) guided by an information flow analysis that takes into account PLC network communication (compositions). Our tools systematically analyse malicious network packets that may force the PLC to send specific control commands to actuators. We evaluate our approach in a real-world system controlling the dosing of chemicals for water treatment. Our tools are able to find 75 attack tactics (56 were novel attacks), and we confirm that 96% of these tactics cause the intended effect in our testbed.

Vehicle-to-Vehicle (V2V) communication takes place among a wide range of wireless nodes such as cars, trucks, buses, motorcycles. However, V2V systems are vulnerable to Global Positioning Systems (GPS) spoofing attacks, where a malicious vehicle may spoof a victim vehicle's GPS coordinates to impersonate it. Thus, the malicious vehicle can act as a traffic injector imperiling traffic safety. In this work, we use simulation software as a platform to develop a practical testbed to execute an impersonation attack along with an in-vehicle defense technique.

With the booming deployment of smart homes, concerns about user privacy keep growing. Recent research has shown that encrypted wireless traffic of IoT devices can be exploited by packet-sniffing attacks to reveal users’ privacy-sensitive information (e.g., the time when residents leave their home and go to work), which may be used to launch further attacks (e.g., a break-in). To address the growing concerns, we propose SniffMislead, a non-intrusive (i.e., without modifying IoT devices, hubs, or platforms) privacy-protecting approach, based on packet injection, against wireless packet sniffers. Instead of randomly injecting packets, which is ineffective against a smarter attacker, SniffMislead proposes the notion of phantom users, “people” who do not exist in the physical world. From an attacker’s perspective, however, they are perceived as real users. SniffMislead places multiple phantom users in a smart home, which can effectively prevent an attacker from inferring useful information. We design a top-down approach to synthesize phantom users’ behaviors, construct the sequence of decoy device events and commands, and then inject corresponding packets into the home. We show how SniffMislead ensures logical integrity and contextual consistency of injected packets, as well as how it makes a phantom user indistinguishable from a real user. Our evaluation results from a smart home testbed demonstrate that SniffMislead significantly reduces an attacker’s privacy-inferring capabilities, bringing the accuracy from 94.8% down to 3.5%.

Due to the dynamic nature of changes in various ICT technologies nowadays, the gaps between industry, research, and academia need to be bridged in order to adequately support STEM students towards their future career paths. With the COVID-19 pandemic, and restrictions on access to university premises, an agile transition of both teaching and experimentation was essential, and adjustments in the curriculum were needed more than ever. Therefore, in this paper we present an adaptive and on-demand education framework for engineering students, thereby enabling remote experimentation and adjustments of exercise content to enhance students' learning experience. We present the two types of practical experimentation environments, i.e., cloud and real-life net-working testbed, for performing remote laboratory exercises, as well as the assessment of students' experience that is used as an input for the dynamic adjustments of the exercise content. Our results show that students consider they significantly improved the baseline skills our courses tend to build and strengthen towards preparing students for their future jobs.

In this contribution, an air-based macroscopic 2 x 2 multiple-input multiple-output (MIMO) molecular communication testbed using fluorescent dyes is presented and the application of MIMO techniques is investigated. It is demonstrated that the data rate can be doubled using spatial multiplexing. Using repetition MIMO coding and Alamouti-type coding, a spatial diversity/array gain can be achieved, resulting in error-free transmission over the entire transmission distance of the testbed.

In the field of molecular communication, molecules or other particles in the nanoscale are used to transmit information. This rather new communication paradigm has a large application potential, ranging from medicine to industrial systems. As testbeds are essential for evaluating the capabilities of this new form of communication, we improve an existing water-based testbed to achieve parallel data transmission within one communication channel in this work. To this end, two different information carriers, superparamagnetic iron oxide nanoparticles and fluorescent particles, are used to simultaneously transmit independent data streams. A detection system for fluorescent particles was specifically developed for the fluid molecular communication testbed. Using parallel data transmission in the modified testbed the channel capacity was nearly doubled in comparison to a transmission using only a single type of information particles.

The network slicing concept of 5G aims to provide the flexibility and scalability required to support a wide array of vertical services. To coordinate the coexistence of network slices, and to guarantee that the required resources are available for each one of them, the 5G core employs a slicing management entity, a slice manager. In this paper, we propose an architecture where the network slicing concept is extended beyond the core and access networks to also include the configuration of the UE's network stack. We exploit the slice manager's global view on the network to feed fine-grained information on slice configuration, health, and status to the UE. This information, together with local policies on the UE, is then used to dynamically create services tailored to the requirements of individual applications. We implement this architecture in a 5G testbed, and show how it can be leveraged in order to enable optimized services through dynamic network protocol configuration, application-to-slice mapping, and network protocol selection.

