Industrial applications with closed-loop control demand ultra-reliable, low-latency communication in fixed and mobile networks. Integrating Time-Sensitive Networking (TSN) with 5G offers a promising solution. However, coordinating prioritization across two different networks is challenging. This paper demonstrates an integrated 5G–TSN solution for end-to-end traffic prioritization. To demonstrate closed-loop stabilization, we employ inverted pendulums, stabilized by controllers over 5G networks. Our solution ensures prioritization by TSN in the wired domain and by 5G Quality-of-Service (QoS) mechanisms in the wireless domain.

Extended Reality (XR) applications are emerging as a transformative technology, significantly enhancing user experiences in various domains, such as marketing and in-store shopping. In this context, object detection is a crucial component that enables XR-based gadgets to identify and interact with products in real-time. Object detection tasks typically employ AI-based models like YOLO (a.k.a. You Only Look Once), which require substantial computing power and currently are often offloaded to cloud servers. However, with advancements in hardware and Software-Defined Networking (SDN) and Network Functions Virtualization (NFV) technologies, In-Network Computing (INC) has emerged as a viable alternative. This paper proposes the application of INC for XR applications, specifically for in-store shopping use case, by leveraging the capabilities of programmable network devices to distribute computational tasks across the network. By integrating and enabling INC, we aim to enhance the efficiency and scalability of the evaluated XR applications while reducing latency, bandwidth consumption and reliance on cloud servers.

The European Union is implementing measures to reduce automotive accidents. In response, the automatic alerting eCall system is a key initiative to respond more quickly and reduce fatalities. However, the system faces significant challenges in areas with limited terrestrial network coverage. Non-Terrestrial Networks (NTNs) present a promising solution by extending connectivity to remote and underserved regions. In this study, we adapted our NTN simulator, INTANS, to implement and evaluate the eCall system within a 6G NTN framework. Our research focuses on measuring and analyzing key performance indicators, such as end-to-end latency and packet loss rate, through detailed simulations. Initial results demonstrate that the latency observed in our simulations meets the requirements set forth in 3GPP Release 19, confirming the system's capability to deliver emergency communications. Furthermore, these findings underscore the necessity of a robust Inter-Satellite Link (ISL) and optimized satellite handovers to enhance the reliability of the eCall system. Future work will aim to refine these results by exploring advanced communication protocols like TCP, thereby optimizing re-transmission strategies and further improving system reliability.

Recent advances in networking technologies, such as in-network computing (INC), have demonstrated significant potential for mobile networks. Among these, Random Linear Network Coding (RLNC), a class of forward error correction codes, has proven especially promising for enhancing reliability. RLNC reduces latency by transmitting additional coded packets, making it well-suited for supporting Ultra-Reliable Low Latency Communications (URLLC), a key requirement in nextgeneration mobile networks. To fully harness the benefits of RLNC in mobile networks, it is essential to closely monitor its operations to design effective RLNC schemes. This need arises from key factors, including the dynamic nature of mobile environments and the continuous emergence of new applications. However, achieving this in traditional mobile networks remains challenging due to their closed architectures and vendor lockin. We propose HawkVision, a monitoring solution for RLNC operations that leverages the flexibility of Open Radio Access Network (O-RAN). HawkVision supports monitoring various RLNC schemes, such as Sliding Window and Systematic Block Code, within mobile networks. It uses xApps in O-RAN to observe RLNC behavior in the RAN. We implement HawkVision using FlexRIC, a widely used O-RAN platform, and deploy a Key Performance Metric (KPM) xApp to collect relevant metrics. Testbed results demonstrate HawkVision’s effectiveness in monitoring the operations for different RLNC schemes.

To meet the stringent requirements of emerging 5G use cases that demand high reliability and low latency, the potential of in-network computing is realized by deploying Random Linear Network Coding (RLNC) recoders directly within the network infrastructure. However, the choice of network infrastructure configuration can significantly impact traffic latency as demonstrated in this paper. To investigate this, we compare the performance of RLNC recoder when implemented on two different switches, the Cisco Catalyst 9500 (C9500) and the Cisco Catalyst 9300 (C9300), focusing on reducing one-way delay (OWD). Our results demonstrate that the C9300 significantly outperforms the C9500, with$\mathbf{9 5. 6 \%}$of haptic packets recovered within 70 ms compared to 85.8% for the C9500. Notably, the C9300 consistently minimizes packet loss better, especially in haptic traffic, where the C9300 with RLNC recoder reduces packet loss to nearly zero. Additionally, under stress conditions with a 10 Mbps transmission rate, the C9300 continues to excel in reducing OWD, although the in-network computing performance in both switches exhibits similar reliability challenges. This study focuses on analysing the performance of RLNC recoding in cloudnative$\mathbf{5 G}$systems under various hardware conditions.

Traditional RAN handover strategies often trigger excessive handovers in V2X systems, leading to unstable links and increased communication overhead. Open and softwarized RAN introduces centralized, intelligent control through custom xApps running on the RAN controller, enabling real-time handover and beam management. However, its application to beamforming-based handovers in multi-vehicle scenarios remains underexplored. In this work, we propose three strategies to minimize handover frequency in both analog and hybrid beamforming systems, with and without relay support. We also outline a simulation framework based on NS-3 and SUMO, the traffic and mobility simulator, to evaluate the effectiveness of our proposed solutions.

Wireless Mesh Networks can be a valuable enhancement to network infrastructures in many contexts. Mesh nodes can extend the range of WiFi networks by acting as relays between source and destination nodes, which can be crucial for contexts where cabling is something to be avoided. Due to unforeseen degradations in the environment, however, certain links within the mesh may fail, leading to a lack of end-to-end connectivity, at which point the nodes must form new routes based on the employed protocol. Better Approach To Mobile Ad-Hoc Networking (BATMAN) is a very commonly employed mesh routing protocol that handles routing within mesh networks. This paper evaluates the end-to-end reconnection establishment performance of the protocol utilizing a hardware testbed, and discusses potential enhancements to improve reconnection latency. We also open source the software developed to automate our operations and measurements done on the testbed, for public reference and usage.

