Juan M. Lopez-Soler

Analysis and modelling of YouTube traffic

YouTube currently accounts for a significant percentage of the Internets global traffic. Hence, understanding the characteristics of the YouTube traffic generation pattern can provide a significant advantage in predicting user video quality and in enhancing network design. In this paper, we present a characterisation of the traffic generated by YouTube when accessed from a regular PC. On the basis of this characterisation, a YouTube server traffic generation model is proposed, which, for example, can be easily implemented in simulation tools. The derived characterisation and model are based on experimental evaluations of traffic generated by the application layer of YouTube servers. A YouTube server commences the download with an initial burst and later throttles down the generation rate. If the available bandwidth is reduced (e.g. in the presence of network congestion), the server behaves as if the data excess that cannot be transmitted because of the reduced bandwidth were accumulated at a servers buffer, which is later drained if the bandwidth availability is recovered. As we will show, the video clip encoding rate plays a relevant role in determining the traffic generation rate, and therefore, a cumulative density function for the most viewed video clips will be presented. The proposed traffic generation model was implemented in a YouTube emulation server, and the generated synthetic traffic traces were compared with downloads from the original YouTube server. The results show that the relative error between downloads from the emulation server and the original server does not exceed 6% for the 90% of the considered videos. Copyright

QoE oriented cross-layer design of a resource allocation algorithm in beyond 3G systems

The provision of high speed access to Internet and IP-based services is one of the main goals of beyond 3G (B3G) wireless systems. These systems will benefit from cross-layer protocol designs that will introduce interactions between different layers to obtain performance gains. The majority of the research in the field of cross-layer in B3G systems aims at improving quality of service (QoS) system centric metrics such as spectral efficiency, service latency, delay variation (jitter), etc. However, minor attention has been paid to the satisfaction of the subjective quality requirements from human users. With the goal of incorporating the subjective human perception into the cross-layer design of B3G systems, this work carries out an experimental survey of the sensitivity of the user subjective quality to the service response time for the Web browsing application. From the experimental results, a mapping from service response time and user data rate (provided by the wireless link) to mean opinion score (MOS) is derived. The presented results will show that the Web page size plays an important role in the mapping function. The derived mapping function is incorporated into a radio resource allocation algorithm for orthogonal frequency division multiple access (OFDMA) systems. This incorporation is carried out maximizing the aggregate utility over all the users in the cell. Its performance has been compared to that of the multicarrier proportional fair (MPF) under heavy load conditions with a 3G LTE simulator. The results have shown that the proposed methodology can provide an interesting enhancement of the user experienced quality compared to the MPF algorithm.

DARGOS: A highly adaptable and scalable monitoring architecture for multi-tenant Clouds

One of the most important features in Cloud environments is to know the status and the availability of the physical resources and services present in the current infrastructure. A full knowledge and control of the current status of those resources enables Cloud administrators to design better Cloud provisioning strategies and to avoid SLA violations. However, it is not easy to manage such information in a reliable and scalable way, especially when we consider Cloud environments used and shared by several tenants and when we need to harmonize their different monitoring needs at different Cloud software stack layers. To cope with these issues, we propose Distributed Architecture for Resource manaGement and mOnitoring in cloudS (DARGOS), a completely distributed and highly efficient Cloud monitoring architecture to disseminate resource monitoring information. DARGOS ensures an accurate measurement of physical and virtual resources in the Cloud keeping at the same time a low overhead. In addition, DARGOS is flexible and adaptable and allows defining and monitoring new metrics easily. The proposed monitoring architecture and related tools have been integrated into a real Cloud deployment based on the OpenStack platform: they are openly available for the research community and include a Web-based customizable Cloud monitoring console. We report experimental results to assess our architecture and quantitatively compare it with a selection of other Cloud monitoring systems similar to ours showing that DARGOS introduces a very limited and scalable monitoring overhead.

Characteristics of mobile youtube traffic

In this work we present the characterization of the mobile network traffic generated by one of the most relevant social networking applications: YouTube. Understanding its characteristics is of major importance to evaluate its impact on mobile networks and optimize network or application design. For this purpose, we have performed a set of experiments capturing traffic from Apple iOS and Android terminals connected to 3G mobile networks. Our results show that the video formats are 3GP formats, which use lower encoding rates than in other networks. As in accesses from other networks, YouTube servers transfer the video clip with an initial burst and a throttling phase, but they apply different parameters for clients located in mobile networks. Additionally, the mobile terminal can strongly influence the download. Our results with high-end Android terminals show that the client implements a dual-threshold buffer policy that interrupts and resumes the download depending on its buffer occupancy. Our results with iOS terminals show that the video clip download is not interrupted by the client, and a large amount of data may be accumulated at the player buffer. In mid-range Android terminals the TCP receive window may additionally throttle the download if the player buffer is filled up.

Traffic Models Impact on OFDMA Scheduling Design

This article studies the impact on the design of scheduling algorithms for Orthogonal Frequency Division Multiple Access (OFDMA) systems of two traffic models described in the evaluation methodology proposals from standardization bodies: the full buffer and the finite buffer traffic models. The analysis concentrates on utility-based scheduling with an α-fair utility function for Non-Real Time (NRT) services. The results show that a gradient scheduling algorithm is able to maximize the aggregate utility over all the users when the less realistic full buffer model is adopted; but not when the finite buffer model is applied. The results also show that with the full buffer model a gradient scheduler exhibits a trade-off between average user throughput and the user throughput at 5% outage, but it does not when the more realistic finite buffer is used. Therefore, it is concluded that designs of scheduling algorithms for NRT services for OFDMA systems carried out under the full buffer model assumption may fail to provide the desired performance benefits in realistic scenarios. Based on the results presented, a recommendation on scheduling design is provided.