Change management has been a long-standing challenge for network operations. The large scale and diversity of networks, their complex dependencies, and continuous evolution through technology and software updates combined with the risk of service impact create tremendous challenges to effectively manage changes. In this paper, we use data from a large service provider and experiences of their operations teams to highlight the need for quick and easy adaptation of change management capabilities and keep up with the continuous network changes. We propose a new framework CORNET (COmposition fRamework for chaNge managEmenT) with key ideas of modularization of changes into building blocks, flexible composition into change workflows, change plan optimization, change impact verification, and automated translation of high-level change management intent into low-level implementations and mathematical models. We demonstrate the effectiveness of CORNET using real-world data collected from 4G and 5G cellular networks and virtualized services such as VPN and SDWAN running in the cloud as well as experiments conducted on a testbed of virtualized network functions. We also share our operational experiences and lessons learned from successfully using CORNET within a large service provider network over the last three years.

For the widely deployed ECN-based congestion control schemes, the marking threshold is the key to deliver high bandwidth and low latency. However, due to traffic dynamics in the high-speed production networks, it is difficult to maintain persistent performance by using the static ECN setting. To meet the operational challenge, in this paper we report the design and implementation of an automatic run-time optimization scheme, ACC, which leverages the multi-agent reinforcement learning technique to dynamically adjust the marking threshold at each switch. The proposed approach works in a distributed fashion and combines offline and online training to adapt to dynamic traffic patterns. It can be easily deployed based on the common features supported by major commodity switching chips. Both testbed experiments and large-scale simulations have shown that ACC achieves low flow completion time (FCT) for both mice flows and elephant flows at line-rate. Under heterogeneous production environments with 300 machines, compared with the well-tuned static ECN settings, ACC achieves up to 20\% improvement on IOPS and 30\% lower FCT for storage service. ACC has been applied in high-speed datacenter networks and significantly simplifies the network operations.

Modern cloud-scale cold storage data centers have begun to support right-provisioning of a rack’s resources (power, cooling, etc.), which allows only a small fraction of all hard disks to be active (spinning) concurrently at any given time to reduce the cost of ownership. Data deduplication is a traditional approach to split files into chunks and eliminate duplicate chunks, which can also cut costs for cold storage systems. However, when combined with right-provisioning, classical deduplication may make a file deduplicated and stored across the disks some of which are not active currently, thus leading to unacceptable access performance caused by spinning up and down of the disks. In this paper, we analyze the deduplication ratio under real-world workloads of cloud cold storage and observe for most workloads: 1) the deduplication ratio generally increases quickly with the first few of versions of the workload, and 2) increases slowly but steadily with the subsequent versions as a long tail. Based on the first observation, we propose an online deduplication way that can improve the deduplication ratio while providing acceptable read performance; based on the second one, we propose an additional offline deduplication way that can achieve comparable deduplication ratios with classical deduplication. We design a cold storage system called DeCold via combining the above two deduplication ways as well as improving deduplication efficiency. We prototype DeCold and conduct testbed experiments on real-world datasets including source code, virtual machine and database. Evaluations show that DeCold achieves better file access performance over the classical deduplication implementation, while maintaining decent deduplication efficiency.

Cyber testbeds provide an important mechanism for experimentally evaluating cyber security performance. However, as an experimental discipline, reproducible cyber experimentation is essential to assure valid, unbiased results. Even minor differences in setup, configuration, and testbed components can have an impact on the experiments, and thus, reproducibility of results. This paper documents a case study in reproducing an earlier emulation study, with the reproduced emulation experiment conducted by a different research group on a different testbed. We describe lessons learned as a result of this process, both in terms of the reproducibility of the original study and in terms of the different testbed technologies used by both groups. This paper also addresses the question of how to compare results between two groups’ experiments, identifying candidate metrics for comparison and quantifying the results in this reproduction study.

