MultiNext: Measuring Concurrent Multipath Transmissions in an Experimental Facility

MultiNext Project Team

Head	Thomas Zinner, University of Würzburg
	Kurt Tutschku, University of Vienna
	Tanja Zseby, Fraunhofer Fokus
	Yuval Shavitt, University of Tel Aviv

Researchers	Dominik Klein
	Albert Rafetseder
	Carsten Schmoll
	Christian Schwartz

Summary

Today, data transport is achieved mainly by the IP (Internet Protocol) protocol. Its routing feature is used to accomplish scalability in the connection of local subnets and end systems. The IP protocol assumes that the interconnecting nodes (i.e. the routers) and the actual transport resources (i.e. the links) are stable and change only in case of faults. The Future Internet will also consist of interconnected subnets and interconnecting nodes. However, the availability of these resources is expected to be provided for a certain lifetime and their capabilities are highly variable. Networks that are based on the model of temporal leasing of variable resources are referred to as federated networks and technologies that enable the save sharing of such network resources are denoted as Network Virtualization (NV) mechanisms.

A major task when creating federated networks for data transport is the selection of resources. The selection has to consider the temporal availability of the resources. Therefore, a measurement-based scheme for selecting the resources is needed. The scheme should adapt the federated transport network to the variable demand for resources as well as to the currently available resources. In addition, the scheme has to scale sufficiently such that it can be applied in large networks.

The concept of Transport Virtualization (TV) enhances the capabilities of future networks. TV can be considered as an alternative mode of Network Virtualization. While NV typically facilitates the sharing of resources, TV creates virtual network based on the aggregation of resources. The simplest form of TV is achieved by collecting multiple transport resources (even from different virtual networks or providers) and selecting the best resources for exclusive or concurrent use. However, the selection and application of concurrent paths (also known as Concurrent Multipath, CMP) is complex and its impact on the transmission is non-intuitive. Path length diversity of different concurrent paths inevitably introduces out-of-order packet delivery. In order to understand the fundamental behavior of resource selection in TV in CMP environments, it is necessary define, verify, and investigate mathematical models for the analysis of packet re-ordering in CMP.

An architecture to enable concurrent multipath transmissions for TV, the so called SORA architecture, was presented by Lane and Nakao. The use of concurrent multipath (CMP) transmissions will bring exceptional advantages to networks, such as higher throughput and increased resilience. However, CMP will introduce additional complexity which has to be understood. First, CMP transmission will inevitably introduce out-of-order packets due to different stochastic packet delay characteristics on the paths. The re-ordering has to be compensated by buffering at the destination, possibly leading to increased e2e delay. Second, the different stochastic delay processes on the paths can amend each other in their negative effects on out-of-order packets. Third, the strength and occurrences of these combination effects are highly non-intuitive.

The importance of multipath transmissions for the Future Internet is reflected also by current research efforts in the IETF and other FP7 projects. Currently the LEDBAT WG within the IETF investigates multi-path TCP. It should be noted that at the last IETF meeting in Stockholm preparations a BoF on multipath TCP was successfully organised. This BoF builds on results obtained within the TRILOGY FP7 project showing that performance between end hosts can be improved by using different paths in parallel.

The aim of the proposed specific EuroNF project is threefold:

To determine how measurements (e.g. one way delay, throughput, etc.) can be conducted within experimental facilites
To perform measurements on the relevant parameters within the SORA Architecture
To link theoretical mathematical models with the obtained measurements

Phone Conferences

February 5th 11-12am, Kick-off call, organizational matters
March 9th 1pm - 2pm, Use of experimental facilities, measurement tools
April 21st 11-12am, Measurement setup
May 20th 11.30-12.30am, Preparation of Kick-off Meeting
June 16th 11.30-12.30am, Discussion of suitable measurement enhancements
July 22nd 11.30-12.30am, Discussion of the first measurements and how to proceed with the measurements
September 28th 11.30-12.30am, Discussion measurements, discussion of possible submissions / suitable conferences/workshops
November 11th, Discussion FIA Book Chapter

Meetings

Kick-Off meeting, Juni 2nd, 2010, co-located with Euro-NF NGI2010 in Paris

Internal project meeting, August 2nd, 2010, during Euroview 2010 in Wuerzburg

Article[ to top ]

