Bjørn Jonny Villa

Group Based Traffic Shaping for Adaptive HTTP Video Streaming by Segment Duration Control

Adaptive video streaming is based on the concept of allowing the quality level of a video stream to change during its lifetime based on certain parameters. This approach makes the video stream more tolerant with respect to fluctuations in available bandwidth during the session lifetime. In order for a video stream to decide which quality level to request from the server, it is typical to use both parameters from the client side such as CPU load and estimated parameters from the network side such as available bandwidth. In order for available bandwidth estimations to become more accurate, it is beneficial with some degree of traffic shaping. This paper describes a new method of achieving a traffic shaping effect for adaptive video streaming sessions delivered to members of a specific user group identified by their unique network destination. The novelty of the method is represented by the use of segment size control per video session, rather than active shaping of each session. The effect of the method is analyzed by means of measuring packet IAT and video segment fetch durations, in a controlled lab environment using the Smooth Streaming framework from Microsoft.

Improving Fairness for Adaptive HTTP Video Streaming

This paper presents an analysis of a suggested method for improving fairness among competing adaptive HTTP video streams. The metrics used as fairness indicators are differences in achieved average rate and stability among the competing streams. The method analyzed is based on changing a fixed and equal video segment request rate of each stream, to either a per session unique or random request rate. The analysis is done by means of measurements in a controlled environment using the Microsoft Smooth Streaming solution. The findings are considered very positive as they show that it is possible to achieve a significant improvement in fairness by applying the suggested method. The main novelty of the paper is that it demonstrates the potential of achieving such improvements without modifying either client or server algorithms.

Detecting Period and Burst Durations in Video Streaming by means of Active Probing

The growth of broadband Internet access across the world for the last 10 years has made new business models possible. Services which before mainly were provided by network operators and in dedicated networks have now migrated to the open Internet. This has created a global service provider market, using the Internet as platform. The network operators are in many cases left with only providing the broadband access service. For the global service providers, continuous effort is put into the field of finding smart methods for bringing new and advanced services to the market without asking for e.g. QoS features from the involved network operators. This type of service delivery is called Over-The-Top (OTT). The concept of making services able to adapt their network and transport requirements during time of delivery is a strong contribution to success for OTT services.The focus of the work in this thesis, is methods for improving various aspects - as defined by my research questions - related to QoS and potentially also QoE for dynamic adaptive video streaming over HTTP (DASH) services. The motivation for focusing on video services is based on their high QoS requirements (i.e. bandwidth) and also popularity in terms of usage.In the work presented in this thesis, I have studied the behaviour and performance of DASH services by means of simulations, measurements and experiments. Based on insight obtained through this, I established a hypothesis on how QoS and QoE aspects could be improved for users present in the same home network environment. My hypothesis was that making more accurate information available about both services and network conditions in near real-time could facilitate improved control methods. The effectiveness of my suggested methods have in most cases been analysed by means of implementation in an experimental lab scenario, and supported by simulations and analytical approaches when appropriate.The findings presented in the included papers are all closely related and map into the research model used. This model is composed of Knowledge Plane, Monitor Plane and Action plane components located in both service endpoints and involved network components. As my main focus has been on OTT service delivery, the main contributions of my work apply to service endpoints, i.e. components in the home network and on the server side. The server side would in many cases be represented by a Content Delivery Network (CDN) node.The main research questions identified are:RQ1: In a home network how to (autonomously) control the performance of DASH based services.RQ2: How to provide fairness and stability for competing DASH sessions using service endpoint functionality.RQ3: How to choose appropriate quality levels for a DASH session.The main contributions from my research are:C1: A method for controlling the quality levels of DASH in the home gateway.C2: A method for improving fairness among competing DASH sessions.C3: A method for shaping traffic aggregates on access links with DASH components.C4: A method for estimating available bandwidth on access links when DASH sessions are present.

A Measurement Study of Active Probing on Access Links

This paper presents a measurement study of two different methods for active probing of cross traffic on access links. The categories used in the study are packet pair probing and one-way-delay probing. The first approach uses measured increase in packet spacing as indicator of cross traffic presence, while the latter uses increase in one-way-delay for probe packets as indicator. These methods have been chosen because they are fundamentally different in terms of requirements, benefits and challenges. The main novelty of this paper is the presentation and discussion of measurement results from an access network using an adaptive video service as cross-traffic. The findings clearly illustrate the potential strengths of a probing method based on one-way-delay measurements, under the condition that the required timing accuracy is achieved and delay characteristics is available for the involved network path. The benefit of using Precision Time Protocol instead of Network Time Protocol is illustrated, even in networks of limited size.