With the advent of next-generation networks, new applications across multiple domains are gaining traction. This shift demands a redefined network paradigm, where ultrareliable, low-latency communication is key. In this work, we explore and extend the concept of in-network programmability in new directions. Unlike conventional approaches, we leverage P4 data plane programmability to implement an in-network collision avoidance algorithm in a UAV scenario. We evaluate our hardware-based implementation under different conditions, including latency and velocity, demonstrating that it efficiently detects and prevents collisions. Our results show the impact of end-to-end latency and highlight how in-network processing can be a valuable ally for time-sensitive tasks, paving the way for future advancements in hardware-based in-network applications.

Time-Sensitive Networking (TSN) is expected to provide reliable, low-latency communication for critical systems. Leveraging the Frame Replication and Elimination for Reliability (FRER) protocol, it protects against packet loss by replicating individual packets and delivering them on disjoint forwarding paths. FRER does not contain any in-built security solutions, and several FRER-related security vulnerabilities have recently been identified that could undermine TSN’s ultimate goal of providing extreme reliability. In this paper, we introduce a comprehensive security-focused testbed for analyzing FRER vulnerabilities. Our self-contained setup employs open and adaptable components, runnable on a single server, ensuring flexibility and accessibility. Leveraging eBPF/XDP, it efficiently implements FRER’s data plane functionalities, accommodating both fast-path and slow-path attacks. Parameters like delay, jitter, loss rate, and bandwidth are easily customizable. We validate the testbed’s effectiveness with various attack scenarios.

This paper presents a deep learning approach for drone navigation aimed at accurately estimating a drone's position based on Received Signal Strength (RSS) in a simulated environment. A predefined route is established, and the drone collects RSS and Inertial Measurement Unit (IMU) data while traversing it multiple times. This dataset serves as the foundation for training various neural network architectures. We create multiple training and test datasets to compare different models and identify the most effective approach for predicting the drone's location. After training, the neural networks are evaluated using distinct test data, showing that deep learning models can reliably predict a drone's position in simulation. The developed predictive models enhance the accuracy and reliability of autonomous drone operations in real-world scenarios.

Manufacturing processes are often monitored by IP cameras. The generated video streams can be transferred via a wireless link to be processed and used by remote industrial controllers that manage and configure the industrial task in real-time. However, the link has limited bandwidth and is not capable of transmitting the aggregated traffic of all the IP cameras. Moreover, the platform provider of the remote controller (e.g., cloud) might charge extra fees for high volumes of network traffic. To optimize the data transport, we propose an in-network video quality control method for IP camera streams that drops non-essential frames from temporally irrelevant IP camera streams even when bandwidth is available. Our solution introduces a high-level API through which the operator can define which camera streams are needed with high quality at which actuator positions/states of the industrial process. Our method assumes an aggregation point that monitors the actuators' states and reduces the quality of camera streams that are not important for the control process, selectively dropping a set of video frames. We have implemented a P4-based prototype of the aggregation point and show that the remote controller can be fed with high-quality video streams while meeting bandwidth limitations and potentially saving data transfer costs without modifying the end-points.

Mean Opinion Score (MOS) and Estimated Mean Opinion Score (EMOS) are originally defined for the evaluation of voice and video codecs. In this paper, we introduce the concept of the application of EMOS on industrial processes to support smart network management. We make measurements in the function of network performance on a cyber-physical system performing a sanding use case. A demonstration video is available at [1].

Industrial networks rely on standard real-time communication protocols like ProfiNet. These protocols are used for cyclic data exchange between IO devices and controllers. Since continuous monitoring of IO devices is important, a large number of data packets can be observed in such field networks between the IO devices and controllers. Each IO device cyclically reports its data or internal state to a controller at a predefined frequency. However, the reported data of most IO devices are not changing all the time, and thus the same bytes are transmitted multiple times. The majority of data packets is only used for checking the availability of such devices. In this paper, we consider an industrial environment where IO devices are located in an industrial site while controllers are running remotely (e.g., a software PLC in a private or edge cloud), and there is a radio link (e.g., 5G radio) between the two sides. We propose an in-network traffic reduction method that filters out the unnecessary data traffic at the two ends of the radio link, detects failure of devices and the radio link fast, and does not require any modification in the IO devices and controllers. Our solution is based on the cooperation of two P4-programmable networking elements deployed at the two sides of the radio link. Our preliminary measurements with P4-programmable hardware switches and emulated ProfiNet devices show that the method can significantly reduce the load on the radio link, while it could seamlessly be deployed in existing industrial environments.