Quality of Experience Management for YouTube: Clouds, FoG and the AquareYoum. Hoßfeld, Tobias; Liers, Florian; Schatz, Raimund; Staehle, Barbara; Staehle, Dirk; Volkert, Thomas; Wamser, Florian. In PIK - Praxis der Informationverarbeitung und -kommunikation (PIK). 2012.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ Download ]
Over the last decade, Quality of Experience (QoE) has become a new, central paradigm for understanding the quality of networks and services. In particular, the concept has attracted the interest of communication network and service providers, since being able to guarantee good QoE to customers provides an opportunity for differentiation. In this paper we investigate the potential as well as the implementation challenges of QoE management in the Internet. Using YouTube video streaming service as example, we discuss the different elements that are required for the realization of the paradigm-shift towards truly user-centric network orchestration. To this end, we elaborate QoE management requirements for two complementary network scenarios (wireless mesh Internet access networks vs. global Internet delivery) and provide a QoE model for YouTube taking into account impairments like stalling and initial delay. We present two YouTube QoE monitoring approaches operating on the network and the end user level. Finally, we demonstrate how QoE can be dynamically optimized in both network scenarios with two exemplary concepts, AquareYoum and FoG, respectively. Our results show how QoE management can truly improve the user experience while at the same time increase the efficiency of network resource allocation.

@article{info3-article-2012-103, abstract = {Over the last decade, Quality of Experience (QoE) has become a new, central paradigm for understanding the quality of networks and services. In particular, the concept has attracted the interest of communication network and service providers, since being able to guarantee good QoE to customers provides an opportunity for differentiation. In this paper we investigate the potential as well as the implementation challenges of QoE management in the Internet. Using YouTube video streaming service as example, we discuss the different elements that are required for the realization of the paradigm-shift towards truly user-centric network orchestration. To this end, we elaborate QoE management requirements for two complementary network scenarios (wireless mesh Internet access networks vs. global Internet delivery) and provide a QoE model for YouTube taking into account impairments like stalling and initial delay. We present two YouTube QoE monitoring approaches operating on the network and the end user level. Finally, we demonstrate how QoE can be dynamically optimized in both network scenarios with two exemplary concepts, AquareYoum and FoG, respectively. Our results show how QoE management can truly improve the user experience while at the same time increase the efficiency of network resource allocation.}, author = {Hoßfeld, Tobias and Liers, Florian and Schatz, Raimund and Staehle, Barbara and Staehle, Dirk and Volkert, Thomas and Wamser, Florian}, journal = {PIK - Praxis der Informationverarbeitung und -kommunikation (PIK)}, keywords = {sys:hidden:article}, month = {08}, title = {Quality of Experience Management for YouTube: Clouds, FoG and the AquareYoum}, year = 2012 }

%0 Journal Article %1 info3-article-2012-103 %A Hoßfeld, Tobias %A Liers, Florian %A Schatz, Raimund %A Staehle, Barbara %A Staehle, Dirk %A Volkert, Thomas %A Wamser, Florian %D 2012 %J PIK - Praxis der Informationverarbeitung und -kommunikation (PIK) %T Quality of Experience Management for YouTube: Clouds, FoG and the AquareYoum %X Over the last decade, Quality of Experience (QoE) has become a new, central paradigm for understanding the quality of networks and services. In particular, the concept has attracted the interest of communication network and service providers, since being able to guarantee good QoE to customers provides an opportunity for differentiation. In this paper we investigate the potential as well as the implementation challenges of QoE management in the Internet. Using YouTube video streaming service as example, we discuss the different elements that are required for the realization of the paradigm-shift towards truly user-centric network orchestration. To this end, we elaborate QoE management requirements for two complementary network scenarios (wireless mesh Internet access networks vs. global Internet delivery) and provide a QoE model for YouTube taking into account impairments like stalling and initial delay. We present two YouTube QoE monitoring approaches operating on the network and the end user level. Finally, we demonstrate how QoE can be dynamically optimized in both network scenarios with two exemplary concepts, AquareYoum and FoG, respectively. Our results show how QoE management can truly improve the user experience while at the same time increase the efficiency of network resource allocation.

Inproceedings[ to top ]

Waiting Times in Quality of Experience for Web Based Services. Egger, Sebastian; Hoßfeld, Tobias; Schatz, Raimund; Fiedler, Markus. In QoMEX 2012. Yarra Valley, Australia, 2012.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ Download ]
A considerable share of applications such as web or e-mail browsing, online picture viewing and file downloads imply waiting times for their users, which is due to the turn-taking of information requests by the user and correspoding response times until each request is fulfilled. Thus, end-user quality perception in the context of interactive data services is dominated by waiting times; the longer the latter, the less satisfied the user becomes. As opposed to heavily researched multimedia experience, perception of waiting times is still not strongly explored in the context of Quality of Experience (QoE). This tutorial will contribute to closing this gap. In its first part, it addresses perception principles and discusses their applicability towards fundamental relationships between waiting times and resulting QoE. It then investigates to which extent the same relationships can also be used to describe QoE for more complex services such as web browsing. Finally, it discusses applications where waiting times determine QoE, amongst other factors. For example, the past shift from UDP media streaming to TCP media streaming (e.g. youtube.com) has extended the relevance of waiting times also to the domain of online video services. In particular, user-perceived quality suffers from initial delays when applications are launched, as well as from freezes during the delivery of the stream. These aspects, which have to be traded against each other to some extent, will be discussed mainly for HTTP video streaming in the last part of this tutorial.