The convergence of 6G and WiFi technologies promises unprecedented online connectivity, enabling immersive experiences such as the Metaverse, with a particular emphasis on VR gaming. Despite these advancements, they bring to light considerable privacy concerns, especially the ability of adversaries to deduce personal information from encrypted gaming network traffic. Focusing on VR headsets and Nintendo Switch consoles, this study explores the privacy implications within such network environments. By simulating the typical network conditions of online multiplayer games, we expose potential privacy breaches by adversaries from both WiFi and WAN, including mobile service providers. Classical machine learning algorithms can successfully classify games and gaming consoles with an accuracy exceeding 90%, further dissecting the network traffic to unveil distinct signatures and assess privacy risks.

Feature-based Visual Simultaneous Localization and Mapping (V-SLAM) employs feature descriptors to recognize landmarks in successive frames. For networked (offloaded and/or collaborative) V-SLAM, the feature descriptors typically need to be transmitted over wireless communication networks with limited bitrates. Conventionally, V-SLAM systems have employed general computer vision feature descriptors whose size (in bytes) accounts for most of the data transmission. We develop a probabilistic model of the V-SLAM feature matching and identify three parameters that govern the probability of the correct feature match: the feature set size, the bit error probability between feature descriptors observing the same landmark, and the probability that the correct feature match exists in the feature set. Based on these three parameters, we formulate a research agenda for achieving high V-SLAM performance (i.e., high correct feature match probability) for reduced V-SLAM feature descriptor sizes (and thus reduced network throughput requirements). We address the first item on this research agenda by pursuing the reduction of the feature set size through a novel Orientation Restriction with Frame-to-Frame Projection (ORFFP). We evaluate ORFFP in comparison to the state-of-the-art ORB-SLAM2 approach using three representative V-SLAM test sequences. Our ORFFP approach reduces the size of the feature set to as little as 9.1% of ORB-SLAM2, allowing for the use of smaller feature descriptors. Thereby, ORFFP reduces the overall required throughput for networked V-SLAM down to roughly a third of the current approach, while achieving essentially the same V-SLAM performance. We also find that for learned descriptor-based models (HashSIFT, BAD), the feature descriptor size can be reduced using ORFFP without severely degrading the performance of V-SLAM. We outline future research directions towards comprehensively addressing the formulated research agenda for enabling networked V-SLAM on low-bitrate networks.

Emerging 5G/6G use cases span various industries, necessitating flexible solutions that leverage emerging technologies to meet diverse and stringent application requirements under changing network conditions. The standard 5G RAN packet error handling using retransmission reduces packet loss but can increase transmission delay. Random Linear Network Coding (RLNC) offers an alternative by proactively sending combinations of original packets, thus reducing both delay and packet loss. Previous research typically only simulates the integration of RLNC in 5G but does not demonstrate nor evaluate this integration in real 5G systems. In contrast, we implement and evaluate our approach through measurements with commercially available servers and switches, running the OpenAirInterface (OAI) 5G software stack. We introduce Flexible Network Coding (FlexNC), which enables the flexible fusion of several RLNC protocols. Specifically, FlexNC provides a forwarder that flexibly interfaces with multiple RLNC protocols in a cloud-native (containerized) solution. Network operators can configure FlexNC based on network conditions and application requirements. For boosting network programmability, our Recoder in the Network (RecNet) leverages In-Network Computing (INC) in intermediate network nodes. We have developed an open-source cloud-native (Docker-based) implementation of both FlexNC and RecNet on OAI, including INC for the Recoder on a Cisco Catalyst switch. Measurements for video, haptic, and audio traffic indicate that i.) FlexNC adapts to various application needs in terms of latency and packet loss, and RecNet significantly reduces packet loss for a remote user with minimal increase in delay compared to pure RLNC. To the best of our knowledge, this is the first article to report testbed measurement results for cloud-native network coding in a 5G system, thus creating a baseline for reliable 5G communication.

Human-robot interaction is crucial in various industries and domains, such as manufacturing, healthcare, and entertainment. Natural and intuitive interactions between humans and robots are crucial. Legacy controllers were designed for two-dimensional visual display and, therefore, suboptimal for interaction in three-dimensional space. Hand gesture recognition with camera-based systems is often hindered by visual obstruction. We demonstrate a hand gesture system leveraging data gloves with inertial measurement units (IMU). This demonstration focuses on gesture recognition quality, enhancing robustness and responsiveness. Audiences can observe and directly participate using the data glove to maneuver a robotic dog remotely in real-time.

Visual Collaborative Simultaneous Localization and Mapping (SLAM) uses multiple mobile devices to capture images of their surroundings and map their environment collectively while simultaneously determining their position. In centralized approaches, a single server merges visual features from these devices to build a shared map. Offloading computationally intensive tasks benefits resource-limited mobile devices. However, due to unpredictable and heterogeneous communication links, maps at the server and on the devices can become asynchronous, potentially leading to real-time localization failures. To address these issues, we present Reliable Agents in Centralized Visual Collaborative SLAM (RAVIC), which employs an optimised server-to-agent communication strategy. This includes a novel keyframe selection method that exploits the server’s global map to improve tracking under high agent mobility and network latency. In addition, adjustments to agent-to-server communication and agent tracking algorithm reduce the computational demands on mobile devices, leading to a reduction in average tracking processing time of up to 27.2 ms. Our evaluation using benchmark datasets and different network conditions shows a drastic reduction in track loss probability from 61.4% to only 1.0% utilising merged map data in collaborative scenarios, demonstrating that collaboration increases the reliability of individual agents.

Edge clouds embrace Network Function Virtualization (NFV) to bring network functions (NFs) closer to end devices. The mobility nature of edge clouds and the emergence of new use cases, such as robot control or metaverse, demand NFs with high resilience. However, numerous NFs like firewalls require stateful processing that depends on historical processing values, referred to as NF states. Consequently, supporting stateful NFs with high resilience necessitates the maintenance of their states, often resulting in significant modification of the NFs. While providing high resilience for NFs, maintaining their processing performance is a critical challenge for existing solutions. Our proposed framework, StateProc, eliminates this concern by enhancing the processing capabilities of NFs through the utilization of NF states for custom processing. Through custom processing for state transfer, the evaluation results show that StateProc significantly improves the state transfer time, up to 93% faster than a notable state-of-the-art framework, without additional processing overhead.

