Click to request a PDF

Internet Protocol Television, or IPTV, is the application that will shape the network over the next decade. With total network bandwidth consumption at 100 times that of such applications as e-mail, gaming, and Web surfing, IPTV deployments will shape billions of dollars of network infrastructure spend.

IPTV is a service offered by a network operator that is intended as a direct competitor for satellite or cable TV. IPTV normally has the same video quality as satellite or cable TV, offers the same channels, and is intended to be watched by the subscriber on a television set. IPTV is usually offered by a fixed line operator that owns the copper or fiber lines to a subscriber. Within the United States, examples of IPTV are Verizon’s FiOS and AT&T’s U-verse services.

Internet TV, on the other hand, refers to services such as YouTube or Joost, which deliver video clips over the Internet. In general, Internet TV is watched on a computer monitor, offers short video clips, and is lesser in quality than IPTV.

Although some new services blur the distinctions between IPTV and Internet TV, the fundamental difference of whether the service is offered by the network operator, as is the case with IPTV, or by third-party companies, like Internet TV, remains.

Business Case for IPTV Network Spend

IPTV represents an enormous opportunity for network operators to compete with cable and satellite providers and is justifying major investments in network infrastructure to support IPTV. A recent forecast from Multimedia Research Group (MRG) shows 2007 IPTV revenue at $3.6 billion and growing to $20 billion in 2011, while the subscriber count grows from 14 million in 2007 to 64 million in 2011.

Figure 1 highlights the regional differences in IPTV uptake, with Asia and Europe holding the majority of IPTV subscribers today and in the future, and North America comprising only 10-15 percent. IPTV service offerings vary dramatically from region to region. North American and European offerings typically provide hundreds of channels, including multiple high-definition channels, and advanced set-top boxes with integrated Personal Video Recording (PVR), also known as Digital Video Recorder (DVR), functionality. IPTV services in China, on the other hand, are offered for three to six dollars per month and provide 50 or 60 low bandwidth channels and low cost set-top boxes.

Diagram Article IPTV Designer 1Figure 1


Click to get full document

There is much more!


To read it all, click HERE!

Speeds and feeds

A single video stream consumes 1.6-3.5 megabits per second (Mbps) for a standard definition stream and 6-15 Mbps for a high-definition stream depending on the codec used to encode the video stream and the level of quality that is desired.

Figure 2 shows a typical IPTV network architecture. IPTV systems use a combination of multicast and unicast streams to provide video services. Multicast streams are used for broadcast channels, while unicast streams are used for video-on-demand services.

Diagram Article IPTV Designer 2

Figure 2

A typical network offering would be 200 standard definition broadcast channels and 20 high-definition channels. This would consume roughly 600 Mbps and would comfortably fit on a single Gigabit Ethernet link. However, a typical access system such as Digital Subscriber Line (DSL) or Fiber-To-The-Home (FTTH) offers 20-30 Mbps to the subscriber, so IPTV systems do not send all channels to the subscriber at the same time. The only channels sent down the DSL link are the channels that the subscriber is watching at a given instant. The mechanism used for this is multicast group membership: when a subscriber wants to watch a channel, the set-top box sends a multicast-group join message (IGMP join) to the Digital Subscriber Line Access Multiplexer (DSLAM). The DSLAM interprets the group join message and starts to send that specific channel to the subscriber. It is the DSLAM that is responsible for monitoring the IGMP messages and replicating the incoming multicast traffic across all subscribers who are watching a given channel.

One important characteristic of IPTV bandwidth is that the average bandwidth usage is close to the peak bandwidth. This is because a television might be on for the entire day, requiring a constant video stream for 24 hours. In contrast, applications like e-mail or Web surfing typically see average bandwidth at 1-10 percent of peak bandwidth. This has important implications for network sizing: typical rules of thumb suggesting that the links from the access system to the core network can be oversubscribed by a factor of 10 are not appropriate in an IPTV deployment. As an example, a single hour of a High-Definition (HD) video stream will consume 4.5 gigabytes of bandwidth, more than a year’s worth of e-mail for a typical user. A single HDTV on for 24 hours a day will consume 39 Terabytes of bandwidth in a year – an astronomical amount.