Resource allocation in wireless access networks has been an intensively researched topic recently: many proposed solutions tackle radio channel access and dynamic spectrum allocation, but traditional issues of queuing, bandwidth sharing and packet processing at wireless access points have been targeted as well. In most of the related work the competition for high quality of service is usually solved by central coordination among users via optimizing a specific target aspect of the overall communication. In this paper we take a turn and provide users with the possibility of resource allocation suggestions. We propose a wireless access sharing framework, in which users have a say in optimizing their quality of service on the long term, and we tackle its analysis with the tool set of stochastic game theory. Our findings show that greedy users become polite against their counterparts when the load is relatively low with the goal of preparing for situations with high load.

Digital Twin (DT) is widely used in various industrial sectors to optimize the operations and maintenance of physical assets, system and manufacturing processes. In this paper our goal is to introduce an architecture in which the radio access control happens automatically to minimize the utilized radio resources while still maximizing the production KPIs of the robot cell. To achieve this, we apply Reinforcement Learning (RL) in a simulated environment to explore the environment fast, while the DT ensures that the learned policy can be applied on the real world environment as well. We show that the application of Ultra Reliable Low Latency Communication (URLLC) connection can be reduced to approx. 30% of the total radio time while achieving real-world accurate robot control. The system in action can be seen on [1].

Safe and efficient Human-Robot Collaboration (HRC) is an essential feature of future Industry 4.0 production systems which requires sophisticated collision avoidance mechanisms with intense computation need. Digital twins provide a novel way to test the impact of different control decisions in a simulated virtual environment even in parallel. In addition, Virtual/Augmented Reality (VR/AR) applications can revolutionize future industry environments. Each component requires extreme computational power which can be provided by cloud platforms but at the cost of higher delay and jitter. Moreover, clouds bring a versatile set of novel techniques easing the life of both developers and operators. Can these applications be realized and operated on today's systems? In this demonstration, we give answers to this question via real experiments.

Current Internet transport protocols are being revisited in an attempt to respond to traffic growth in size, diversity and the emergence of new applications. Designed to reduce Web TCP latency and connection establishment time, QUIC uses UDP and defines its own congestion control. To verify some of QUIC's performance claims, we carried out an extensive set of controlled experiments reflecting Internet traffic conditions by configuring various parameters, such as the round trip time (RTT), the packet loss ratio, web pages, and caching. Surprisingly, our experimental results show that QUIC presented worst page load time (PLT) results than TCP. We next present some of our interpretations of these results.

This paper investigates the possibilities of relaxing the wireless resource consumption of a remote robot cell control use case. We propose a Quality of Control (QoC)-aware wireless resource allocation strategy that is based on the categorization of the robotic arm movement phases into high and low QoC requiring movements. We evaluate our proposed method in a simulation environment against the local controlled based solution of the Agile Robotics for Industrial Automation Competition (ARIAC) [1]. The evaluation shows that 54% of the radio time on average can be saved without affecting the productivity of the robot cell.

In this paper, we propose an architecture for network traffic prioritization. This architecture is application and protocol agnostic. We also give an interface for this architecture which enables QoE triggered fast network interface switching without interrupting the communication sessions. We give an algorithm in order to minimize the cost of the frequent network changes (reordering and retransmissions) and evaluate its performance in simulations and demonstrating it on real Wi-Fi and cellular access network. To demonstrating the effectiveness of our QoE triggered switching approach we eliminate the stalling events on YouTube video playback.

We propose an app agnostic QoE deduction method that can be used for on-line traffic prioritization. Our implemented method extracts QoE information directly from the screen of mobile devices and this information is utilized for video traffic prioritization to improve user QoE. We evaluate our method and show that it can realize 80-100% of the achievable gain of the state-of-the-art methods with 50% less traffic that needs to be prioritized to achieve this. We provide deployment alternatives for traffic prioritization and evaluate performance impact on achievable gain.