Teleoperation has become mainstream in several applications, such as telesurgery, manufacturing, and construction. The human operator relies on video, audio, and haptic feedback data from the remote site to adjust operations, ensuring safety and collision avoidance. These applications require real-time perception and synchronization of modalities to maintain a high quality of experience, which presents a significant challenge. Human sensitivities are different among modalities, while communication networks rely on statistical multiplexing to maximize bandwidth utilization, often neglecting the specific latency requirements of each modality. We introduce SynCoDel to minimize asynchrony between video and haptic streams by considering event pairs, which are synchronized events. By prioritizing video packets within these pairs, we effectively utilize bandwidth resources initially reserved for high-priority haptic streams. Evaluation on a practical testbed with the programmable data plane and synthetic data shows that SynCoDel reduced asynchrony over half while maintaining low latency for haptic and video streams in congestion scenarios.

Using virtualized network management functions, the Service Based Architecture will replace the Reference Point Architecture for managing future mobile networks. It naturally lacks a clear security perimeter and has an increased attack surface, so defending the control plane against attacks requires a novel protection paradigm. Both National Institute of Standards and Technology and 3rd Generation Partnership Project suggest moving from perimeter security to a Zero Trust Architecture (ZTA), authenticating all request initiators and controlling access to all resources for each request. However, it insofar remains somewhat unclear to which extent the suggested management protocols do indeed meet the ZTA. We are exploring the standardized communication management protocols in this paper. Our analysis indicates that with careful implementation, the existing network functions and protocols can indeed achieve comprehensive authentication and access control so that the ZTA can be met.

Emerging use cases, such as Tactile Internet, typically demand high reliability and low-latency communication. To fulfill these stringent requirements, 5G networks employ re-transmission within the Radio Access Network (RAN) in the event of packet loss, but at the expense of increased latency. An alternative approach to address this challenge involves leveraging Random Linear Network Coding (RLNC) to recover lost packets, thereby eliminating the necessity for retransmission within the RAN. We demonstrate the ability of in-network computing to run RLNC, particularly with the use of RLNC recoder, to enhance reliability and reduce latency within the network.

Network Functions (NFs) in edge clouds are required to provide scalability, fault tolerance, and mobility support. They all require maintaining NF states (i.e., processing results), e.g., for recovery, especially for stateful NFs like firewalls. Even though current solutions provide an alternative to storing states in memory, their design can support only a single requirement, either fault tolerance or scaling. Advocating the versatility, we propose StateOS - an operating system of NF states for user-defined programs supporting different requirements. Additionally, we propose a state transfer scheme, namely Divide-and-Conquer (DAC), to accelerate StateOS. The combination of DAC and StateOS demonstrates its efficiency for all three scenarios: scaling, fault tolerance, and service function chain acceleration.

As part of the most recent mobile broadband standards, 5G and 6G, network slicing solutions will shape the future of networks. By employing network slicing, network operators can protect participants against different attacks over the network by isolating devices and providing resource guarantees. Currently, to evaluate new network-slicing solutions, network researchers or engineers have to either evaluate the solution in a virtual environment that can not accurately scale to the performance of hardware platforms or invest significant portions of time in network configuration on hardware platforms. To quickly prototype network slicing solutions and other solutions depending on software-defined networking (SDN) features, we introduce a hardware abstraction layer featuring a vendor-independent API to manage hardware platforms and a new testbed solution featuring topologies similar to common network evaluation platforms like mininet.In this demo, we showcase a network-slicing solution leveraging SDN. The audience can visually observe the performance of the communication channel before and after denial-of-service attacks and with and without the secure network slicing.

Emerging network application paradigms, such as the Tactile Internet, re-emphasize the need for different Quality of Service (QoS) levels. Due to the large packet buffers in the underlying network data plane, Active Queue Management (AQM) is generally required to curtail packet latencies for flows requiring high QoS levels. At the same time, programmable data planes, such as P4, enable packet processing at line-speed, albeit with limited packet processing functionalities. However, the existing AQM mechanisms that support QoS differentiation are too complex to readily run on P4, while the existing AQM mechanisms that run on P4 do generally not support effective QoS differentiation. We address this gap by developing to the best of our knowledge the first AQM mechanism that supports effective QoS differentiation while running on P4. Specifically, we propose SP-CoDel, which combines the well-known Controlled Delay (CoDel) AQM mechanism with Static Priority (SP) scheduling for QoS differentiation. Also, we propose RED-SP-CoDel, which adds a RED AQM component to SP-CoDel so as to make the AQM with priorities essentially parameterless. As a community resource contribution, we substantially extend the existing P4Simulator to the novel fused P4-NS3 Simulator so as to enable the evaluation of packet processing mechanisms through the combined functionalities of P4 device emulation and NS3 packet simulation. Evaluations conducted with the P4 reference switch model in the P4-NS3 Simulator indicate that SP-CoDel and RED-SP-CoDel provide high QoS to high-priority data streams, i.e., significantly reduce latency and packet loss compared to CoDel, while effectively mitigating bufferbloat.

In order to provide consistent low-latency communication network services, Time-Sensitive Networking (TSN) unites a set of standards for time-synchronization, flow control, enhanced reliability, and management. We design the TSN-FlexTest testbed with generic commodity hardware and open-source software components to enable flexible TSN measurements. We have conducted extensive measurements to validate the TSN-FlexTest testbed and to examine TSN characteristics. The measurements provide insights into the effects of TSN configurations, such as increasing the number of synchronization messages for the Precision Time Protocol, indicating that a measurement precision of 30 ns can be achieved. The TSN measurements included extensive evaluations of the Time-Aware Shaper (TAS) for sets of Tactile Internet (TI) packet traffic streams. The measurements elucidate the effects of different scheduling and shaping approaches, while revealing the need for pervasive network control that synchronizes the sending nodes with the network switches. We present the first measurements of distributed TAS with synchronized senders on a commodity hardware testbed, demonstrating the same Quality-of-Service as with dedicated wires for high-priority TI streams despite a 200% over-saturation cross traffic load. The testbed is provided as an open-source project to facilitate future TSN research.

