Feature Article

Transcoding Cuts through Codec Complexity

This article originally appeared in the August issue of INTERNET TELEPHONY

With the advent of VoIP technology, continued advancements in telecom engineering and growing multimedia consumption, the variety of audio and video codecs in the network is increasing. This explosion of codecs is creating additional challenges for service providers in terms of increased codec complexity in their networks. In addition, the volume of media streams – particularly video media streams in the modern 3G mobile and LTE (News - Alert) networks – continues to grow.

There are now hundreds of codecs to meet the needs of new service models delivered by all network types, including wireless, wireline, enterprise and satellite. For instance, in certain pockets of the evolving network where end-to-end IP broadband connectivity is available, the industry is seeing increased adoption of high-definition audio codecs, which deliver a high fidelity audio experience. At the same time, a huge installed base of subscribers continue to use the PSTN or 2G mobile services, where legacy narrowband codecs designed for 64kbps circuit-switched networks are the norm. Without a transcoding function somewhere in the call path, these endpoints would not be able to connect.

Challenges in IP Network Transcoding

Transcoding is the process of converting from one encoding format to another. Transcoding and other media conditioning functions require significant processing power in a network. This is best achieved with network equipment specifically designed to handle large volumes of IP media stream processing.

Codec complexity originates at the remote edges of the network, as many different device types frequently operate with their own preferred codec standards. For the network operator wishing to gain control over this complexity, a key objective in a transcoding network design is to reduce the total number of codecs supported in the core network. With fewer codecs supported in the core IP services network, the network operator can reduce operational complexity, lowering media processing investment and minimizing equipment capital expenditures and ongoing operating expenditures.

Another transcoding design objective is to reduce the media processing steps in the call path. Because transcoding is a processor-intensive function, each transcoding step adds a slight delay to the overall call delay. Therefore, by reducing the transcoding and media processing steps, overall call delay is reduced and call quality improved.

Network operators need optimized transcoding solutions that deliver economics, network simplicity and call quality. But to approach these questions, the first decision that must be made is where to perform transcoding.

Transcoding today is performed in one or more of three network device types: in the media gateway, or MGW; by session border controllers, better known as SBCs; or via the IP media server, or MS. The question then becomes: Which can best meet operators’ cost, complexity and quality requirements?

Locations for Transcoding in the Network

One network location for transcoding is often found in MGW equipment located between circuit-switched access networks and the IP services core. The core function of a MGW is to convert a TDM circuit to an IP packet stream. However, TDM investment has been declining, and with networks like LTE increasingly delivering end-to-end IP connectivity to the devices themselves, IP traffic growth will increasingly dominate.

Enterprise connectivity is also changing. In the past, enterprises connected their PBX (News - Alert) equipment to the PSTN using E1/T1 or ISDN PRI circuit-switched trunks. Here also, TDM connectivity is giving way to SIP trunking and again, is diminishing while IP connectivity is exploding.

A second location for transcoding is sometimes found in SBCs located at IP network peering points. SBCs were originally designed to focus primarily on the IP signaling and security interface between two autonomous IP networks, such as a carrier network and an enterprise IP VPN. The SBC’s core function is to analyze and update IP packet headers flowing across the border, but in turn, it is not always optimized for full-scale transcoding or other media processing needs.

However, operators have two SBC architecture alternatives: integrated SBC, signaling and transcoding in a single SBC element; or a decomposed SBC, signaling only in the SBC with transcoding in an IP media server. IP media servers are already found in many networks today to support a large variety of IP media stream processing functions, but relevant to this discussion, the IP media server has been delivering transcoding as an underlying function for years.  

The integrated SBC was, as described previously, not originally intended for transcoding, so when you apply transcoding in both the SBC and the MRF, it results in an inefficient duplication of media processing resources. The second option, the decomposed SBC, allows the SBC to focus on what it was intended to do, while the media server focuses on what it was designed for – media processing. This approach meets providers’ need for improved economics, reduced network complexity and enhanced call quality by providing a cost advantage over alternative approaches, scalability, media conditioning capabilities and deployment flexibility.

Transcoding in the IP Media Server

Using an IP media server provides an optimized approach for IP-to-IP transcoding in 3G mobile, LTE and IMS networks. A media server can deliver audio and video media processing for IP telecom services offering multimedia conferencing, ringback tones, IVVR applications and more. In addition to these media processing capabilities, it can support built-in transcoding. An IP media server with such capabilities can perform large-scale, real-time transcoding between different codecs in large service deployments.

A media server-based transcoding solution with signaling only performed in the SBC can deliver up to a 50 percent cost advantage over other approaches with integrated transcoding. Due to a dense digital signaling processor platform and centralizing media processing in a general purpose media server, transcoding efficiencies are achieved and costs decreased. Meanwhile, the signaling-only SBC will typically have much higher call per second performance, compared to performance when the SBC needs to perform signaling and transcoding together. Also, by reducing the number of transcoding operations in an end-to-end call path, cost is further reduced and call quality is improved.

In addition to audio transcoding, carrier networks often require other media stream processing capabilities, referred to as media conditioning. IP media servers have the capability to perform transcoding on IP media streams in real time while also performing other functions including voice quality enhancement, video transcoding and transrating and IPv4/v6 address scheme normalizing.

Voice quality is an important requirement for many network operators that offer telephony services based on VoIP technology. Despite providing tremendous economic benefits, VoIP also presents new voice quality challenges such as dropped packets, variable packet delay, and packet reordering and corruption. In-network quality enhancement solutions, generally referred to as VQE, are available to effectively address these issues using techniques like acoustic echo cancellation, dynamic noise reduction and packet loss concealment.

Voice quality improvements can be achieved as part of the transcoding solution design itself. As mentioned earlier, each media processing step adds some level of delay in the call path. Rather than media being processed twice as in an integrated SBC approach, first in the SBC and again in the media server, by focusing transcoding and media conditioning functions on the media server, media is processed only once in the media server, reducing delay by up to 50 percent and improving overall voice quality.

An IP media server approach to transcoding also provides more flexibility in terms of network location support.

For example, in one approach the media server is controlled by a signaling element in the network using a control interface such as SIP or H.248. The signaling element can be an application server, call state control function or the signaling component of an SBC. By using this approach, not all calls are processed by the media server, only calls requiring transcoding. This reduces equipment investment, while also allowing per-stream control of services media processing and media conditioning.

Another approach involves the media server being deployed directly in the call path. This approach eliminates the need for external control integration, which simplifies overall network design deployment. It also allows selective media conditioning based on rules and triggers.

Network operators have a variety of transcoding solution approaches available on the market today. By choosing to perform transcoding and media processing in the IP media server, service providers are supported by a solution that is optimized for IP-to-IP transcoding in 3G mobile, LTE and IMS networks, scalable across wide range of codecs, economical and flexible.