The killer application

The killer application for IPTV, in both a positive and negative sense, is the network-PVR. With network-PVR, a subscriber can watch any content that has been broadcast in the last week or month without having to remember to record the program in a local PVR. Such devices, for example, TiVo, record video in a digital format to a local hard disk. The networkPVR feature also removes limitations from the subscriber on the number of programs that can be simultaneously recorded. In addition to offering benefits to the subscriber, network-PVR has the potential to lower the costs of IPTV by eliminating the expense of a local hard disk, thus reducing the cost of the set-top boxes. Given that many IPTV deployments require one IPTV set-top box per TV, not just one set-top box per subscriber, this cost savings can quickly add up.

However the cost savings on the PVR are offset by the increased costs at the network. Network-PVR has the potential to dramatically increase the network bandwidth consumed by IPTV. In a typical broadcast IPTV system, a single Gigabit Ethernet link from the video servers to the DSLAM can carry the entire set of programming regardless of the number of subscribers. The DSLAM handles packet replication and distribution based on the multicast group membership of each subscriber. Network-PVR turns this traffic into a unicast video stream for each TV in the subscriber household. Assuming 2.5 TVs per household, a community of 1,000 subscribers using the network-PVR service would require 22 Gigabits per second (Gbps) of network bandwidth.

Finally, network-PVR is the subject of ongoing litigation between Cablevision and a group of television networks and studios, which contend that the network-PVR is unauthorized duplication of the broadcasts despite the fact that local PVRs, or subscriber PVRs, are legal. The outcome of this litigation will determine whether this feature can be deployed in North America.

Security threats

IPTV networks are more exposed to threats than cable or satellite TV because the video traffic runs over an IP network that is shared with PC data traffic and voice traffic. It is comparatively straightforward for an application running on a PC to create and inject packets into the network that mimic the set top box or other elements of the IPTV network.

Five broad categories of security threats apply to IPTV:

  • Content theft
  • Service theft
  • Denial-of-Service (DoS) attacks
  • Pranking/spoofing
  • Bandwidth over consumption from P2P applications

Although each threat is different, they can all be addressed with deep-packet inspection and content filtering.

Content theft is the copying and distribution of video content. In IPTV systems, the video streams are encrypted between the video server and the set-top box, but if the encryption keys are exposed at any point, the IP video stream can be captured, decrypted, and distributed. Given the protectiveness of the content owners and their unwillingness to license content unless they feel it has been protected from duplication, this threat has received the bulk of attention in early IPTV deployments. As a result, content is relatively well protected against duplication, and it is the other IPTV security threats that warrant further attention.

Service theft refers to unauthorized access to content. Recall that all of the channels are generally available as multicast streams at the DSLAM, and the local set-top box sends IGMP join messages to the DSLAM to instruct the DSLAM to send the video stream over the DSL line. If the system has no other authentication, and a subscriber is able to forge an IGMP join message, the subscriber can watch premium content without paying for it. Combating this threat requires that the DSLAM have a mechanism to validate IGMP messages or that the control messages themselves are encrypted and protected.

DoS attacks in IPTV systems consist of overloading a network element and preventing it from offering service to other subscribers. This can be as simple as sending a flood of IGMP messages to the DSLAM and preventing other subscribers from changing channels, or overloading billing/application servers in the network and preventing subscribers from ordering new content.

Pranking or spoofing attacks rely on the open IP nature of IPTV and the fact that a PC can be used to forge an IPTV message. In a system with loose security, it is possible to direct a set-top box to play a video stream from a video server somewhere on the Internet or at another subscriber in the IPTV network – video server spoofing. Similarly, an unprotected billing system could expose the PayPer-View ordering habits of subscribers – leading to unwanted publicity for the IPTV operator the first time that a celebrity’s IPTV records are released.

Finally, although in a different category of threat, successful IPTV deployments rely on having sufficient bandwidth between the core network and the subscriber. The growth of P2P traffic threatens that bandwidth and requires that IPTV systems have careful traffic management and traffic shaping to ensure that surges in Internet or P2P traffic do not affect the bandwidth needed by IPTV.

Although the set of security challenges around IPTV is varied, a common thread exists around the need for a platform at the edge of the network that can act as a firewall and traffic shaper. In addition, deep-packet inspection should look for specific attacks against the IPTV network. Figure 3 shows the appropriate location for such a security and shaping platform – deployed immediately behind the access DSLAM. With the coming challenges of network-PVR requiring unicast bandwidth per subscriber, this traffic management and shaping platform needs to offer 25+ Gbps per 1,000 subscribers, requiring a high-density packet processing platform.