Computing in the Network (COIN) has the potential to reduce the data traffic and thus the end-to-end latencies for data-rich services. Existing COIN studies have neglected the impact of the size of the data unit that the network nodes compute on. However, similar to the impact of the protocol data unit (packet) size in conventional store-and-forward packet-switching networks, the Computing Data Unit (CDU) size is an elementary parameter that strongly influences the COIN dynamics. We model the end-to-end service time consisting of the network transport delays (for data transmission and link propagation), the loading delays of the data into the computing units, and the computing delays in the network nodes. We derive the optimal CDU size that minimizes the end-to-end service time with gradient descent. We evaluate the impact of the CDU sizing on the amount of data transmitted over the network links and the end-to-end service time for computing the convolutional neural network (CNN) based Yoho and a Deep Neural Network (DNN) based Multi-Layer Perceptron (MLP). We distribute the Yoho and MLP neural modules over up to five network nodes. Our emulation evaluations indicate that COIN strongly reduces the amount of network traffic after the first few computing nodes. Also, the CDU size optimization has a strong impact on the end-to-end service time; whereby, CDU sizes that are too small or too large can double the service time. Our emulations validate that our gradient descent minimization correctly identifies the optimal CDU size.

Cellular-connected Unmanned Aerial Vehicles (UAVs) require reliable wireless connections and will benefit from ultra-reliable and low-latency communication (URLLC) in 5G and beyond networks. 5G campus solutions, when integrating with Time-Sensitive Networking (TSN), can provide extended coverage and deterministic behavior for wireless links. However, one of the challenges for research in this direction is reliable yet cost-effective and flexible evaluation, which applies to academia and industry. We believe that open-source testbeds can effectively address the challenge due to their maturity and transparency in verifying evaluation results. Subsequently, we systematically survey and categorize testbeds in the literature that use commercial-off-the-shelf hardware. Our study outlines options for building a full-fledged testbed for integrating TSN with 5G and beyond for research and development. This study initiates one of the first steps toward facilitating future research toward practical deployment of TSN-5G integration.

Compressed sensing aims to retrieve sparse signals from very few samples. It relies on dedicated reconstruction algorithms and well-chosen measurement matrices. In combination with network coding, which operates traditionally over finite fields, it leverages the benefits of both techniques. However, compressed sensing has been primarily investigated over the real field. F2OMP is one of the few recovery algorithms to reconstruct signals over finite fields. However, its use in practical cases is limited since its performance depends mainly on binary matrices for signal recovery. This paper reports results of extensive simulations enhancing the features of well-performing measurement matrices for F2OMP as well as methods to build them. Moreover, a modified version of the algorithm, F2OMP-loop, is proposed. It offers a compromise between performance, stability, and processing time. This allows to design a joint compressed sensing and network coding framework over finite fields.

Human activity recognition is crucial for tactile internet, virtual reality, and digital-twin applications. Previous works have analyzed machine learning for this purpose but often need to pay more attention to typical challenges when deploying these machine learning models in production. First, data must be transmitted to the network node on which the machine-learning model is running. However, scaling human activity recognition by the number of devices and users puts additional constraints on the available transmission channel. While transmitting less data saves bandwidth, removing redundant information from the data can also benefit machine learning. This paper addresses the problem of transmitting wearable sensor data for human activity recognition. To this end, we analyze the extent to which wearable sensor data can be compressed via sparse coding without sacrificing loss in recognition performance. We empirically illustrate, on various datasets, that only a fraction of sensor information is relevant for human activity recognition.

Acoustic data from the Industrial Internet of Things (IIoT) are widely used in anomaly detection because audio information reflects richer internal statuses of monitored working machines than the video does. Since multiple acoustic data sources interfere with each other by nature, source data estimation is a prerequisite of subsequent anomaly detection. Existing schemes often use a centralized manner to separate full data on a remote node in clouds. However, such a centralized manner may delay reactions to anomalies due to data transmission delay and the complexity of solving data separation problems. This article shows that the data separation phase can be substantially accelerated with an in-network computing approach. The key idea is to offload data processing jobs to intermediate network nodes along the forwarding path. We first propose a distributed algorithm so that the data separation jobs can be done in a progressive manner; likewise, we modify the forwarding layer in order to eliminate hop-by-hop data transmission delay that hurts the performance of using in-network computing. We further derive theoretical upper and lower bounds of the required number of intermediate nodes that achieve the maximum acceleration. We also implement our proposed solution in a full-stack network emulator. Based on an open and professional data set, evaluation results justify the feasibility and advantages of our idea with nearly 32.18% acceleration on total processing time. This work exemplifies the convergence of IIoT, edge, and clouds.

Existing Network Function Virtualization (NFV) service placements that reuse already deployed network functions either reuse an entire Service Function Chain (SFC) or only individual network functions while ignoring the chain configuration cost for configuring the SFC traffic steering and ignoring the reliability of the network functions. Also, the Mobile Edge Cloud (MEC) frameworks that are required to implement an NFV service placement should ideally seamlessly cooperate with the various existing NFV Management and Orchestration (MANO) frameworks. However, the existing MEC frameworks lack multi-MANO support. We formulate the novel Subchain-Aware NFV service Placement (SAP) optimization model that accounts for the configuration cost for stitching together reused network functions to an SFC and strives to reuse existing subchains of consecutive network functions (with already deployed SFC traffic steering), while accounting for the recovery cost of network functions with limited reliability. We develop Tabu-SAP, a Tabu search approach to solve the SAP optimization problem. Furthermore, we introduce the novel Automated Provisioning framework for MEC (APMEC) with open-source OpenStack implementation to enable the deployment of Tabu-SAP in real networks; APMEC supports multiple MANOs through a loose coupling MANO-MEC design. Our Tabu-SAP evaluations indicate an around eightfold increase of the number of supported SFCs compared to the state-of-the-art reuse of individual network functions, while substantially reducing the total cost, which includes the chain configuration cost. Also, for long SFCs of seven or more network functions, the Tabu-SAP total cost is less than 10% higher than the optimal solution (which requires over ten times longer execution time).