@inproceedings{info3-inproceedings-2012-452, abstract = {A considerable share of applications such as web or e-mail browsing, online picture viewing and file downloads imply waiting times for their users, which is due to the turn-taking of information requests by the user and correspoding response times until each request is fulfilled. Thus, end-user quality perception in the context of interactive data services is dominated by waiting times; the longer the latter, the less satisfied the user becomes. As opposed to heavily researched multimedia experience, perception of waiting times is still not strongly explored in the context of Quality of Experience (QoE). This tutorial will contribute to closing this gap. In its first part, it addresses perception principles and discusses their applicability towards fundamental relationships between waiting times and resulting QoE. It then investigates to which extent the same relationships can also be used to describe QoE for more complex services such as web browsing. Finally, it discusses applications where waiting times determine QoE, amongst other factors. For example, the past shift from UDP media streaming to TCP media streaming (e.g. youtube.com) has extended the relevance of waiting times also to the domain of online video services. In particular, user-perceived quality suffers from initial delays when applications are launched, as well as from freezes during the delivery of the stream. These aspects, which have to be traded against each other to some extent, will be discussed mainly for HTTP video streaming in the last part of this tutorial.}, address = {Yarra Valley, Australia}, author = {Egger, Sebastian and Hoßfeld, Tobias and Schatz, Raimund and Fiedler, Markus}, booktitle = {QoMEX 2012}, keywords = {sys:hidden:inproceedings}, month = {07}, title = {Waiting Times in Quality of Experience for Web Based Services}, year = 2012 }

%0 Conference Paper %1 info3-inproceedings-2012-452 %A Egger, Sebastian %A Hoßfeld, Tobias %A Schatz, Raimund %A Fiedler, Markus %B QoMEX 2012 %C Yarra Valley, Australia %D 2012 %T Waiting Times in Quality of Experience for Web Based Services %X A considerable share of applications such as web or e-mail browsing, online picture viewing and file downloads imply waiting times for their users, which is due to the turn-taking of information requests by the user and correspoding response times until each request is fulfilled. Thus, end-user quality perception in the context of interactive data services is dominated by waiting times; the longer the latter, the less satisfied the user becomes. As opposed to heavily researched multimedia experience, perception of waiting times is still not strongly explored in the context of Quality of Experience (QoE). This tutorial will contribute to closing this gap. In its first part, it addresses perception principles and discusses their applicability towards fundamental relationships between waiting times and resulting QoE. It then investigates to which extent the same relationships can also be used to describe QoE for more complex services such as web browsing. Finally, it discusses applications where waiting times determine QoE, amongst other factors. For example, the past shift from UDP media streaming to TCP media streaming (e.g. youtube.com) has extended the relevance of waiting times also to the domain of online video services. In particular, user-perceived quality suffers from initial delays when applications are launched, as well as from freezes during the delivery of the stream. These aspects, which have to be traded against each other to some extent, will be discussed mainly for HTTP video streaming in the last part of this tutorial.
Initial Delay vs. Interruptions: Between the Devil and the Deep Blue Sea. Hoßfeld, Tobias; Egger, Sebastian; Schatz, Raimund; Fiedler, Markus; Masuch, Kathrin; Lorentzen, Charlott. In QoMEX 2012. Yarra Valley, Australia, 2012.
- [ Abstract ]
- [ BibTeX ]
- [ EndNote ]
- [ Download ]
End user quality perception in the context of Internet applications and services is often characterized by waiting times before service consumption as well as interruptions during service consumption. In particular in case of bad network conditions, network and service providers have to trade off between these two impairment types, i.e. between the devil and the deep blue sea. In this paper we investigate this trade-off in order to guide the design and development of Internet applications and network management approaches. The contribution of this paper is twofold. Firstly, we quantify the impact of initial delays on the user perceived Quality of Experience (QoE) for different application scenarios by means of subjective laboratory and crowdsourcing studies. We show that QoE for a given waiting time strongly depends on the concrete application at hand but that rating diversity remains fairly application-invariant. Secondly, using the example of YouTube video streaming we compare the influence of initial delays and interruptions (stallings) during watching. Our results demonstrate that users are extremely sensitive to interruptions and that services should be designed accordingly e.g. by increasing initial delay for prebuffering to overcome lack of resources.

