A5060308984- Software-Defined Networks and 5G
- Network Traffic and Congestion Control
- Advanced Optical Network Technologies
- Caching and Content Delivery
- Interconnection Networks and Systems
- Peer-to-Peer Network Technologies
- Cloud Computing and Resource Management
- IoT and Edge/Fog Computing
- Network Security and Intrusion Detection
- 3D Surveying and Cultural Heritage
- Internet Traffic Analysis and Secure E-voting
- Wireless Networks and Protocols
- Service-Oriented Architecture and Web Services
- Robotics and Sensor-Based Localization
- Software System Performance and Reliability
- Network Time Synchronization Technologies
- Advanced Graph Neural Networks
- Distributed and Parallel Computing Systems
- Image and Video Quality Assessment
- Advanced Malware Detection Techniques
- Energy Efficient Wireless Sensor Networks
- EEG and Brain-Computer Interfaces
- Energy Harvesting in Wireless Networks
- Explainable Artificial Intelligence (XAI)
- Image and Signal Denoising Methods
Eötvös Loránd University
2015-2024
ORCID
2021
University of Debrecen
2013
The localization of Internet hosts opens space for a wide scope applications, from targeted, location aware content provision to localizing illegal content. In this paper we present novel probabilistic approach, called Spotter, estimating the geographic position devices with remarkable precision. While existing methods use landmark specific calibration building their internal models show that delay-distance data follow generic distribution each landmark. Hence, instead describing in manner...
Smart cities offer services to their inhabitants which make everyday life easier beyond providing a feedback channel the city administration. For instance, live timetable service for public transportation or real-time traffic jam notification can increase efficiency of travel planning substantially. Traditionally, implementation these smart require deployment some costly sensing and tracking infrastructure. As an alternative, crowd be involved in data collection via mobile devices. This...
Although the programmability of control planes has been thoroughly examined in past years, only a limited number studies go beyond consideration that data plane is collection simple packet forwarding devices. Even OpenFlow, popular, very expressive programming language, still restricted to supporting subset existing protocol headers. To overcome such limitations, new models have recently emerged. One them P4, high-level language for processors enables great flexibility description structures...
To manage and monitor their networks in a proper way, network operators are often interested identifying the applications generating traffic traveling through networks, doing it as fast (i.e., from few packets) possible. State-of-the-art packet-based classification methods either based on costly inspection of payload several packets each flow or basic statistics that do not take into account packet content. In this paper we consider intermediate approach analyzing only first bytes (or few)...
P4 is a high level language for programming network switches that allows great flexibility in the description of packet structure and processing, independent specifics underlying hardware. In this demo, we present our prototype compiler which hardware specific functionalities are separated. We have identified requisites latter, form interface target Hardware Abstraction Library (HAL); turns code into core program linked to library invokes its operations. The two stage separation improves...
High level, network programming languages, like P4, enable students to gain hands-on experience in the structure of a switch or router. Students can implement packet processing pipeline themselves, without prior knowledge circuit design. However, when choosing P4 programmable target for use classroom, instructors face lack options. On one hand, software solutions, such as behavioral model (BMv2) switch, are overly simplified and offer low performance. other existing hardware solutions closed...
Novel applications and network scenarios challenge existing traffic management strategies. Hierarchical Quality of Service (HQoS) provides a fine control resource sharing delay, but traditional HQoS solutions have challenging complexity that prevents their deployment in the engine high-speed switches. Programmable switches emerged to make packet processing pipelines flexible reconfigurable, capabilities still rely on fixed functions cannot handle approaches. In this paper, we show how...
Industrial Internet of Things (I-IoT) applications require a large number sensor data to be processed under tight delay and jitter constraints. In such applications, flexible event detection fast reaction critical events is an important building block. Traditional approaches use either proprietary networks dedicated hardware or transmit towards processing elements in the Cloud at Network Edge, using distributed stream frameworks. For scalability, servers are needed on commodity CPUs...
This study outlines two novel techniques which can be used in the area of IP geolocation. First we introduce a detailed path-latency model to able determine overall propagation delays along network paths more accurately. knowledge then leads precise geographic distance estimation between routers and measurement nodes. In addition application model, describe method utilizes high-precision one-way delay measurements further increase accuracy router geolocation techniques. The values are as...
The geographic layout of the physical Internet inherently determines important network properties and traffic characteristics. To give insight into geography Internet, we examine spatial topology routing. represent conducted a geographically dispersed traceroute campaign, embedded extracted space by applying novel IP geolocalization service, called Spotter. In this paper present frequency analysis link lengths, quantify path circuitousness explore symmetry end-to-end routes.