Teaching heterogeneous robotic cells is difficult and time-consuming. We present a context-aware robot teaching solution based on an open-source framework that decouples the task instruction from the task execution to simplify the teaching workflow significantly. To demonstrate the advantages of the solution, a human operator uses a VR headset and controllers to teach a virtual robot to perform a task, e.g., pick-and-place in a virtual world. The task is automatically adapted to different settings of operation cells and executed by physical robots located at different geographical locations.

Simultaneous Localization and Mapping (SLAM), among other critical functions of mobile robots, such as navigation, are computationally expensive. When deployed at the robot, those functions demand high energy consumption and result in shorter operation time. Offloading SLAM to an Edge Cloud (EC) can significantly reduce the robot’s computing demand and resources, subsequently reducing energy consumption. We offload intelligence of mobile robot control functionality, i.e., navigation, localization, and control to an EC. The EC processes sensor data and sends the robot the directional velocities. Meanwhile, a 5G wireless connection ensures the necessary low latencies and high throughputs. We demonstrate the feasibility of offloading SLAM and navigation in an EC based on a use case in automotive production. Additionally, we developed a digital twin of the robot and visualized its current sensor data.

The future Tactile Internet with Human in the Loop (TaHiL) [1] enables a perceived real-time interaction between a human and a remote physical or virtual object. Human intention needs to be inferred from data captured by sensors throughout the body and even the brain. Therefore, the prediction can happen already in a machine attached to the human body (or so-called a Body Computing Hub, BCH). This human-machine coaugmentation requires Tactile electronics [2] with extreme requirements, such as ultra-small, stretchable, and ultra-low-energy consumption, allowing for sensing at extremely low latency. However, for a human to interact with a real or virtual object across the globe, tactile electronics require tight integration with softwarised networks [3] , wireline or wireless, with extremely low latency. Such networks can also bring the computing capability to human’s proximity, such as a network edge, by leveraging Network Function Virtualization (NFV) and Software-Defined Networking (SDN). Even though critical, the integration between tactile electronics in a body area network (BAN) and external softwarised networks is uncovered in the literature. This paper explores vital connections between tactile electronics and fully softwarised networks, focusing on adapting all layers from electronics to network and application.

This demonstration focuses on in-network computing as an enabler for low-latency Industrial Internet of Things (IIoT) applications, such as audio source separation for anomaly detection. By demonstrating a specific industrial application, we show that our method Progressive ICA (pICA), improves accuracy and reduces overall service latency progressively. The idea is to parallelize data transmission and processing along a multi-hop path consisting of in-network computing nodes. The audience can experience the benefits of the novel concept of in-network computing by interacting with the demonstration remotely via the Internet or in person.

Edge Simultaneous Localization and Mapping (SLAM) retains only the tracking on the mobile device, while offloading the compute-intensive local mapping and loop close to edge computing. Existing Edge SLAM approaches incur relatively high delays for offloading, resulting in high failure probabilities, i.e., low reliability, for commonly used public SLAM datasets. We discovered that two parameters which had not previously been studied in detail, namely the number of features and the number of keyframes that are bundled for a local map update, play a critical role in the offloading delay. Also, previous approaches updated the local map in the mobile device in a serial manner, incurring map update latencies. We study the numbers of features and bundled keyframes in detail and we parallelize the local map update. We find that judicious parameter settings, namely relatively small numbers of features (750 per frame) and bundled keyframes (1, i.e., effectively no bundling), reduce the map update latency to less than half compared to the previously common settings (1000 features per frame and 6 keyframes used for a map update). For a low network latency of 20ms, these judicious parameter settings in conjunction with our parallelized local map updating, reduce the 79% failure rate of the previous Edge SLAM systems down to 2%.

Mobile robots are an essential part of the vision of flexible production in a smart factory. To monitor and connect the robots, reliable and low latency communication is necessary. In this work, we conduct packet-based active measurements to evaluate the performance of a state-of-the-art 5G standalone system in a production environment. The focus is on whether 5G connections can meet the requirements specified by 3G PP in terms of delay and reliability. The results indicate that without cross-traffic, the requirement of a delay of less than 10 ms for 99.9 % of the packets can be met for the remote control and fleet management of mobile robots. However, as soon as cross-traffic is injected, especially in the uplink, the upper percentiles of the delay increase significantly, thus failing to hold the reliability requirements.

Deterministic communications are required for industrial environments, yet their realization is a challenging task. Time-Sensitive Networking (TSN) is intended to enable deterministic communication over inexpensive Ethernet networks. Standardized by the IEEE TSN working group, TSN enables precise control of time synchronization, traffic shaping, reliability enhancements, and network administration to answer the demands of industrial control applications. Subsequently, there is a significant need to enable turnkey research and implementation efforts. However, a current lack of open-sourced testbed implementations to investigate and study the behavior of TSN network devices limits verification to simulation and theoretical models. We introduce a publicly available, flexible, and open-sourced measurement testbed for evaluating TSN in the context of industrial automation applications to address the need to perform real-world measurements. In this contribution, we describe our testbed combining Commercial-Off-The-Shelf (COTS) hardware and existing open-source tools as a platform for in-depth evaluation of TSN devices. Providing detailed TSN backgrounds, we describe an in-depth performance analysis for our implementation. For a common Tactile Internet scenario, we observe an accuracy of close to 5 ns achievable with our publicly available COTS setup.

