Abstract — In this paper we share our experience in building an ADSL video delivery test-bed. As a result, we are able to measure the impact of important parameters such as the loop length, the presence of background traffic and line protection against transmission errors on video quality. We show that although these access technologies promise a broadband pipe and low delay communication, a careful configuration of line parameters is paramount to achieve optimal access and is needed for the success of multimedia triple play (3P) services. We particularly analyze the delay, jitter, packet loss and bitrate consumption obtained from real ADSL measurements. These results are an important step towards understanding the adequate deployment of such services. As a special case, we study the effect of repetitive noise on SDTV. Index Terms— ADSL system performance, Measurements, Video over ADSL, Noise, SDTV, HDTV, Triple Play services. I. INTRODUCTION any efforts have been made to stream multimedia content through the existing data network infrastructure and more recently over telecommunication networks. Triple Play Services (Voice, Video and Data) are becoming the business portfolio of telecom companies. With it comes the need to find new approaches for the engineering of solutions capable of the cost effective delivery of Voice over IP (VoIP), Standard Definition TV (SDTV) and/or High Definition TV (HDTV) as well as data access. Video delivery over the last mile has been a challenging task due to its high data rate requirement as compression techniques can only help a little. Despite advances in broadband technologies, such as ADSL, there are a number of physical and link layers impairments that may potentially degrade video quality. In such hostile environments, packet loss along with other factors such as delay and jitter, may lead to poor video transmission. The main objective of this work is to conduct a detailed investigation of the impact of some ADSL physical and link level parameters on network metrics when carrying video traffic in a real scenario. There are several video quality of service (QoS) or experience (QoE) studies available in the literature, but these are often conducted using simulations and often using over naive, simplistic and even unrealistic error models. We build a test-bed for an ADSL2+ broadband Manuscript received February 27th, 2009. Glauco E. Gonçalves, Gustavo G. B. Santos, Joseane F. Fidalgo, Judith Kelner and Djamel Sadok are with Federal University of Pernambuco, Recife, PE, Brazil; e-mail: [glauco, gustavogbs, joseane, jk, jamel]@gprt.ufpe.br. Stenio Fernandes is currently with University of Ottawa, Ottawa, Canada. He is on leave of his faculty position of the Federal Center for Education in Technology in Maceio, Alagoas, Brazil; e-mail: stenio.fernandes@ ieee.org. access, using hardware and software similar to those used by 3P telecom operators, in order to identify the impact of noise, loop length and background traffic on video delivery under different scenarios. These scenarios contemplate the variation of impairments such as noise, concurrent background traffic, and the loop distance from the Central Office (CO) to the Customer Premise Equipment (CPE) side. This study is crucial for the successful provision of triple play and particularly video services. The authors hope that their results may serve as guidelines, in addition to the standards´ institutions such as the Broadband Forum [14], for adequate configuration when deploying these new multimedia services on DSL systems. In the rest of the paper, section II presents some theoretical background and discusses important related work. Section III presents our methodology and scenarios, whereas section IV describes the corresponding results. Section V presents summarized guidelines obtained through the experiments presented in this paper and section VI summarizes and draws some conclusions along with proposals for future work. II. BACKGROUND AND RELATED WORK While multimedia applications can tolerate some network impairments, excessive packet loss, delay and jitter yield unacceptably low user perceived quality. Unlike traditional data services that depend on network level quality metrics, video quality also depends on other physical aspects such as the coding process, which introduces perceptual quality degradation. Therefore, it is recommended that additional protection of video frames must be applied to establish an upper bound to packet loss, delay and jitter. In addition, an important impairment in the video streaming process is the noise that can be coupled together with the transmitted signal. Noise can pose a considerable threat to DSL transmission and the analysis of its consequences is a daunting task due to its complex statistical nature and the tangled error mitigation and framing techniques used in DSL systems. From the transport and network layers perspective, there are an overwhelming number of research studies. Last mile video quality relies primarily on the physical setup used and the techniques employed to mitigate the different forms of noise using adaptive modulation and low power margins. In [6], the author evaluated the impact of several types of noise when streaming MPEG2 video over DSL. Nonetheless, the analysis lacked parameters concerned with DSL systems and lower network layers. In [2] the authors presented a numerical study of the impact of Impulse Noise (IN), an unpredictable and often deadly form of noise impairment over copper lines, on ADSL. They showed that, while a combination of coding, On the Use of an ADSL2+ Testbed for Video Quality Assessment Glauco E. Gonçalves, Gustavo G. B. Santos, Joseane F. Fidalgo, Judith Kelner, Djamel H. F. Sadok, Stenio F. de L. Fernandes M This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE ICC 2009 proceedings 978-1-4244-3435-0/09/$25.00 ©2009 IEEE