FTTP, FTTH, perhaps from 10Mbps to 100 Mbps Because of the high speed network, it’s clearly that the IP networks will become a delivery mechanism for broadcast television content
Since FTTP has a high speed, various traffic will transmit over it, such as … Any network link that handles many subscribers, each capable of demanding one or more IP TV video streams, must have enough bandwidth to meet the demand. There are two kinds of definition, SD and HD, The problem comes up, when there are different definition or IPTV traffic mixed together, what’s the actual demand we need? After evaluating the bandwidth demand, we can further analyze the aggregate traffic of data, voice, and video (triple play)
The graph shows a portion of a FTTP network The PON typically serves up to 32 subscribers, while the OLT serves roughly 2000 subscribers. The link from the edge router to the OLT must deliver content to those 2000 subscribers. While not all of them may subscribe to IP TV, we can reasonably expect to find hundreds of IP TV subscribers on an OLT
This paper divided the video stream into two states, one is steady state, the other is unsteady state, which is disrupted by channel surfing. When in steady state, all viewers have settled into channels, then we can use multicast to serve viewers since every viewer stably receives the video stream, and for those who watch the same program, we can only send one stream and split it into more than one substream to achieve reducing the bandwidth demand. When we watch a program, there will be a short time that the program breaks and start broadcasting advertisements, we call it commercial break. Often last for few minutes, maybe one or three minutes. At this moment, viewer often quickly changes the channel from one to another, we call it channel surfing. In this paper, they claim that the additional bandwidth demand is required, I will explain it later.
The graph shows the multicast in steady state scenario  In the steady state, multicast reduces IP TV traffic volume: if two viewers watch the same broadcast program, the network needs to deliver only one video stream to the point (e.g., the OLT) where it must divide the stream in two.
When viewer changes channels, we have to provide fast channel change mechanism  In the paper, the scenario is like the graph, we use a channel change server to provide higher than usual rate to fill the buffer, Then the buffer can play in a constant rate.  One way the network can make channel changes fast is to send surfers unicast (one per viewer) streams at higher than usual rates. While surfing period may be short-lived (say, a minute), each episode superimposes a significant additional demand on top of the steady state demand. Surfing’s recurring nature means that capacity planning and engineering must include surfing’s transient effects.
From above scenarios, we can imagine that the bandwidth demand might be similar to the graph. The vertical coordinate is bandwidth demand while the horizontal is time This is a graph for single viewer, we will combine multiple viewers later. Here comes the issues, what is the bandwidth demand of IPTV subscribers, and how to model the IPTV traffic, Furthermore, I mean in our project, we have to find out what happened when video, voice, and data traffic combined together.
When the paper wants to build a traffic demand model, they have to made some assumptions. They define a viewer is ……………, this is because one subscriber may have several TV or many set top boxes, and each of them plays different channel. Second is ………, because there is still no rigorous approach to know if the user is in front of the tv or not, maybe in the future TV will close automatically in order to save some resources, we will see in the future The paper neglect the other traffics and focus only the IPTV traffic Last one is to assume that when viewer make channel changes, the paper use a kind of stochastic process to model it, which indicate we expect the user behavior follows the same pattern and channel change probability also follows the same pattern
 means there is no relation between program definition and the channel changes …… I think, it is perhaps, maybe, a little strong, it’s not definitely to be all in the same time, but if we don’t make this assumption, the problem will become very complicate
I use a graph to demonstrate the whole architecture of the mathematical model in this paper, Steady state multicast model is a easier one, and very intuitive In the channel surfing model, more complicate one, we need to know the time of filling the playout buffer and the single viewer behavior Afterword, the paper extended both and combine them together, Finally, we get the total bandwidth demand of IPTV traffic
(1) => Zip’s law , by reference 
bs is buffer size in second
In the terminating Poisson process, the times between renewals (channel changes) follow an exponential distribution; the mean time between renewals is 4 seconds. At each renewal and at time 0, the viewer tosses a coin to decide whether to change the channel. We choose the mean number of channel changes to be 10, which makes the probability of heads in thecoin toss 0.91. Note that rCH = 2.5rH implies rCH = 37.5 Mbps, which is high by today’s broadband standards. If use instead rCH = 1.67rH , we obtain a more tractable rCH = 25 Mbps. However, the results described below change hardly at all. Fig. 2 shows the mean surfing demand E(BU (t)) (labeled “Surfing”), the mean multicast demand E(BM (t)) (labeled “Mcast”), and and the mean total demand E(B(t)) (labeled “Total”) over 100 seconds. The multicast demand increases as time progresses and surfers quit surfing and settle down into multicast groups. The Surfing curve decreases as viewers stop surfing. The mean total bandwidth decays asymptotically to about 1000 Mbps as a result of multicast savings. Fig. 2 makes clear the implications of surfing and multicast. As a reference point, if 400 viewers were watching unicast streams with average bandwidth per stream pSDrS + pHDrH = 4.875 Mbps, the mean total demand would be 1950 Mbps. The asymptotic demand of about 1000 Mbps in Fig. 2 shows that multicast cuts the demand in half. On the other hand, the maximum in Fig. 2 reaches almost to 1700 Mbps. Channel surfing gives back nearly all multicast’s gains.