5G campus networks, whose advantages include flexible deployment, can be a promising candidate for production plants to complement existing Wifi-based networks. Toward that goal, 5G has to satisfy strict requirements about real-time communication to facilitate novel use cases. However, the realtime-capability of 5G is not well understood yet. In this work, we deliver insights into the functioning of 5G NR RAN Release 15, which includes actual one-way delay and Round-Trip Time (RTT) measurements for Downlink and Uplink in a private 5G Standalone campus network. The extensive measurement results reveal that these delays are correlated, and the corresponding RTT, i.e. the sum of Downlink and Uplink delays, is discreetly clustered, ranging between 12ms and 40ms. The measurements also show that the distribution of RTTs is mainly dependent on the packet rates and their inter-arrival times. Our study helps expand the current understanding of 5G used for latency-critical applications. We make the code and the measurement data traces publicly available as the IEEE DataPort 5G Campus Networks: Measurement Traces dataset (DOI 10.21227/xe3c-e968).

Delivering content from a network via a client-server architecture is expensive not only for content owners but also for network operators. Moving content closer to the end user is already used in Content Delivery Networks (CDN). Multi-Access Edge Computing (MEC) enables us to shift the content even closer by using the storage of end users. But, due to the large media files, storage and transport costs for peers increase significantly. Network Coding can reduce these costs. However, peers in CDNs tend to be highly fluctuating and often need to be restored, making continuous availability of data at the network edge a problem. While for uncoded data, individual packets lost due to peer failures can be tracked to determine availability, the availability of coded data is currently distinguished only in two cases: either there are still enough linearly independent packets to decode the file, or there are not. However, we have found that the network’s combined coded cache loses quality over time due to recovery. This quality loss, which we refer to as grade, can be measured by very cost-effective monitoring. If the grade falls below a certain limit, we can intervene in the network by performing a cache refresh to prevent data becoming unavailable preemptively. In this paper, we present the cases in which such monitoring is useful, how the grade is calculated, and when a cache refresh is necessary. The results show that we can reduce network traffic by up to 34% with minimal storage costs through efficient monitoring.

Live service migration with low service downtime is one of the main challenges in mobile edge cloud (MEC). In other words, it is important to guarantee that users continue to receive good service even if they cross through multiple MEC servers. In this paper, we propose a seamless service migration framework and apply it to an autonomous driving demonstration. We expect this paper to provide a perspective on how MEC can be used for practical autonomous driving in the near future.

Time-sensitive applications on wireless networks often demand high reliability and high throughput. Random Linear Network Coding (RLNC) is a forward error correction (FEC) method for wireless lossy channels that organizes the information in blocks, commonly called generations. Hybrid Automatic Repeat reQuest (HARQ) is a combination of a FEC method, such as RLNC, and an ARQ often used to ensure reliable data transmission. Many RLNC-based protocols provide in-order delivery, which helps in general cases but backfires in high-jitter channels. Multigeneration RLNC protocols are more resilient in high-jitter channels than conventional RLNC protocols because they are capable of handling multiple generations at the same time. However, they still struggle against losses. Despite the considerable amount of research done in this field, there is still no solution that combines HARQ with multigeneration RLNC protocols. Our contribution in this paper is two-fold: (i) we propose a new ARQ mechanism adapted for multigeneration RLNC protocols, and (ii) we compare its performance in terms of per-packet delay and throughput to conventional RLNC, HARQ RLNC, and multigeneration RLNC. The results show an increase in the decoding probability from 25% without ARQ to 80% with ARQ. Moreover, we observe a decrease of 35% in the average packet delay.

It is vital for Tactile Internet to constantly maintain a low-latency control loop between sensors, actuators, and their controlling software applications. Mobile Edge Computing (MEC) is an essential technology that brings the elasticity of cloud computing to run controlling applications at a close proximity to controlled objects, thus reducing latency. To support the mobility of the objects, the underlying network has to be capable of migrating MEC applications seamlessly to guarantee the close proximity. However, it is challenging to migrate application states quickly and flexibly without interrupting the control loop. We propose FAST, a flexible scheme for direct and low-latency state transfer leveraging Software-Defined Networking (SDN). Evaluation results show that compared to state of the art, FAST reduces the service migration time by 80%.

Fulcrum decoders can trade off between computational complexity and the number of received packets. This allows heterogeneous nodes to decode at different level of complexity in accordance with their computing power. Variations of Fulcrum codes, like dynamic sparsity and expansion packets (DSEP) have significantly reduced the encoders and decoders' complexity by using dynamic sparsity and expansion packets. However, limited effort had been done for recoders of Fulcrum codes and their variations, limiting their full potential when being deployed at multi-hop networks. In this paper, we investigate the drawback of the conventional Fulcrum recoding and introduce a novel recoding scheme for the family of Fulcrum codes by limiting the buffer size, and thus memory needs. Our evaluations indicate that DSEP recoding mechamism increases the recoding goodput by 50%, and reduces the decoding overhead by 60%-90% while maintaining high decoding goodput at receivers and small memory usage at recoders compared with the conventional Fulcrum recoding. This further reduces the resources needed for Fulcrum codes at the recoders.

Data transmission in wireless networks is vulnerable to errors, due to the nature of wireless characteristics. As a result, corrupted packets are a common case in wireless data transmissions. Many techniques have been proposed to tackle these issues and one of the recommended techniques is the partial packet recovery (PPR) scheme. There are various existing PPR methods that have been proposed in wireless networks over the past two decades. In addition, the recent works have shown the possibility of network coding along with its capabilities in recovering the partial packets. A review study of PPR approaches in wireless networks is presented in this study. The authors classify the approaches into several groups based on the soft information in the PHY layer. Furthermore, they describe the PPR techniques that have used in each of these groups. Their studies found that there are 24 protocols have been proposed using PPR to improve the performance in wireless networks.

5G ecosystems will benefit significantly from Multi-access Edge Computing (MEC) and Network Function Virtualization (NFV). While NFV allows for dynamical deployment of virtualized network functions, MEC allows applications to be deployed close to mobile users, thus reducing latency. When used together, NFV and MEC bring flexibility and enhanced performance to meet the user's demand. However, to increase service availability and maximize the convenience of MEC users, an MEC framework has to interface with multiple NFV orchestrators to utilize running network functions in an efficient manner. We introduce in this demonstration APMEC, a framework addressing the above two challenges. Via a combination of an interactive GUI, KPI views and a live-demo setup, we will showcase the advanced features of the framework.