Diagram Article IPTV Designer 3

Figure 3

Standards-based platforms in IPTV implementations Standards-based platforms are an ideal match for the key IPTV network elements. The rapid growth in IPTV deployments and quickly evolving feature set demand solutions that can be quickly deployed to bring new applications into service. The release of high-bandwidth 10 Gbps per slot AdvancedTCA systems in 2007, such as that of Continuous Computing, marks the first time that a standards-based platform can be used for high-bandwidth packet processing and packet inspection.

Diagram Article IPTV Designer 5

Table 1 highlights the key requirements for each of the IPTV elements and describes an AdvancedTCA implementation for each element.

Diagram Article IPTV Designer 4Figures 4 illustrates an implementation of an 80 Gbps packet inspection and shaping platform capable of handling 4,000 subscribers at a full 20 Mbps line rate per subscriber. This platform uses the Continuous Computing FlexPacket PP50 packet processing blade, an AdvancedTCA blade that can inspect 10 Gigabits of full-duplex traffic. The PP50 is combined with a 10 Gbps per slot AdvancedTCA switch and includes a full suite of high availability middleware and management software.

Summary

The rapid pace of IPTV rollouts, with 10 million subscribers being added in 2008, demands solutions that are available today and can quickly ramp new applications into the network. Internal development of an application and platform can easily take 24 to 36 months – in that time, 40 million IPTV subscribers will be added. Open platforms based on AdvancedTCA that include integrated middleware, management software, and control plane protocols can shorten this development time to 12 months, dramatically increasing the market window and potential revenue for innovative IPTV applications.

IPTV Q&A with Mike Coward and Joe Pavla

Mike is CTO and cofounder of Continuous Computing. Joe is PICMG president and editorial director for CompactPCI and AdvancedTCA Systems.

Joe: With the shift away from the carriers to the content providers, how are the carriers, who are increasingly just roadways, going to come up with the money to build the next-generation high-bandwidth infrastructure to support things like IPTV?

Mike: That answer has two parts: First, the AdvancedTCA community has done a fantastic job of driving down the cost of these new platforms, making it more and more economical to deploy high capability IPTV to one’s subscriber base.

The second part of the answer is strategic. For the phone operators, IPTV looks to be the main weapon in their fight against the cable companies. The introduction of VoIP earlier this decade allowed the cable companies to take revenue from the incumbent phone companies, and they’ve been able to take 20-30 percent market share in some of their territories. IPTV is the first chance for the wire line operators to take a piece of the TV market, and they will invest huge sums in order to get that chance to compete.

Joe: How do you see the AdvancedTCA adoption rates?

Mike: AdvancedTCA is the dominant open platform for new design wins. Five years ago, CompactPCI was lucky to achieve 10 percent of design wins for core telecom equipment. With AdvancedTCA, it’s more like 75 percent. AdvancedTCA is the platform with the right set of power, capability, and cost structures to address a wide swatch of telecom applications.

Joe: I have noticed that people overestimate the rate of adoption for technologies such as AdvancedTCA and underestimate the overall total adoption for a 20-year period.

Mike: If the adoption of AdvancedTCA has been slower than expected, it’s because it has taken customers much longer to integrate blades and software together than they expected. What started out as a six-month integration exercise becomes 18 to 24 months, and this has delayed the rollout of AdvancedTCA into the mass market.

This is the challenge that we’re addressing with our system-level solutions that include high availability, integrated protocol stacks, hardware platforms, and system management software – everything but the application.

Joe: Agreed. It seems like everyone is trying to adopt a system-level strategy because they need to add value beyond the hardware and because of technologies like AdvancedTCA, where there are literally thousands of options. The ability to find a class of integrators who can put all that stuff together and integrate it with the software simply does not exist. And the customers want to buy things that are application-ready, with the OS and high availability firmware and middleware in place. A system-level strategy is absolutely what everyone is struggling to develop because there is not an integration base out there that knows how to do it.

Mike: Exactly. More than ever before, customers are embracing system-level vendors and are looking for system-level expertise. At Continuous Computing we have had very positive customer response to our strategy of developing much deeper network expertise than found in traditional blade manufacturers.

Related Information