In this paper we propose evaluation models and calculate range of potential gains for enhancing the average web-browsing QoE of all the users in a mobile network with given capacity. We investigate a method that takes capacity from users during QoE insensitive periods. The gained capacity is distributed among users being in QoE affecting state. Our models of web page network prioritization show that the potential gain for QoE affecting periods highly depends on the correlation between data downloaded during QoE affecting and QoE insensitive periods. The numerical evaluation of wide range of scenarios with web traffic shows that the achievable gain is up to 104% i.e., halving the download time of QoE affecting part. Network prioritization of web pages has good potential to improve web-browsing QoE.

Commercial implementations of adaptive streaming systems are either too aggressive or too conservative, not being able to efficiently deliver video content to the client. Because of these issues, a great number of adaptive streaming systems have been proposed in the academia. They employ different strategies, to select the proper video level that the server should send to the client. One trend is the use of control theory, because it guarantees system stability, and enables simulation and tuning of the system prior to its implementation. However, the current proposals that employ control theory do not take full advantage of this technique, because they do not thoroughly analyze the system parameters and their influence on the overall system performance. Therefore, it is possible that they use suboptimal solutions. In this paper, we propose a control-theoretic approach for selecting the video quality in a DASH environment. We use a controller added to a state machine, to improve the QoE and to maximize bandwidth utilization. We also present a comparison of several system settings performances, by changing the controller types, their gains, and the state machine parameters. Experimental results show the effectiveness of our proposed system for buffer time control and video level smoothness, and that it outperforms a current proposal in the academia.

Compressed Deterministic Finite Automata (DFA) promises same representation power as traditional DFAs while using less memory for representing Regular Expressions (RE). Experimental evaluations of DFA-based Deep Packet Inspection (DPI) systems focus mainly on memory consumption without observing other important related aspects, such as the matching speed. Proper design of DPI systems requires the assessment of several performance metrics at hardware level, in order to make sure that its implementation will not compromise the overall performance. This paper proposes a novel and systematic evaluation of DPIs and reveals the impact of DFA's data-structures and the correspondent memory layout implementation to hardware-level metrics. Experimental results show that some DFA model and memory layout combinations are almost 100 times faster than others. Results also show that choosing the incorrect model-layout pair can lead to significant performance issues. Our methodology and results will certainly help researchers and developers to design efficient DPI engines, through the selection of the best DFA model and memory layout combination to achieve the targeted overall performance.

We present the versatile functionality of our novel user behavior based traffic emulation system in this paper. We show the unique feature of the system, i. e., it is capable of working on different platforms (Windows, Android), on different access technologies (wired, WiFi, 3G) and as a remote controlled system on different sites (Europe, Asia, South America). Our examples exhibit some of the manifold traffic analysis possibilities as a result of this key functionality. We have also made our system available to the public [1].

Channel switching in HSPA networks is used to reduce the channel occupancy when there is no data transfer for the given user, this way reducing battery consumption. This paper is the first dealing with another important aspect that is the CPU load on the radio network controller (RNC) caused by channel switching. This is an optimization task, in which both the channel switching and staying on the high-bandwidth channel have costs. In this paper we propose a system to minimize the costs by applying a predictor based method which uses time slice based features in order to reduce the high variance in the feature values. The proposed system is evaluated and compared to other state-of-the-art methods.

Traffic Identification is a key function performed by Internet Service Providers' (ISP) administrators to evaluate and improve network services. However, traffic identification needs to be done in real-time and at wire speed to be useful for network tuning. Deep Packet Inspection (DPI) is widely used for identifying normal applications and attacks in the network by looking for well-known patterns within the packets. Such patterns are mostly expressed by Regular Expressions (RE), which are then evaluated by abstract machines known as Deterministic Finite Automata (DFA). Some previous studies grouped DFAs together to evaluate multiple patterns on a single DFA match's run. Efficient grouping algorithms would combine several DFAs without exceeding the available machine's memory. This work proposes and evaluates a new method to combine several DFAs into a single one. Additionally we compared this algorithm to state-of-the-art approaches using a compressed DFA model. Experimental results show that our algorithm generates less groups and transitions than existent algorithms.