Congestion detection is the cornerstone of end-to-end congestion control. Through in-depth observations and understandings, we reveal that existing congestion detection mechanisms in mainstream lossless networks (i.e., Converged Enhanced Ethernet and InfiniBand) are improper, due to failing to cognize the interaction between hop-by-hop flow controls and congestion detection behaviors in switches. We define ternary states of switch ports and present Ternary Congestion Detection (TCD) for mainstream lossless networks. Testbed and extensive simulations demonstrate that TCD can detect congestion ports accurately and identify flows contributing to congestion as well as flows only affected by hop-by-hop flow controls. Meanwhile, we shed light on how to incorporate TCD with rate control. Case studies show that existing congestion control algorithms can achieve 3.3x and 2.0x better median and 99th-percentile FCT slowdown by combining with TCD.

This paper describe our team’s experience using minimega, a network emulation system using node and network virtualization, to support evaluation of a set of networked and distributed systems for topology discovery, traffic classification and engineering in the DARPA Searchlight program [18]. We present the methodology we developed to encode network and traffic definitions into an experiment description model, and how our tools compile this model onto the underlying minimega API. We then present three cases studies which demonstrate the ability of our EDM to support experiments with diverse network topologies, diverse traffic mixes, and networks with specialized layer-2 connectivity requirements. We conclude with the overall takeaways from using minimega to support our evaluation process.

Security datasets with various operating characteristics and abnormal situations of industrial control system (ICS) are essential to develop artificial intelligence (AI)-based control system security technology. In this study, we built a hardware-in-the-loop (HIL)-based augmented ICS (HAI) testbed and developed ICS security datasets. Here, we introduce the second dataset (HAI 21.03), which was developed with the user feedback of the first released version (HAI 20.07). All HAI datasets are publicly available at https://github.com/icsdataset/hai. HAI 21.03 was expanded by adding data points and normal/attack scenarios to HAI 20.07. We also held an AI-based anomaly detection contest (HAICon 2020) utilizing the HAI datasets developed so far, giving many AI researchers an opportunity to discuss and share ideas for ICS anomaly detection research. This paper presents the results of the HAICon 2020. The results of the top teams in the competition can be used as a performance comparison criterion when using HAI 21.03.

Fiber cut events reduce the capacity of wide-area networks (WANs) by several Tbps. In this paper, we revive the lost capacity by reconfiguring the wavelengths from cut fibers into healthy fibers. We highlight two challenges that made prior solutions impractical and propose a system called Arrow to address them. First, our measurements show that contrary to common belief, in most cases, the lost capacity is only partially restorable. This poses a cross-layer challenge from the Traffic Engineering (TE) perspective that has not been considered before: “Which IP links should be restored and by how much to best match the TE objective?” To address this challenge, Arrow's restoration-aware TE system takes a set of partial restoration candidates (that we call LotteryTickets) as input and proactively finds the best restoration plan. Second, prior work has not considered the reconfiguration latency of amplifiers. However, in practical settings, amplifiers add tens of minutes of reconfiguration delay. To enable fast and practical restoration, Arrow leverages optical noise loading and bypasses amplifier reconfiguration altogether. We evaluate Arrow using large-scale simulations and a testbed. Our testbed demonstrates Arrow's end-to-end restoration latency is eight seconds. Our large-scale simulations compare Arrow to the state-of-the-art TE schemes and show it can support 2.0x--2.4x more demand without compromising 99.99% availability.

Millimeter-wave communication with high throughput and high reliability is poised to be a gamechanger for V2X and VR applications. However, mmWave links are notorious for low reliability since they suffer from frequent outages due to blockage and user mobility. We build mmReliable, a reliable mmWave system that implements multi-beamforming and user tracking to handle environmental vulnerabilities. It creates constructive multi-beam patterns and optimizes their angle, phase, and amplitude to maximize the signal strength at the receiver. Multi-beam links are reliable since they are resilient to occasional blockages of few constituent beams compared to a single-beam system. We implement mmReliable on a 28 GHz testbed with 400 MHz bandwidth, and a 64 element phased array supporting 5G NR waveforms. Rigorous indoor and outdoor experiments demonstrate that mmReliable achieves close to 100\% reliability providing 2.3x improvement in the throughput-reliability product than single-beam systems.