A QoE-Aware Scheduler for HTTP Progressive Video in OFDMA Systems

HTTP progressive video downloading is becoming increasingly relevant in wireless and mobile networks. Therefore, improving the quality of experience (QoE) for this service is important for customer satisfaction. In this letter we propose a QoE-aware scheduling solution for the HTTP progressive video downloading in OFDMA systems. Whereas most scheduling strategies consider QoS metrics such as delay, jitter or throughput, our proposal takes into account a metric that directly affects the end-users experience. In particular, we propose to use an estimation of the amount of video data stored in the player buffer as the main criterion for resource assignment. The ultimate goal of using this criterion is the avoidance of playback interruptions. In particular, we have evaluated this scheduler in an LTE network simulator for the YouTube service, which is the most visited HTTP video streaming site. Our results show a significant improvement compared to the Multicarrier Proportional Fair scheduler, e.g. the number of videos with pauses decrease from 16.1% to 0% for 12 Mbps of cell offered load.

A content-aware bridging service for publish/subscribe environments

The OMG DDS (Data Distribution Service) standard specifies a middleware for distributing real-time data using a publish-subscribe data-centric approach. Until now, DDS systems have been restricted to a single and isolated DDS domain, normally deployed within a single multicast-enabled LAN. As systems grow larger, the need to interconnect different DDS domains arises. In this paper, we consider the problem of communicating disjoint data-spaces that may use different schemas to refer to similar information. In this regard, we propose a DDS interconnection service capable of bridging DDS domains as well as adapting between different data schemas. A key benefit of our approach is that is compliant with the latest OMG specifications, thus the proposed service does not require any modifications to DDS applications. The paper identifies the requirements for DDS data-spaces interconnection, presents an architecture that responds to those requirements, and concludes with experimental results gathered on our prototype implementation. We show that the impact of the service on the communications performance is well within the acceptable limits for most real-world uses of DDS (latency overhead is of the order of hundreds of microseconds). Reported results also indicate that our service interconnects remote data-spaces efficiently and reduces the network traffic almost N times, with N being the number of final data subscribers.

Bloom Filter Based Discovery Protocol for DDS Middleware

Abstract The Data Distribution Service (DDS) middleware has recently been standardized by the OMG. Prior to data communication, a discovery protocol had to locate and obtain remote DDS entities and their attributes. Specifically, DDS discovery matches the DataWriters (DWs) and DataReaders (DRs) entities ( Endpoints ) situated in different network nodes. DDS specification does not specify how this discovery is translated “into the wire”. To provide interoperability and transparency between different DDS implementations, the OMG has standardized the DDS Interoperability Wire Protocol (DDS-RTPS). Any compliant DDS-RTPS implementation must support at least the SDP (Simple Discovery Protocol). The SDP works in relatively small or medium networks but it may not scale as the number of DDS Endpoints increases. This paper addresses the design and evaluation of an SDP alternative–which uses Bloom Filters (BF)–that increases DDS scalability. BFs use Hash functions for space-efficient probabilistic data set representation. We provide both analytical and experimental studies. Results show that our approach can improve the discovery process (in terms of network load and node resource consumption), especially in those scenarios with large Endpoint per Participant ratios.

DDS-enabled Cloud management support for fast task offloading

Cloud computing has become an essential technology not only for web provisioning, but also in mobile scenarios. Mobile devices are usually resource constrained due to processing and power limitations, so typical applications are not easy portable. Battery draining and application performance (resource shortage) have a big impact on the experienced quality, so shifting applications and services to the Cloud may improve mobile users satisfaction. However, available Cloud solutions are mostly focused on scenarios with slowly changing provisioning, which are unable to support and promptly react to short-term provisioning requests. To address the new scenario, this paper proposes a novel Cloud monitoring and management architecture based on the data-centric publish-subscribe Data Distribution Service (DDS) standard. We present not only an architecture proposal, but also a real prototype that we have deployed in our experimental testbed. The experimental results show that our architecture is able to support the scheduling of highly dynamic tasks in the Cloud while maintaining low overheads.

Video Tester — A multiple-metric framework for video quality assessment over IP networks

This paper presents an extensible and reusable framework which addresses the problem of video quality assessment over IP networks. The proposed tool (referred to as Video-Tester) supports raw uncompressed video encoding and decoding. It also includes different video over IP transmission methods (i.e.: RTP over UDP unicast and multicast, as well as RTP over TCP). In addition, it is furnished with a rich set of offline analysis capabilities. Video-Tester analysis includes QoS and bitstream parameters estimation (i.e.: bandwidth, packet inter-arrival time, jitter and loss rate, as well as GOP size and I-frame loss rate). Our design facilitates the integration of virtually any existing video quality metric thanks to the adopted Python-based modular approach. Video-Tester currently provides PSNR, SSIM, ITU-T G.1070 video quality metric, DIV and PSNR-based MOS estimations. In order to promote its use and extension, Video-Tester is open and publicly available.