Current network traffic analysis systems heavily rely on Deep Packet Inspection (DPI) techniques, such as Finite Automata (FA), to detect patterns carried by regular expression (regex). However, traditional Finite Automata cannot keep up with the ever-growing speed of the Internet links. Although there are a number of efficient FA compressing mechanisms for DPIs, there is no standardized or common way to evaluate and compare them. In this scenario, this paper proposes a methodology to evaluate and compare automaton models and the data-structures that materialize them. We also adapt state-of-the-art memory layouts to better fit in today's computer architectures. Finally, we apply our methodology to most important automaton models, memory layouts, and well-known signature sets. The results show us that some memory layouts are not efficient for regexes that represent small automata and other ones which fit only with uncompressed automata. Further, we also found out that theoretical studies about memory usage from memory encodings are not as accurate as they should be.

In this paper we introduce a novel framework for traffic identification that employs machine learning techniques focusing on the estimation of multiple traffic influencing factors. The effect of these factors is handled with the training of several machine learning models. We utilize the outcome of the multiple models via a recombination algorithm to achieve high overall true positive and true negative and low overall false positive and false negative classification ratio. The proposed method can improve the performance of every kind of machine learning based traffic identification engine making them capable of efficient operation in changing network environment i.e., when the probing node is trained and tested in different sites.

Deep Packet Inspection (DPI) systems have been becoming an important element in traffic measurement ever since port-based classification was deemed no longer appropriate, due to protocol tunneling and misuses of well-defined ports. Current DPI systems express application signatures using regular expressions and it is usual to perform pattern matching through the use of Finite Automaton (FA). Although DPI systems are essentially more accurate, they are also resource-intensive and do not scale well with link speeds. Looking to this area of interest, this paper proposes a novel Deterministic Finite Automaton, called Ranged Compressed Deterministic Finite Automaton (RCDFA), that compresses transitions without additional memory lookups. Experimental results show that RCDFA yields space savings of 97% over the original DFA and up to 93% better compression when compared to the DFA's state-of-the-art compression techniques.

Traffic identification and classification are essential tasks performed by Internet Service Provider (ISPs) administrators. Deep Packet Inspection (DPI) is currently playing a key role in traffic identification and classification due to its increased expressive power. To allow fair comparison among different DPI techniques and systems, workload generators should have the following characteristics: (i) synthetic packets with meaningful payloads; (ii) TCP and UDP traffic generation; (iii) a configurable network traffic profile, and (iv) a high-speed sending rate. This paper proposes a workload generator framework which inherits all of the above characteristics. A performance evaluation shows that our flexible workload generator system achieves very high sending rates over a 10Gbps network, using a commodity Linux machine. Additionally, we have configured and tested our workload generator following a real application traffic profile. We then analyzed its results within a DPI system, proving its accuracy and efficiency.

Game traffic depends on two main factors, the game protocol and the gamers' behavior. Based on a few popular real-time multiplayer games, this paper investigates the latter factor showing how a set of typical game phases-e.g., player movement, changes in environment- impacts traffic. By understanding the nature of this impact an algorithm is introduced to grab specific events and states from passive traffic measurements. Further, a measurement example including a detailed analysis is shown from an operational broadband network. The collected data of player behavior in the gaming environment was analyzed and it was also shown that the long range dependence (LRD) property of gaming traffic is not due to the popular explanation based on the heavy-tailed periods of player activities.

The analysis of network traffic can provide important information for network operators and administrators. One of the main purposes of traffic analysis is to identify the traffic mixture the network carries. A couple of different approaches have been proposed in the literature, but none of them performs well for all different application traffic types present in the Internet. Thus, a combined method that includes the advantages of different approaches is needed, in order to provide a high level of classification completeness and accuracy. According to our best knowledge, this study is the first attempt where the currently known traffic classification methods are benchmarked on network traces captured in operational mobile networks. The pros and cons of the classification methods are analyzed, based on the experienced accuracy for different types of applications. Using the gained knowledge about the strengths and weaknesses of the existing approaches, a novel traffic classification method is proposed. The novel method is based on a complex decision mechanism, in order to provide an appropriate identification mode for each different application type. As a consequence, the ratio of the unclassified traffic becomes significantly lower. Further, the reliability of the classification improves, as the various methods validate the results of each other. The novel method is tested on several network traces, and it is shown that the proposed solution improves both the completeness and the accuracy of the traffic classification, when compared to existing methods.