Sandor Laki

Eötvös Loránd University · Department of Information Systems

The traditional method of data plane programming involves deploying a single P4 program to a single target. However , different targets have varying capabilities, functionalities, and support for various programming languages beyond P4. Therefore, disaggregating a single data plane program into multiple subprograms that run on different targets can...

The typical approach to data plane programming involves deploying a single P4 program to a single target. However , different targets have different capabilities, functionalities, and support for various programming languages apart from P4. Consequently, disaggregating a single data plane program into multiple subprograms that run on different targ...

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 secu...

In automated radio network control, understanding the effect of different factors on network performance is crucial. Although there are machine learning (ML) solutions that can reliably anticipate network performance expressed as key performance indicator (KPI) values, these models are typically black-box or provide only partial explanations. Most...

GPUs and programmable data planes have gone through an enormous evolution in the past years. GPUs can be used for modeling the real-world environment accurately, while programmable data planes can monitor the network in real-time and implement novel packet processing and decision logic. In this paper, we investigate how these two distant technologi...

Throughput and latency critical applications could often benefit of performing computations close to the client. To enable this, distributed computing paradigms such as edge computing have recently emerged. However, with the advent of programmable data planes, computations cannot only be performed by servers but they can be offloaded to network swi...

Industrial networks rely on standard real-time communication protocols, such as ProfiNet RT. These real-time protocols use cyclic data exchange between IO devices and controllers. Each IO device reports its internal state to the controller at a predefined frequency, even if the state of the device is unchanged. These reports are essential to accura...

There is a great interest in utilizing P4 for in-network computingalong with programmable data planes. This use is emerging as anew network paradigm that can not just reduce the complexitybut the delay as well. Beamforming is now an integral feature ofmodern wireless communication systems and its implementationcalls for an accurate beam alignment b...

Novel applications and network scenarios challenge existing traffic management strategies. Hierarchical Quality of Service (HQoS) provides a fine control of resource sharing and delay, but traditional HQoS solutions have challenging complexity that prevents their deployment in the traffic management engine of high-speed switches. Programmable switc...

Throughput and latency critical applications could often benefit of performing computations close to the client. To enable this, distributed computing paradigms such as edge computing have recently emerged. However, with the advent of programmable data planes, computations cannot only be performed by servers but they can be offloaded to network swi...

In-network computing has emerged as a new computational paradigm made possible with the advent of programmable data planes. The benefits of moving computations traditionally performed by servers to the network have recently been demonstrated through different applications. In this paper, we argue that programmable data planes could be a key technol...

Local 3D point feature descriptors play an important role in many areas of computer vision, such as object recognition, registration, etc. There are many well-functioning feature descriptors, but they are typically real-valued and multidimensional vectors, leading to high computational complexity in nearest neighbor searches. To overcome this chall...

This paper focuses on hybrid pipeline designs for User Plane Function and next-generation NodeB leveraging target-specific features and an insightful discussion of P4 and target challenges and limitations. The entire or disaggregated UPF runs on P4 targets and allocates packet processing data paths in P4 hardware or DPDK/x86 software based on flow...

Resource sharing is of utmost importance in networking environments where substantial overprovisioning is economically infeasible. Different centralized solutions have recently been proposed for Wide Area Networks (WANs), Access Aggregation Networks and other closed networking domains, relying on different ideas from exploiting the capabilities of...

Nowadays, depth data has an important role in many applications. The sensors which can capture depth data became essential parts of autonomous vehicles. These sensors record a huge amount of 3D data (point clouds with x, y, and z coordinates). Furthermore, for many point cloud processing applications, it is important to calculate feature vectors th...

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 the packet processing pipeline themselves, without prior knowledge of circuit design. However, when choosing a P4 programmable target for use in the classroom, instructors face a lack of opti...

Reinforcement learning has gone through an enormous evolution in the past ten years. It's practical applicability has been demonstrated through several use cases in various fields from robotics to process automation. In this paper, we examine how the tools of deep Q-learning can be used in an AQM algorithm to reduce queuing delay and ensure good li...

LIDAR and depth cameras have gone through a profound technological evolution, making large-scale recording of 3D point cloud data possible which raises new challenges for data processing. Most of the existing 3D point cloud processing methods were developed to work properly when the entire data set fits into the memory of a single server. When poin...

In order to solve the bufferbloat problem, many Active Queue Management (AQM) algorithms have been proposed, such as Random Early Detection (RED), Proportional Integral Controller Enhanced (PIE), Controlled Delay (CoDel), etc. Among them, CoDel has widely been adapted by devices at the last mile. It maintains the sojourn times of transmitted packet...

Network QoS, fairness and resource sharing control are not completely solved problems. Available solutions lack scalability due to maintaining flow state, require re-tuning if traffic changes, focus on a limited set of networking scenarios or require complex, centralized controllers and feedback loops. In this paper, we propose a core-stateless sol...