Active Queue Management (AQM) has been used in routers for early detection of traffic congestion at the bottleneck link. In AQM, packets can be drop before buffer becomes full, based on many different parameters or conditions. Many algorithms have proposed to efficiently control network. The scope this paper is analyze most recent and well-known AQM evaluate their performance under heavy hybrid load (TCP UDP flows), using NS-3 network simulator. For comparative analysis, various metrics are...
The killer features of the next-generation 5G mobile standard, including edge computing and network slicing, will be very difficult to support with traditional fixed-function appliances. Rather, core depend on programmable switches, which allow packet processing functionality reconfigured fly in order deploy virtualized functions service chains instantaneously. With close horizon, it has become crucial identify price for programmability software data plane, considering expected complexity...
Active Queue Management (AQM) addresses the problem arising from using unnecessarily large, unmanaged buffers and thus it aims at improving network application performance. AQM methods introduce different drop policies to proactively packets according queue states parameters. Though we are living in an renaissance, comprehensive methodology framework for performance evaluation under realistic traffic loads still missing. With advent of P4, description, validation algorithms a generic have...
Reinforcement learning has gone through an enormous evolution in the past ten years. It's practical applicability been demonstrated several use cases various fields from robotics to process automation. In this paper, we examine how tools of deep Q-learning can be used AQM algorithm reduce queuing delay and ensure good link utilization at same time. The proposed method called RL-AQM advantage that it is less prone parameterization automatically adapt new network conditions. prototype...
Social networks like Twitter and Facebook have gained a significant popularity with people from all parts of the society in past decade, providing new kind data source for novel social-aware applications. A great majority users are online time, posting real-time information on various topics including unpredicted events. An accident or natural disaster is often posted social hours before appearing traditional news. In this paper, we outline framework event detection data. contrast to prior...
To manage and monitor their networks in a proper way, network operators are often interested automatic methods that enable them to identify applications generating the traffic traveling through as fast (i.e., from first few packets) possible. State-of-the-art packet-based classification either based on costly inspection of payload several packets each flow or basic statistics without taking into account packet content. In this paper, we consider an intermediate approach analyzing only bytes...
In this paper, we propose CSAQM, a novel approach of Active Queue Management (AQM) for network traffic with multiple Congestion Control algorithms. Our goal is similar to that PI2 AQM supports both Classic and Scalable TCP in single queue. contrast existing solutions, CSAQM has two key advantages: 1) arbitrary congestion controls (including unresponsive flows) can also be handled 2) it rich set resource sharing policies beyond equal sharing. based on the core-stateless Per Packet Value...
The excess buffering of packets in network elements, also referred to as bufferbloat, results high latency. Considering the requirements traffic generated by video conferencing systems like Zoom, cloud rendered gaming platforms Google Stadia, or even streaming services such Netflix, Amazon Prime and YouTube, timeliness is important. Ensuring low latency IP flows with a throughput calls for application Active Queue Management (AQM) schemes. This introduces yet another problem co-existence...
Industrial networks rely on standard real-time communication protocols, such as ProfiNet RT.These protocols use cyclic data exchange between IO devices and controllers.Each device reports its internal state to the controller at a predefined frequency, even if of is unchanged.These are essential accurately monitor health devices, but network resources limited it not advisable overload with unnecessary packets.The traffic generated by single insignificant, in an industrial site hundreds number...
Active queue management (AQM) techniques have evolved in the recent years, after defining bufferbloat problem. In parallel novel congestion control (CC) algorithms been developed to achieve better data transport performance, often assuming simple tail dropping buffers. On other hand, AQM usually assume legacy CC (Cubic). Though all of and improve performance under these assumptions, their co-existence has not or only partially tested so far. Similarly, router buffer sizing was studied mainly...
In the recent years a new Internet service called Low-Latency Low-Loss Scalable-Throughput (L4S) has been proposed to provide scalable sources with ultra-low latency over Internet, while keeping delay and resource sharing reasonable for flows classic loss-based congestion control. At same time both controls are being evolved resulting in unfair share among using different controls, even within control family. The fairness is further degraded by heterogeneous RTTs real-world Internet. this...
This paper focuses on hybrid pipeline designs for User Plane Function and next-generation NodeB leveraging target-specific features an insightful discussion of P4 target challenges limitations. The entire or disaggregated UPF runs targets allocates packet processing data paths in hardware DPDK/x86 software based flow characteristics (e.g., heavy hitters) QoS requirements low-latency slices). For the gNodeB, most is executed commodity Tofino hardware, while unsupported functions such as...