@inproceedings{info3-inproceedings-2012-444, abstract = {End user quality perception in the context of Internet applications and services is often characterized by waiting times before service consumption as well as interruptions during service consumption. In particular in case of bad network conditions, network and service providers have to trade off between these two impairment types, i.e. between the devil and the deep blue sea. In this paper we investigate this trade-off in order to guide the design and development of Internet applications and network management approaches. The contribution of this paper is twofold. Firstly, we quantify the impact of initial delays on the user perceived Quality of Experience (QoE) for different application scenarios by means of subjective laboratory and crowdsourcing studies. We show that QoE for a given waiting time strongly depends on the concrete application at hand but that rating diversity remains fairly application-invariant. Secondly, using the example of YouTube video streaming we compare the influence of initial delays and interruptions (stallings) during watching. Our results demonstrate that users are extremely sensitive to interruptions and that services should be designed accordingly e.g. by increasing initial delay for prebuffering to overcome lack of resources.}, address = {Yarra Valley, Australia}, author = {Hoßfeld, Tobias and Egger, Sebastian and Schatz, Raimund and Fiedler, Markus and Masuch, Kathrin and Lorentzen, Charlott}, booktitle = {QoMEX 2012}, keywords = {sys:hidden:inproceedings}, month = {07}, title = {Initial Delay vs. Interruptions: Between the Devil and the Deep Blue Sea}, year = 2012 }

%0 Conference Paper %1 info3-inproceedings-2012-444 %A Hoßfeld, Tobias %A Egger, Sebastian %A Schatz, Raimund %A Fiedler, Markus %A Masuch, Kathrin %A Lorentzen, Charlott %B QoMEX 2012 %C Yarra Valley, Australia %D 2012 %T Initial Delay vs. Interruptions: Between the Devil and the Deep Blue Sea %X End user quality perception in the context of Internet applications and services is often characterized by waiting times before service consumption as well as interruptions during service consumption. In particular in case of bad network conditions, network and service providers have to trade off between these two impairment types, i.e. between the devil and the deep blue sea. In this paper we investigate this trade-off in order to guide the design and development of Internet applications and network management approaches. The contribution of this paper is twofold. Firstly, we quantify the impact of initial delays on the user perceived Quality of Experience (QoE) for different application scenarios by means of subjective laboratory and crowdsourcing studies. We show that QoE for a given waiting time strongly depends on the concrete application at hand but that rating diversity remains fairly application-invariant. Secondly, using the example of YouTube video streaming we compare the influence of initial delays and interruptions (stallings) during watching. Our results demonstrate that users are extremely sensitive to interruptions and that services should be designed accordingly e.g. by increasing initial delay for prebuffering to overcome lack of resources.
Prospects for Realizing User-Centric Network Orchestration: FEC-protected SVC Streaming. Zinner, Thomas; Liers, Florian; Hoßfeld, Tobias; Rauscher, Dirk; Reuther, Bernd; Günther, Daniel; Volkert, Thomas; Fiedler, Markus. In 7. GI/ITG KuVS Fachgespräch ’Future Internet’. München, 2012.
- [ BibTeX ]
- [ EndNote ]
- [ Download ]
@inproceedings{info3-inproceedings-2012-436, address = {München}, author = {Zinner, Thomas and Liers, Florian and Hoßfeld, Tobias and Rauscher, Dirk and Reuther, Bernd and Günther, Daniel and Volkert, Thomas and Fiedler, Markus}, booktitle = {7. GI/ITG KuVS Fachgespräch 'Future Internet'}, keywords = {sys:hidden:inproceedings}, month = {01}, title = {Prospects for Realizing User-Centric Network Orchestration: FEC-protected SVC Streaming}, year = 2012 }

%0 Conference Paper %1 info3-inproceedings-2012-436 %A Zinner, Thomas %A Liers, Florian %A Hoßfeld, Tobias %A Rauscher, Dirk %A Reuther, Bernd %A Günther, Daniel %A Volkert, Thomas %A Fiedler, Markus %B 7. GI/ITG KuVS Fachgespräch 'Future Internet' %C München %D 2012 %T Prospects for Realizing User-Centric Network Orchestration: FEC-protected SVC Streaming

Misc[ to top ]

Waiting Times in Quality of Experience for Web Based Services. Egger, Sebastian; Hoßfeld, Tobias; Schatz, Raimund; Fiedler, Markus. Yarra Valley, Australia, 2012, July.
- [ BibTeX ]
- [ EndNote ]
@misc{info3-tutorial-2012-15, address = {Yarra Valley, Australia}, author = {Egger, Sebastian and Hoßfeld, Tobias and Schatz, Raimund and Fiedler, Markus}, booktitle = {QoMEX 2012}, keywords = {sys:hidden:tutorial}, month = {07}, title = {Waiting Times in Quality of Experience for Web Based Services}, year = 2012 }

%0 Generic %1 info3-tutorial-2012-15 %A Egger, Sebastian %A Hoßfeld, Tobias %A Schatz, Raimund %A Fiedler, Markus %B QoMEX 2012 %C Yarra Valley, Australia %D 2012 %T Waiting Times in Quality of Experience for Web Based Services