5G Radio Access Networks consists of two key services: User Plane Function (UPF) and next generation NodeB (gNB). Though several papers have recently demonstrated that the high-level UPF can be described in P4, for the lowest-level gNB service it is more challenging and cannot purely be done with existing programmable switches. In this paper, we sh...

Industrial Internet of Things (I-IoT) applications require a large number of sensor data to be processed under tight delay and jitter constraints. In such applications, flexible event detection and fast reaction to critical events is an important building block. Traditional approaches use either proprietary networks and dedicated hardware or transm...

In addition to classic ones, a new family of congestion controls regarded as "scalable" has recently emerged. Scalable congestion control exploits ECN to provide much finer control as rate scales, enabling reactions proportional to the congestion level. Nowadays, different initiatives propose the use of Scalable congestion control over the public I...

Nowadays, three dimensional point cloud processing plays a very important role in a wide range of areas: autonomous driving, robotics, cartography, etc. Three dimensional point cloud registration pipelines have high computational complexity, mainly because of the cost of point feature signature calculation. By selecting keypoints and using only the...

Analyzing EEG signals can help us make implications about the user’s activities or even thoughts which can result in a myriad of applications. However, clinical EEG monitoring tools are expensive, often immobile and in need of professional supervision. Lately a couple of companies started the production of relatively cheap, easy-to-use, and mobile...

Novel interactive applications require small end-to-end latency for providing end users with good Quality of Experience. To prevent the bufferbloat problem causing increased delay, various AQM solutions have recently emerged. One of the most widely adopted method is called CoDel that detects the increased delay by maintaining the per packet sojourn...

Core-Stateless resource sharing control identifies flow aggregates in the edge and marks packets based on the relevant policies. Within the core of the network, no flow identification or policy knowledge is needed to realize the desired resource sharing. As packet marking needs trust and the behavior of core-routers is independent of the number of...

Active Queue Management (AQM) addresses the problem arising from using unnecessarily large, unmanaged buffers and thus it aims at improving network and application performance. AQM methods introduce different drop policies to proactively drop packets according to queue states and parameters. Though we are living in an AQM renaissance, comprehensive...

Extending fairness to multiple timescales creates the right incentives for users and provides better QoE for short sessions, e.g. for web page download. In this paper, we show how to define and implement multi-timescale fairness among flows independently of actual traffic mixes and resource capacities. The proposed method is built on the top of the...

Network QoS, fairness and resource sharing control are open challenges of network slicing and virtualization in 5G and future networks providing ultra-high speed Internet access. Traditional stateful solutions either employ the one-size-fits-all approach to provide services to end-users, regardless of the requirements of vertical services, or requi...

The killer features of the next-generation 5G mobile standard, including mobile edge computing and network slicing, will be very difficult to support with traditional fixed-function network appliances. Rather, the 5G core will depend on programmable switches, which allow packet processing functionality to be reconfigured on the fly in order to depl...

Various augmented and virtual reality (AR/VR) applications have emerged in the past years and their popularity are still increasing. Most of these applications work near real-time, posing low-latency requirements against the algorithms to be used. The high computational complexity of 3D mesh generation from continuous 3D point cloud streams provide...

With P4 gaining traction to define datapath pipelines along auto-generated control plane APIs, the protocol-independence and increased flexibility add non-trivial hazards when it comes to functional and in-depth performance evaluation. P4-dependent workload traces are needed along automated methods to populate the tables of the datapath under test...

The Finite Impulse Response (FIR) filter is one of the most important components in digital communications. Therefore, any enhancement of the FIR filter design will improve the efficiency of digital communications. There are several methods proposed to design or improve FIR filters. Involving Artificial Neural Network (ANN) in the FIR designing pro...

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 of PI2 AQM that supports both Classic and Scalable TCP in a single queue. In contrast to existing solutions, CSAQM has two key advantages: 1) arbitrary congestion controls (in...

Although the programmability of control planes has been thoroughly examined in the past years, only a limited number of studies go beyond the consideration that the data plane is only a collection of simple packet forwarding devices. Even OpenFlow, a popular, very expressive data plane programming language, is still restricted to supporting a subse...

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 the buffer becomes full, based on many different parameters or conditions. Many algorithms have been proposed to efficiently control the congestion in the network. The scope of this paper is to...

Before intruding into a system attackers need to collect information about the target machine. Port scanning is one of the most popular techniques for that purpose, it enables to discover services that may be exploited. In this paper we propose an accurate port scan detection method that can detect port scanning attacks earlier with higher reliabil...

In this paper, we propose the PVPIE Active Queue Management (AQM) method that combines the packet scheduling and dropping algorithms of PIE AQM and the packet marking-based resource sharing of the Per Packet Value (PPV) concept. The algorithm calculates dropping probabilities needed for keeping the queueing delay at a predefined level using the PIE...

P4 is a high level language for programming network switches that allows for great flexibility in the description of packet structure and processing, independent of the specifics of the underlying hardware. In this demo, we present our prototype P4 compiler in which the hardware independent and hardware specific functionalities are separated. We ha...

