arXiv subjects

Klaus Wehrle

Publications and source records attributed to Klaus Wehrle.

52 records · Page 3Linked to original sources

arXiv2019

DeePCCI: Deep Learning-based Passive Congestion Control Identification

Transport protocols use congestion control to avoid overloading a network. Nowadays, different congestion control variants exist that influence performance. Studying their use is thus relevant, but it is hard to identify which variant is used. While passive identification approaches exist, these require detailed domain knowledge and often also rely on outdated assumptions about how congestion control operates and what data is accessible. We present DeePCCI, a passive, deep learning-based congestion control identification approach which does not need any domain knowledge other than training traffic of a congestion control variant. By only using packet arrival data, it is also directly applicable to encrypted (transport header) traffic. DeePCCI is therefore more easily extendable and can also be used with QUIC.

cs.NI↗

arXiv2019

A Performance Perspective on Web Optimized Protocol Stacks: TCP+TLS+HTTP/2 vs. QUIC

Existing performance comparisons of QUIC and TCP compared an optimized QUIC to an unoptimized TCP stack. By neglecting available TCP improvements inherently included in QUIC, comparisons do not shed light on the performance of current web stacks. In this paper, we can show that tuning TCP parameters is not negligible and directly yields significant improvements. Nevertheless, QUIC still outperforms even our tuned variant of TCP. This performance advantage is mostly caused by QUIC's reduced RTT design during connection establishment, and, in case of lossy networks due to its ability to circumvent head-of-line blocking.

cs.NI↗

arXiv2019

Blitz-starting QUIC Connections

In this paper, we revisit the idea to remove Slow Start from congestion control. To do so, we build upon the newly gained freedom of transport protocol extendability offered by QUIC to hint bandwidth estimates from a typical web client to a server. Using this bandwidth estimate, we bootstrap congestion windows of new connections to quickly utilize available bandwidth. This custom flow initialization removes the common early exit of Slow Start and thus fuels short flow fairness with long-running connections. Our results indicate that we can drastically reduce flow completion time accepting some losses and thereby an inflated transmission volume. For example, for a typical DSL client, loading a 2 MB YouTube video chunk is accelerated by nearly 2x. In the worst case, we find an inflation of the transfer volume by 12% due to losses.

cs.NI↗

arXiv2019

Application-Agnostic Offloading of Packet Processing

As network speed increases, servers struggle to serve all requests directed at them. This challenge is rooted in a partitioned data path where the split between the kernel space networking stack and user space applications induces overheads. To address this challenge, we propose Santa, a new architecture to optimize the data path by enabling server applications to partially offload packet processing to a generic rule processor. We exemplify Santa by showing how it can drastically accelerate kernel-based packet processing - a currently neglected domain. Our evaluation of a broad class of applications, namely DNS, Memcached, and HTTP, highlights that Santa can substantially improve the server performance by a factor of 5.5, 2.1, and 2.5, respectively.

cs.NI↗

arXiv2019

The Dagstuhl Beginners Guide to Reproducibility for Experimental Networking Research

Reproducibility is one of the key characteristics of good science, but hard to achieve for experimental disciplines like Internet measurements and networked systems. This guide provides advice to researchers, particularly those new to the field, on designing experiments so that their work is more likely to be reproducible and to serve as a foundation for follow-on work by others.

cs.NI↗

arXiv2018

Interoperability-Guided Testing of QUIC Implementations using Symbolic Execution

The main reason for the standardization of network protocols, like QUIC, is to ensure interoperability between implementations, which poses a challenging task. Manual tests are currently used to test the different existing implementations for interoperability, but given the complex nature of network protocols, it is hard to cover all possible edge cases. State-of-the-art automated software testing techniques, such as Symbolic Execution (SymEx), have proven themselves capable of analyzing complex real-world software and finding hard to detect bugs. We present a SymEx-based method for finding interoperability issues in QUIC implementations, and explore its merit in a case study that analyzes the interoperability of picoquic and QUANT. We find that, while SymEx is able to analyze deep interactions between different implementations and uncovers several bugs, in order to enable efficient interoperability testing, implementations need to provide additional information about their current protocol state.

cs.SE↗

arXiv2018

Is the Web ready for HTTP/2 Server Push?