INFOCOM, 2007 Chen Bin Kuo (20077202) Young J. Won (20063292) DPNM Lab.
<ul><li>Introduction </li></ul><ul><li>Problem Description and Assumptions </li></ul><ul><li>Mathematical Model </li></ul><ul><li>Simulation Result </li></ul><ul><li>Conclusions </li></ul><ul><li>Discussion </li></ul>05/19/10
<ul><li>Fiber optic access networks (fiber-to-the-premises and fiber-to-the-node) have boosted individual user’s broadband access speeds. </li></ul><ul><li>IP networks may soon become a delivery mechanism for broadcast television content. </li></ul>05/19/10
<ul><li>IP data packets in FTTP network might include: </li></ul><ul><ul><li>Broadcast video </li></ul></ul><ul><ul><li>Video-on-demand (VOD) </li></ul></ul><ul><ul><li>Web applications such as web surfing, online gaming, and p2p file transfers </li></ul></ul><ul><li>Demand of a video stream (under MPEG-2 encoding) </li></ul><ul><ul><li>Standard definition (SD) : 3.75 Mbps </li></ul></ul><ul><ul><li>High definition (HD) : 15 Mbps </li></ul></ul><ul><li>Evaluating bandwidth demand is necessary. </li></ul>05/19/10
05/19/10 Portion of a FTTP network How big do these links need to be? Core router and edge router deliver content to the edge of the network . An OTL forwards the content over a PON to an ONT at each subscriber’s premise. 32 subscribers 2000 subscribers Core Router Edge Router Optical Line Terminal (OLT) Passive Optical Network (PON) Optical Network Terminal (ONT)
<ul><li>Steady state demand </li></ul><ul><ul><li>Analytical methods for engineering links often assume stationary (steady state) busy hour traffic. </li></ul></ul><ul><ul><li>All viewers have settled into channels </li></ul></ul><ul><ul><li>Using multicast to satisfy demand </li></ul></ul><ul><li>Channel surfing </li></ul><ul><ul><li>Disrupting the steady state at every commercial break </li></ul></ul><ul><ul><li>Significant additional bandwidth demand is required. </li></ul></ul>05/19/10
If there exist two viewers watch the same channel (5 different in total) 05/19/10 If all viewers watch different channels (6 different channels) 6 video stream required 5 video stream required Viewers OTL Edge Router The network needs to deliver only one video stream for the same channel to the OLT where it must divide the stream for two viewers.
<ul><li>The way to make channel changes fast is to send surfers unicast (one per viewer) streams at higher rates. </li></ul>05/19/10 Channel change server Playout buffer (in the set top box) Higher rate Usual rate
05/19/10 Settled in one channel (steady state) Channel surfing (commercial break) Bandwidth Time <ul><li>Bandwidth planning issue </li></ul><ul><li>What is the demand? </li></ul><ul><li>How to model the traffic? </li></ul>
<ul><li>A viewer: a device capable of receiving an IPTV stream </li></ul><ul><li>Each set top box always remains on and receives some video stream </li></ul><ul><li>This paper only to characterize broadcast IP TV’s contribution (from edge router to an OLT) </li></ul><ul><li>Homogeneous viewers </li></ul>05/19/10
<ul><li>The program definition is independent of the renewal process </li></ul><ul><li>The number of active viewers is constant during commercial break </li></ul><ul><li>Commercial breaks on all channels begin at the same time and all active viewers begin surfing at the same time </li></ul>05/19/10
05/19/10 Steady State Multicast Model Channel Surfing Model – Single User Extend to multiple users Focusing on the viewers who are not surfing The Time to Fill the Playout Buffer Single Viewer Behavior Combined Model (Multiple Users) Bandwidth Demand
05/19/10 Channel i , probability that a viewer will choose
05/19/10 The time to start channel surfing when commercial break starts The channel surfing ends
<ul><li>The paper developed a model to quantify bandwidth demand during the transition from surfing to steady state viewing. </li></ul><ul><li>The example with 400 viewers shows mean demand during surfing peaking at almost two times the steady state level if the service provider offers fast IP TV channel changes. </li></ul>05/19/10
<ul><li>Assumptions of the mathematical model </li></ul><ul><li>How to model the user behavior more realistic? </li></ul><ul><li>What else can we contribute to the IP TV filed? </li></ul>05/19/10