We have digital imprints generated by Smart City, banking systems, wearable sensors, IoT, mobile networks, social networks. The growing volumes of data generated by the information systems obtain values after proper analysis. Data algorithms and methods applicable for small volumes of data, do not work with several terabytes of data. There is a gro...

This article presents the design and pilot implementation of a suite of intelligent methods, algorithms, and tools for federating heterogeneous experimental platforms (domains) toward a holistic Future Internet experimentation ecosystem. The proposed framework developed within the NOVI research and experimentation European collaborative effort, aim...

The NOVI Information Model (IM) and the corresponding data models are the glue between the software components in the NOVI Service Layer. The IM enables the communication among the various components of the NOVI Architecture and supports the various functionalities it offers. The NOVI IM consists of three main ontologies: resource, monitoring and p...

To manage and monitor their networks in a proper way, network operators are often interested in automatic methods that enable them to identify applications generating the traffic traveling through their networks as fast (i.e., from the first few packets) as possible. State-of-the-art packet-based traffic classification methods are either based on c...

To understand the long-term dynamics of networks engineers and network scientists collect tremendous amount of data and distribute them across many different data warehouses. In EU FP7 OpenLab project we developed the nmVO, which helps handling distinct data sources together in a common way efficiently. It also supports data collecting systems with...

Smart cities offer services to their inhabitants which make everyday life easier beyond providing a feedback channel to the city administration. For instance, a live timetable service for public transportation or real-time traffic jam notification can increase the efficiency of travel planning substantially. Traditionally, the implementation of the...

In the past decade, Internet has radically changed our lives, enabling us to obtain information on everything (disasters, political decisions, ordinary events, etc.) we are interested in in almost real-time. Downloading a web page by a browser, instant messaging or file sharing generate huge amount of network traffic that carry valuable information...

Social networks like Twitter and Facebook have gained a significant popularity with people from all parts of the society in the past decade, providing a new kind of data source for novel social-aware applications. A great majority of the users are online all the time, posting real-time information on various topics including unpredicted events. An...

In this paper we introduce a novel neighbor selection strategy for tracker-based peer-to-peer systems like BitTorrent that can uniformly distribute the load among peers in the network. Our method is based on a balanced multiple choice algorithm which takes into account not only the actual load of a peer, but the possibility as well that it will be...

In this paper we propose a new way of constructing the evolving graph in BitTorrent (BT) as new peers join the system one by one. The maximum degree in the constructed graph will be O(1) while the diameter will remain O(ln n), with high probability, where n is the number of nodes. Considering a randomized upload policy, we prove that the distributi...

Monitoring and measurement is a fundamental building block for developing and testing new protocols, routing algorithms and networked applications. In a federated virtualized testbed they allow other service components and testbed-users to follow the current state of the network, and on the other hand they enable intelligent automatic decision-maki...

The geographic layout of the physical Internet inherently determines important network properties and traffic characteristics. In this paper, we examine the spatial properties of the Internet topology and routing. To represent the network we conducted a geographically dispersed traceroute campaign, and embedded the extracted topology into the geogr...

To manage and monitor their networks in a proper way, network operators are often interested in identifying the applications generating the traffic traveling through their networks, and doing it as fast (i.e., from as few packets) as possible. State-of-the-art packet-based traffic classification methods are either based on the costly inspection of...

The geographic layout of the physical Internet inherently determines important network properties and traffic characteristics. To give insight into the geography of the Internet, we examine the spatial properties of the topology and routing. To represent the network we conducted a geographically dispersed traceroute campaign, and embedded the extra...

The geographic layout of the physical Internet inherently determines important network properties. In this paper, we analyze the spatial properties of the Internet topology. In particular, the distribution of the lengths of Internet links is presented — which was possible through spatial embedding of a representative set of IP addresses by applying...

The localization of Internet hosts opens space for a wide scope of applications, from targeted, location aware content provision to localizing illegal content. In this paper we present a novel probabilistic approach, called Spotter, for estimating the geographic position of Internet devices with remarkable precision. While the existing methods use...

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 be able to determine the overall prop-agation delays along the network paths more accurately. The knowledge of accurate prop-agation delay values then leads to more precise geographic distance estimation betw...

To characterize the structure, dynamics and operational state of the Internet it requires distributed measurements. Although in the last decades several systems capable to do this have been created, the easy access of these infrastructures and orchestration of complex measurements is not solved. We propose a system architecture that combines the fl...

ETOMIC is a network traffic measurement platform with high precision GPS-synchronized monitoring nodes. The infrastructure is publicly available to the network research community, supporting advanced experimental techniques by providing high precision hardware equipments and a Central Management System. Researchers can deploy their own active measu...

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 be able to determine the overall propagation delays along the network paths more accurately. This knowledge then leads to more precise geographic distance estimation between network routers and measurement no...

Efficient and reliable available bandwidth measurement remains an important goal for many applications. In this paper we introduce an empirical bandwidth estimation tool based on neural networks. Training the neural network on simulation data, it provides reliable estimation of physical and available bandwidth for simulated single and multi-hop net...