This demonstrator, which is completely realized in software and hardware, shows the influence of latency in a very descriptive and interactive way with the example of a Mobile Edge Cloud or legacy cloud operated ball sorting machine. The demonstrator has every essential element that would be found in a real production machine or control loop. It consists of a sensor, a controller and several actuators. Through the Control and Monitoring Interface, the audience of the demo has a simple and self-explanatory way to interact with the demonstrator. In addition the advantages of Multi-access Edge Computing compared to legacy cloud computing and embedded computing are explained in the context of industry 4.0.

Virtualization of services in factory production allows achieving higher service reliability and smaller cost of industrial equipment. The demonstrator shows Mobile Edge Cloud (MEC) implementation for services such as path planning and movement control of a robot arm, collaboration between two robot arms, and remote control of a robot arm. The audience can participate in the demonstrator through guiding a robot arm remotely using a wireless controller. The MEC resources required for such virtualized services are shown in terms of traffic (throughput and inter-arrival time) and server load.

This demonstration shows both the utility of Mobile Edge Cloud (MEC) for 5G connected cars, as well as the impact of MEC server selection (i.e. migration) strategy on latency. The demonstration simulates cars moving according to a realistic model inside the city of Munich, and implements different MEC server selection strategies. A player can steer an ambulance using a controller, and can also choose one among four server selection strategies while driving. With each strategy, the player will experience a different latency.

Random Linear Network Coding (RLNC) is widely considered a key enabler for 5G networks. It compensates for lost and corrupted packets by sending redundant coding packets. Despite the considerable attention RLNC received from the research community, the impact of the number of redundant coded packets on the network performance is poorly understood. An excessive number of redundancies would pollute the network with useless packets that give no additional information, and an insufficient number of redundancies would not provide enough packet loss resilience. We introduce a novel formal model for predicting the number of losses a system would have when RLNC is applied, by optimizing simulation results with the Newton-Raphson method. The model allows the sender to set the minimum amount of RLNC redundancies needed to keep the packet loss under a certain threshold. We validate our model against simulations by calculating the MSE between simulation results and our model.

Mobile Edge Computing (MEC) and Network Function Virtualization (NFV) are widely considered to be key players in the 5G era. Whereas MEC enables to reduce latency significantly by allowing applications to be deployed close to end users, NFV allows for flexible deployment of virtualized network functions. The performance and flexibility can be improved further by combining MEC and NFV. Existing frameworks for managing and orchestrating MEC applications and NFV are either tightly coupled or completely separated. The former design is inflexible and increases the complexity of one framework, while the latter leads to inefficient use of computation resources. In this paper, we extend our Automated Provisioning Framework for MEC (APMEC), which combines each MEC application and its respective network service (comprising a chain of virtual network functions) in a MEC service. We propose a novel MEC service placement algorithm allowing to reuse a subset of yet underloaded network functions. Our evaluation results, obtained from a testbed implementation and simulations, show that our solution allows to remarkably increase the utilization of network functions and simultaneously reduce routing cost. Specifically, it allows to accept at least 60% more user requests, and at the same time lowering the routing cost by more than 30%, as compared to the baseline approach.

A wide variety of applications ranging from distributed storage to Forward Error Correction (FEC) benefit from Random Linear Network Coding (RLNC). Recent research activities exploited Network Coding as a Service (NCaaS) as FEC for Software-Defined Networking (SDN). Based on the aforementioned research, in this paper we propose a more efficient approach to use RLNC in a reliability-centric network. By leveraging both, the knowledge possessed by the SDN switches on channel conditions and the dynamic flexibility provided by the virtualized coding instances, we developed a novel method to adapt the code rate of the underlying FEC, to support two different transport protocols with varying reliability and throughput requirements. We have implemented our approach in a realistic SDN emulator to evaluate it in a communication network with time-variant links for TCP and UDP flows comparing against existing fixed-code rate approaches. The results show that our method adapts the flows respectively to the channel conditions and thus delivers a good trade-off between reliability, bandwidth, and overhead. Specifically, our loss estimation algorithm can precisely estimate future losses with a deviation of up to 3%. Our approach is also able to precisely determine the delivery probability with a maximum deviation of 1.5%. In addition, we show that our adaptive redundancy enables TCP to achieve a stable throughput despite losses.

Software-defined networking (SDN) and network function virtualization (NFV) processed in multi-access edge computing (MEC) cloud systems have been proposed as critical paradigms for achieving the low latency requirements of the tactile Internet. While virtual network functions (VNFs) allow greater flexibility compared to hardware-based solutions, the VNF abstraction also introduces additional packet processing delays. In this paper, we investigate the practical feasibility of NFV with respect to the tactile Internet latency requirements. We develop, implement, and evaluate Chain-based Low latency VNF ImplemeNtation (CALVIN), a low-latency management framework for distributed Service Function Chains (SFCs). CALVIN classifies VNFs into elementary, basic, and advanced VNFs; moreover, CALVIN implements elementary and basic VNFs in the kernel space, while the advanced VNFs are implemented in the user space. Throughout, CALVIN employs a distributed mapping with one VNF per Virtual Machine (VM) in a MEC system. Furthermore, CALVIN avoids the metadata structure processing and batch processing of packets in the conventional Linux networking stack so as to achieve short per-packet latencies. Our rigorous measurements on off-the-shelf conventional networking and computing hardware demonstrate that CALVIN achieves round-trip times from a MEC ingress point via two elementary forwarding VNFs (one in kernel space and one in user space) and a MEC server to a MEC egress point on the order of 0.32 ms. Our measurements also indicate that MEC network coding and encryption are feasible for small 256 byte packets with an MEC latency budget of 0.35 ms; whereas, large 1400 byte packets can complete the network coding, but not the encryption within the 0.35 ms.

With the explosion of the number of compute nodes, the bottleneck of future computing systems lies in the network architecture connecting the nodes. Addressing the bottleneck requires replacing current backplane-based network topologies. We propose to revolutionize computing electronics by realizing embedded optical waveguides for onboard networking and wireless chip-to-chip links at 200-GHz carrier frequency connecting neighboring boards in a rack. The control of novel rate-adaptive optical and mm-wave transceivers needs tight interlinking with the system software for runtime resource management.

