IP-TV pushes silicon, software encoded

Increasing worldwide broadband penetration and the emergence of newer, more efficient video compression technologies enable telecommunication and service providers to deliver Internet Protocol (IP)-based video services to consumers. The emerging IP-TV and networked entertainment applications bring networking and audio/video (A/V) processing together, placing new requirements on silicon and software for devices such as IP set-top boxes (IP-STBs), digital media adapters (DMAs), triple-play gateways and similar IP video appliances.

Digital video consumer equipment such as DVD players, STBs and personal video recorders (PVRs) primarily use MPEG-2 video coding; the integrated system-on-chip for these systems have largely implemented MPEG-2 in hardware. In an IP-TV scenario where the media content is sent over a broadband connection, MPEG-2 is inadequate. Newer, more-efficient compression algorithms are needed to enable consumer-quality video-over-broadband IP service. For example, Windows Media (WMV9) and H.264/MPEG-4 part 10 (AVC) are newer low-bit-rate video codecs offering significantly higher compression than MPEG-2.

These newer codecs can offer a good viewer experience using much lower bandwidth than what is needed for MPEG-2. These low-bit-rate codecs are a requirement for making IP-TV a reality for broadband connections like asymmetric digital subscriber line (ADSL). New silicon for IP-TV applications must support these multiple, newer compression technologies like H.264, WMV9and MPEG-4 in addition to legacy codecs such as MPEG-2. There are several approaches for handling this multiformat A/V processing, including both hardware and programmable methods.

Historically, MPEG-2 systems have usually used hardware. Some solutions providers are attempting to use the same hardware-based approach for the newer video codecs like WMV9 and H.264; however, there are many challenges and issues with this approach. First, this approach requires dedicated hardware for each codec. Second, some of the newer codecs don't lend themselves well to hardware-based implementation as they are still in flux-many of the standards continue to evolve, with regular revisions at both the codec level and at the transport-stream level.

The hardware-based implementations have limited flexibility and upgradeability and, thus, are ill-suited to these emerging video standards and varying requirements for service-provider-driven IP-TV deployments. Some silicon providers have tried a hybrid approach in which some aspects of the video processing are done in software. Most of these are near-hardware implementations, essentially sharing the same limited upgradeability and flexibility as the fully hardware-based deployments.

Programmable platforms

Fully programmable solutions that handle multiformat audio and video processing have so far been the only proven way for handling the newer, more efficient low-bit-rate video codecs like WMV9 and H.264 along with supporting legacy codecs, like MPEG-2. Many codec formats, such as WMV9, have multiple variants, including Main Profile, Advanced Profile and VC-9 (SMPTE standard). Programmability provides maximum flexibility and upgradeability that are extremely important attributes for the demanding requirements for telco and other service-provider-driven IP-TV deployments. IP STBs and DMAs need to work with video portals and handle any content they encounter. There may be large variations in the way in which content may be encoded or encapsulated.

Although a fully programmable platform is the best approach for a multiformat A/V solution, it has its own set of challenges. Until recently, programmable solutions have been either essentially embedded PCs or expensive very-long-instruction-word architectures, neither of which lent themselves well to consumer IP-TV solutions for many reasons, including system cost, complexities in design and power dissipation. Recently, however, some new breakthrough programmable solutions that overcome these limitations are emerging. Programmable solutions require optimized software for A/V processing including support for a number of audio, video codecs and post-processing functions.

Networking requirements

IP-TV applications need a wired or wireless network connection. The connection is typically 10/100 Ethernet, 802.11, HPNA, powerline or MOCA. A simple IP-TV client, such as an IP-STB or DMA, would receive media content over the IP connection and would need to decode the compressed video to NTSC/PAL for A/V out to a TV. This basic IP-STB can be combined with many other functions, including gateway, VoIP, videoconferencing and PVR. The networking protocols that need to be supported depend on the functions integrated into the system. Service-provider-driven IP-TV applications typically require IP-multicast support for live TV and a number of protocols for supporting video-on-demand (VoD). For a VoD-type application, an IP-TV client needs to communicate with a VoD server. A number of protocols used for this, including RTP, RTCP, RTSP and HTTP. These protocols are for controlling the media stream and streaming of content. In addition to the standards-based protocols, most of the VoD server vendors have their own proprietary quality-of-service (QoS) extensions.

When IP-TV functions are integrated into the gateway along with other services like VoIP, virtual private network and a firewall, a more sophisticated network processor that supports wire-speed forwarding engines, IP-QoS and hardware-based encryption and authentication becomes important.

IP-TV requires support for digital rights management and/or conditional access, depending on the application.

Roshan Gudapati (roshan.gudapati@analog.com) is director of product strategy for Network Media Platforms in the Media Platforms and Services Group of Analog Devices Inc. (Norwood, Mass.).