When playing inter-personal sports games remotely, the time lag between user actions and feedback decreases the user’s performance and sense of agency. While computational assistance can improve performance, naive intervention independent of the context also compromises the user’s sense of agency. We propose a context-aware assistance method that retrieves both user performance and sense of agency, and we demonstrate the method using air hockey (a two-dimensional physical game) as a testbed. Our system includes a 2D plotter-like machine that controls the striker on half of the table surface, and a web application interface that enables manipulation of the striker from a remote location. Using our system, a remote player can play against a physical opponent from anywhere through a web browser. We designed the striker control assistance based on the context by computationally predicting the puck’s trajectory using a real-time captured video image. With this assistance, the remote player exhibits an improved performance without compromising their sense of agency, and both players can experience the excitement of the game.

The Long Range (LoRa) protocol for low-power wide-area networks (LPWANs) is a strong candidate to enable the massive roll-out of the Internet of Things (IoT) because of its low cost, impressive sensitivity (-137dBm), and massive scalability potential. As tens of thousands of tiny LoRa devices are deployed over large geographic areas, a key component to the success of LoRa will be the development of reliable and robust authentication mechanisms. To this end, Radio Frequency Fingerprinting (RFFP) through deep learning (DL) has been heralded as an effective zero-power supplement or alternative to energy-hungry cryptography. Existing work on LoRa RFFP has mostly focused on small-scale testbeds and low-dimensional learning techniques; however, many challenges remain. Key among them are authentication techniques robust to a wide variety of channel variations over time and supporting a vast population of devices. In this work, we advance the state of the art by presenting (i) the first massive experimental evaluation of DL RFFP and (ii) new data augmentation techniques for LoRa designed to counter the degradation introduced by the wireless channel. Specifically, we collected and publicly shared more than 1TB of waveform data from 100 bit-similar devices (with identical manufacturing processes) over different deployment scenarios (outdoor vs. indoor) and spanning several days. We train and test diverse DL models (convolutional and recurrent neural networks) using either preamble or payload data slices. We compare three different representations of the received signal: (i) IQ, (ii) amplitude-phase, and (iii) spectrogram. Finally, we propose a novel data augmentation technique called DeepLoRa to enhance the LoRa RFFP performance. Results show that (i) training the CNN models with IQ representation is not always the best combo in fingerprinting LoRa radios; training CNNs and RNN-LSTMs with amplitude-phase and spectrogram representations may increase the fingerprinting performance in small and medium-scale testbeds; (ii) using only payload data in the fingerprinting process outperforms preamble only data, and (iii) DeepLoRa data augmentation technique improves the classification accuracy from 19% to 36% in the RFFP challenging case of training on data collected on a different day than the testing data. Moreover, DeepLoRa raises the accuracy from 82% to 91% when training and testing 100 devices with data collected on the same day.

Ookami is a computer technology testbed supported by the United States National Science Foundation. It provides researchers with access to the A64FX processor developed by Fujitsu in collaboration with RIK{\Xi}N for the Japanese path to exascale computing, as deployed in Fugaku, the fastest computer in the world. By focusing on crucial architectural details, the ARM-based, multi-core, 512-bit SIMD-vector processor with ultrahigh-bandwidth memory promises to retain familiar and successful programming models while achieving very high performance for a wide range of applications. We review relevant technology and system details, and the main body of the paper focuses on initial experiences with the hardware and software ecosystem for micro-benchmarks, mini-apps, and full applications, and starts to answer questions about where such technologies fit into the NSF ecosystem., Comment: 14 pages, 7 figures, PEARC '21: Practice and Experience in Advanced Research Computing, July 18--22, 2021, Boston, MA, USA

Real-time Distributed Media Processing Workflows (DMPW) are popular for online media delivery. Combining distributed media sources and processing can reduce storage costs and increase flexibility. However, high request rates may result in unacceptable latency or even failures in incorrect configurations. Thus, testing DMPW deployments at scale is key, particularly for real-time cases. We propose the new MPEG Network Based Media Processing (NBMP) standard for this and present a testbed implementation that includes all the reference components. In addition, the testbed includes a set of configurable functions for load generation, monitoring, data-collection and visualization. The testbed is used to test Dynamic Adaptive HTTP streaming functions under different workloads in a standardized and reproducible manner. A total of 327 tests with different loads and Real-Time DMPW configurations were completed. The results provide insights in the performance, reliability and time-consistency of each configuration. Based on these tests, we selected the preferred cloud instance type, considering hypervisor options and different function implementation configurations. Further, we analyzed different processing tasks and options for distributed deployments on edge and centralized clouds. Last, a classifier was developed to detect if failures happen under a certain workload. Results also show that, normalized inter-experiment standard deviation of the metric means can be an indicator for unstable or incorrect configurations.