HTTP/2 supersedes HTTP/1.1 to tackle the performance challenges of the modern Web. A highly anticipated feature is Server Push, enabling servers to send data without explicit client requests, thus potentially saving time. Although guidelines on how to use Server Push emerged, measurements have shown that it can easily be used in a suboptimal way and hurt instead of improving performance. We thus tackle the question if the current Web can make better use of Server Push. First, we enable real-world websites to be replayed in a testbed to study the effects of different Server Push strategies. Using this, we next revisit proposed guidelines to grasp their performance impact. Finally, based on our results, we propose a novel strategy using an alternative server scheduler that enables to interleave resources. This improves the visual progress for some websites, with minor modifications to the deployment. Still, our results highlight the limits of Server Push: a deep understanding of web engineering is required to make optimal use of it, and not every site will benefit.

cs.NI↗

arXiv2016

The SensorCloud Protocol: Securely Outsourcing Sensor Data to the Cloud

The increasing deployment of sensor networks, ranging from home networks to industrial automation, leads to a similarly growing demand for storing and processing the collected sensor data. To satisfy this demand, the most promising approach to date is the utilization of the dynamically scalable, on-demand resources made available via the cloud computing paradigm. However, prevalent security and privacy concerns are a huge obstacle for the outsourcing of sensor data to the cloud. Hence, sensor data needs to be secured properly before it can be outsourced to the cloud. When securing the outsourcing of sensor data to the cloud, one important challenge lies in the representation of sensor data and the choice of security measures applied to it. In this paper, we present the SensorCloud protocol, which enables the representation of sensor data and actuator commands using JSON as well as the encoding of the object security mechanisms applied to a given sensor data item. Notably, we solely utilize mechanisms that have been or currently are in the process of being standardized at the IETF to aid the wide applicability of our approach.

cs.NI↗

arXiv2016

Finite Blocklength Performance of Multi-Terminal Wireless Industrial Networks

This work focuses on the performance of multi-terminal wireless industrial networks, where the transmissions of all terminals are required to be scheduled within a tight deadline. The transmissions thus share a fixed amount of resources, i.e., symbols, while facing short blocklengths due to the low-latency requirement. We investigate two distinct relaying strategies, namely best relay selection among the participating terminals and best antenna selection at the access point of the network. In both schemes, we incorporate the cost of acquiring instantaneous Channel State Information (CSI) at the access point within the transmission deadline. An error probability model is developed under the finite blocklength regime to provide accurate performance results. As a reference, this model is compared to the corresponding infinite bocklength error model. Both analytical models are validated by simulation. We show that the average Packet Error Rate (PER) over all terminals is convex in the target error probability at each single link. Moreover, we find that: (i) The reliability behavior is different for the two strategies, while the limiting factors are both finite blocklengths and overhead of acquiring CSI. (ii) With the same order of diversity, best antenna selection is more reliable than best relay selection. (iii) The average PER is increasing in the number of participating terminals unless the terminals also act as relay candidates. In particular, if each participating terminal is a candidate for best relay selection, the PER is convex in the number of terminals.

cs.IT↗

arXiv2015

Privacy in the Internet of Things: Threats and Challenges

The Internet of Things paradigm envisions the pervasive interconnection and cooperation of smart things over the current and future Internet infrastructure. The Internet of Things is, thus, the evolution of the Internet to cover the real-world, enabling many new services that will improve people's everyday lives, spawn new businesses and make buildings, cities and transport smarter. Smart things allow indeed for ubiquitous data collection or tracking, but these useful features are also examples of privacy threats that are already now limiting the success of the Internet of Things vision when not implemented correctly. These threats involve new challenges such as the pervasive privacy-aware management of personal data or methods to control or avoid ubiquitous tracking and profiling. This paper analyzes the privacy issues in the Internet of Things in detail. To this end, we first discuss the evolving features and trends in the Internet of Things with the goal of scrutinizing their privacy implications. Second, we classify and examine privacy threats in this new setting, pointing out the challenges that need to be overcome to ensure that the Internet of Things becomes a reality.

cs.CR↗

arXiv2014

User-driven Privacy Enforcement for Cloud-based Services in the Internet of Things