Future communication systems, such as those enabling the Tactile Internet, will face disruptive changes compared to the state-of-the-art systems, which are 1) highly dynamic topology changes; 2) replacement of the end-to-end paradigm by real mesh topologies; and 3) a massive number of devices. To overcome these disruptive changes, future communication systems will substitute specialized hardware with generic hardware boxes and the softwarization paradigm. Furthermore, this approach will allow for a quick deployment of new services, which was known to the cloud service already. In this paper, we will introduce the most prominent candidates for softwarization such as software-defined networking (SDN) and network function virtualization (NFV) and explain the importance of these technologies for the upcoming 5G communication system and Tactile Internet applications realizing novel mobile edge computing, storage, and networking solutions. Specifically, we will discuss use cases of SDN/NFV such as network coding as a service, and ultrareliable distributed edge caching. Finally, we will describe our holistic testbed at the 5G Lab Germany as a fundamental step toward creating an experiment infrastructure that anticipates the 5G communication systems and Tactile Internet applications.

The 5G communication system will provide flexible and programmable networks by leveraging softwarization technologies such as NFV and SDN. That advanced feature of 5G networks will be demonstrated in the context of real-time video surveillance for public safety. Specifically, we demonstrate for the first time a practical deployment of Random Linear Network Coding (RLNC) with NFV and SDN technologies to improve video quality against packet loss due to congestion at the core network and signal impairment of lossy channels at network edges. The demonstration implements NFV and SDN applications on COTS devices to prove its flexibility and portability.

The sharp increase in the video traffic and hand-held devices in the last decade warrant for new paradigms of applications including massive video multicasting, such as in the football stadiums, or concert halls where multiple angles of the live feed could be multicasted to the users' smart phones or tablets. However, the current state-of-the-art cellular solution is simply individually unicast the video stream to each node, which is extremely resource-inefficient. In this demo, we show how we perform reliable massive video multicasting over standard LTE links by combining the principle of cooperative networking and Random Linear Network Coding. We offload the majority of the traffic from LTE onto Wifi multicasts, thus increasing the efficiency of the LTE channel usage by multiple orders of magnitude. This demo has been demonstrated at the Mobile World Congress 2017 as a part of Nokia's booth to demonstrate the 5G stadium experience.

Massive content delivery in cellular networks is in the spotlight of the research community as data traffic is increasing at an incredibly fast pace. The existing LTE-A implementation for content broadcast presents several issues such as indoor coverage, along with low energy and spectral efficiency. Therefore, novel systems that provide efficient massive content delivery and reduced energy consumption are needed. In this paper we present a massive content distribution protocol that combines the benefits of cooperative mobile clouds (CMCs) with Random Linear Network Coding (RLNC) through multicast WiFi links. Our main goal is to offload data traffic from the LTE-A link and to reduce the energy consumption at the cooperating UEs. We solve the problem of excessive signaling that oftentimes arises in cooperative approaches by eliminating feedback messages within the CMCs. Instead, we provide a simple but accurate analytic model to correctly configure the number of coded transmissions to be performed within the CMCs. Results show that energy savings of more than 37 percent can be achieved with our protocol when compared to direct content download from the cellular base station. Furthermore, bandwidth utilization at the LTE-A link is sharply reduced.

Service Function Chaining allows the flexible and efficient deployment of network functions for different applications. With Network Function Virtualization the elements of the chain can be provisioned in virtual environments on any COTS hardware. This introduces the question of where to position the individual network functions within the virtualization environment. This problem of network function placement has been studied in theory as an optimization problem on a graph. However it is challenging to apply theoretical work on practical deployments. In this paper, we perform a measurement campaign to study the delay introduced by Service Function Chaining. We propose placement heuristics and evaluate the performance on OpenStack. With the proposed heuristics, the service delay can be reduced by more than 20%. We measured the overhead introduced by a network function implemented in user space. The processing delay in user space can be twice as much as the same function in kernel space. More interestingly, we identified a service interruption of more than 1 second after activation of the chain.

In pull-based Peer-to-Peer video streaming systems, peers exchange buffer maps to reveal the availability of video chunks in their buffer. When collecting these buffer maps, a malicious party can infer the system's overlay structure and even identify head nodes, the direct communication partners of the stream's source. Attacking these head nodes can isolate peers from the source resulting in a disruption of the video dissemination for most peers in the system. We introduce a lightweight SWAP scheme, which allows peers to proactively change their partners, to reduce the chance of head nodes to be identified by such an inference attacker. Extensive simulation studies demonstrate that our scheme effectively undermines the attack's accuracy in identifying head nodes. So, SWAP lowers the chunk miss ratio while causing only a slight increase in signaling overhead.

Hybrid Peer-to-Peer streaming systems combine the advantages of an efficient push-based with a more resilient pull-based system to deliver video streams over the Internet. In this manner, hybrid systems offer low latency and an increased robustness to failures and node churn. However, current hybrid systems is vulnerable to misbehaving nodes and deliberate attacks. By taking central positions in the overlay, malicious nodes can perform extremely harmful Denial-of-Service (DoS) attacks. We propose RBCS, a novel backbone construction scheme, that is highly resilient against DoS attacks while maintaining fast content dissemination. RBCS incorporates stable peers into a manipulation-resistant multi-tree backbone overlay, which is resilient against both attacks and node churn. Additionally, RBCS securely identifies stable peers by using only local knowledge about the participation time of others. Extensive simulations indicate that RBCS outperforms the state-of-the-art in being more resilient against attacks at the price of a slightly increased overhead.

Our main contribution in this work is a deployable multitree-push system for P2P-based live streaming. It runs on both desktop PCs and Android-based mobile devices. Additionally, it provides controlling, monitoring, and measurement functionalities which help with debugging in the development phase, visualize the topology during a demonstration, and support the deployment of test scenarios in a distributed setting. Besides, the generic architecture of the system also allows for the extension to other classes of streaming systems.