Internet of Things devices are envisioned to penetrate essentially all aspects of life, including homes and urbanspaces, in use cases such as health care, assisted living, and smart cities. One often proposed solution for dealing with the massive amount of data collected by these devices and offering services on top of them is the federation of the Internet of Things and cloud computing. However, user acceptance of such systems is a critical factor that hinders the adoption of this promising approach due to severe privacy concerns. We present UPECSI, an approach for user-driven privacy enforcement for cloud-based services in the Internet of Things to address this critical factor. UPECSI enables enforcement of all privacy requirements of the user once her sensitive data leaves the border of her network, provides a novel approach for the integration of privacy functionality into the development process of cloud-based services, and offers the user an adaptable and transparent configuration of her privacy requirements. Hence, UPECSI demonstrates an approach for realizing user-accepted cloud services in the Internet of Things.

cs.SE↗

arXiv2014

HotBox: Testing Temperature Effects in Sensor Networks

Low-power wireless networks, especially in outside deployments, are exposed to a wide range of temperatures. The detrimental effect of high temperatures on communication quality is well known. To investigate these influences under controlled conditions, we present HotBox, a solution with the following properties: (1) It allows exposition of sensor motes to a wide range of temperatures with a high degree of accuracy. (2) It supports specifying exact spatial orientation of motes which, if not ensured, interferes with repeatable experiment setups. (3) It is reasonably easy to assemble by following the information (code, PCB schematics, hardware list and crafting instructions) available online, facilitating further use of the platforms by other researchers. After presenting HotBox, we will show its performance and prove its feasibility as evaluation platform by conducting several experiments. These experiments additionally provide further insight into the influence of temperature effects on communication performance in low-power wireless networks.

cs.NI↗

arXiv2014

POSTER: Privacy-preserving Indoor Localization

Upcoming WiFi-based localization systems for indoor environments face a conflict of privacy interests: Server-side localization violates location privacy of the users, while localization on the user's device forces the localization provider to disclose the details of the system, e.g., sophisticated classification models. We show how Secure Two-Party Computation can be used to reconcile privacy interests in a state-of-the-art localization system. Our approach provides strong privacy guarantees for all involved parties, while achieving room-level localization accuracy at reasonable overheads.

cs.CR↗

arXiv2014

Enabling Distributed Simulation of OMNeT++ INET Models

Parallel and distributed simulation have been extensively researched for a long time. Nevertheless, many simulation models are still executed sequentially. We attribute this to the fact that many of those models are simply not capable of being executed in parallel since they violate particular constraints. In this paper, we analyze the INET model suite, which enables network simulation in OMNeT++, with regard to parallelizability. We uncovered several issues preventing parallel execution of INET models. We analyzed those issues and developed solutions allowing INET models to be run in parallel. A case study shows the feasibility of our approach. Though there are parts of the model suite that we didn't investigate yet and the performance can still be improved, the results show parallelization speedup for most configurations. The source code of our implementation is available through our web site at code.comsys.rwth-aachen.de.

cs.NI↗

arXiv2013

Support for Error Tolerance in the Real-Time Transport Protocol

Streaming applications often tolerate bit errors in their received data well. This is contrasted by the enforcement of correctness of the packet headers and payload by network protocols. We investigate a solution for the Real-time Transport Protocol (RTP) that is tolerant to errors by accepting erroneous data. It passes potentially corrupted stream data payloads to the codecs. If errors occur in the header, our solution recovers from these by leveraging the known state and expected header values for each stream. The solution is fully receiver-based and incrementally deployable, and as such requires neither support from the sender nor changes to the RTP specification. Evaluations show that our header error recovery scheme can recover from almost all errors, with virtually no erroneous recoveries, up to bit error rates of about 10%.

cs.NI↗

arXiv2013

SensorCloud: Towards the Interdisciplinary Development of a Trustworthy Platform for Globally Interconnected Sensors and Actuators

Although Cloud Computing promises to lower IT costs and increase users' productivity in everyday life, the unattractive aspect of this new technology is that the user no longer owns all the devices which process personal data. To lower scepticism, the project SensorCloud investigates techniques to understand and compensate these adoption barriers in a scenario consisting of cloud applications that utilize sensors and actuators placed in private places. This work provides an interdisciplinary overview of the social and technical core research challenges for the trustworthy integration of sensor and actuator devices with the Cloud Computing paradigm. Most importantly, these challenges include i) ease of development, ii) security and privacy, and iii) social dimensions of a cloud-based system which integrates into private life. When these challenges are tackled in the development of future cloud systems, the attractiveness of new use cases in a sensor-enabled world will considerably be increased for users who currently do not trust the Cloud.

cs.DC↗