Dilithium Technology

Video

The delivery of video services is far more complex than that of pure voice or data. Dilithium engineers and technical experts have been involved in broad range of standard and technology development in the audio/video compression and multimedia communication areas since the early 1990s. Today's Dilithium products and solutions are underpinned by many of these technologies, either in a unique offering or as part of the standards.

Of particular interest to Dilithium is the topic of video quality and customer experience. Dilithium has developed a range of unique technologies implemented today in its products. We summarize these in the next few sections.

Dilithium is the leading vendor of advanced video transcoding and transrating solutions. The Dilithium patented Video "Unicoding" technology represents a breakthrough in video processing at the algorithmic level and enables efficient and on-the-fly transformation from one codec (e.g. H.263, MPEG4-part2, AVC/H.264) to another (e.g. H.263, MPEG4-part2, AVC/H.264).

Dilithium Video Unicoding offers several significant benefits compared to the brute force and far less efficient tandem decoding-recoding method used by competing solutions. As a starting point, the Dilithium Video Unicoding offers equal or better quality than tandem transcoding. Unicoding also improves the customer experience by minimizing end-to-end delays and maximizing lip-sync. End-to-end delays are minimized because Unicoding eliminates the need to wait for video frame bits and the brute force transcoding and burst emission target codec bits. Unicoding allows the transcoding and transmission of video bits in a progressive manner and without huge burst of bits. The lip-synch is maximized because Unicoding video delays is significantly shorter that brute force transcoding making the emitted video aligned with the audio by construction.

The superior session capacity enabled though Unicoding results in a smaller footprint and equipment count leading to lower CAPEX and OPEX. Other benefits include the pooling of gateways or application servers across many services and applications.

Most video value added services involve the streaming of some form of video content (clips), whether these are the actual clips themselves, the greetings or the navigation menus visible to the user. Hence any interaction between a subscriber and a service is likely to involve the streaming of a video clip. Other applications such as video ring-back tones, short-code video access, video push, etc, all also involve the transmission of some form of pre-coded clips.

The streaming of video clips is typically performed by a streaming server or the pump of a media server. Such servers do not have facilities to encode video in real-time, and are only capable of streaming out the pre-coded clips which consist of a sequence of key frames and non-key (predicted) frames.

The transmission of video (and any data) over mobile networks incurs corruption which is to interference of fading communication channels. Transmitted streams are not error-free and such corruption results in various video corruption artifacts that can range by some out-of-place blocks, ghosting, and more severe and unrecognizable video frames. The duration of the corruption can typically last 1-5 seconds depending on the time intervals between key-frames in the clip. Video is restored by the arrival of the next key frame in the clip to the terminal or to the server in case of a blogging or video mail (message deposit) application. Note that in the case of a person to person communication (handset to handset), the video corruption also exists but the recovery is much quicker (a fraction of a second) as handsets are equipped with facilities to request and transmit key frames on demand.

Video Refresh is a unique Dilithium patent-pending technology that minimizes video corruption while maintaining the highest video quality. Video Refresh operates in the Dilithium Gateways family of products and performs two functions: (1) the fulfillment of handset request of key frame generation and transmission on-demand, and (2) the continuous detection of video corruption on streams received from handsets (e.g. blogging application or video mail message deposit) and the immediate generation and transmission of key frame requests to the handsets. Because Video Refresh operates on demand, it dynamically responds to handsets requests and network conditions. Therefore, the duration of any corruption in streaming to handsets or from handsets to servers is minimized. Another important benefit of Video Refresh is that it automatically adapts to the visual scenes, whether they are natural videos with smooth long movements and slow varying scenes or rapid visual scenes movements such as in music videos. Video Refresh is the only technology that allows service providers to eliminate the high costs associated with special encoding of video and the degraded quality that results from high frequency key frames which impacts video smoothness, frame rate and video quality.

Video Transrating (changing of bit-rate, mostly downward) is a key feature for today's video communication applications. Most video value added services require a form of transrating, in particular when video streams originate from one network or a source where the bit-rate can vary dynamically and can be higher than a destination network/terminal where the target bit-rate has to be lower. The source video bitstream may be real live video or a video clip. An example is the transmission of video from a broadband IP terminal (e.g. SIP/IMS) towards a 3G mobile terminal (circuit-switched or packet-switched) where the bit-rate is lower. This scenario is very common, for example in call centers, surveillance, mobile-fixed peer-to-peer communication, etc. Transrating is most situations has to be done on the fly, in conjunction with or without transcoding. For example both end-points may be using H.263 or MPEG4 but one terminal could be transmitting video at one rate (e.g. 96 kbps) and the other terminal can only receive video at a lower rate (e.g. 45 kbps). In most mobile-fixed convergence operation transrating is unavoidable.

If transrating is not done, then serious artifacts may result – for example video streams can backlog and lead to either lost packets or video delay build-up resulting in serious lip-synch problems.

Video transrating is a computationally intensive operation. Typically the input video for the transrating operation may have higher frame rate and bit-rate, and the target bit-rate would necessitate reduction in frame rate and bit rate. If bit rate is only reduced, a degraded video quality may result.

The Dilithium patent-pending Video Transrating technology is a key element of the Unicoding technology: transrating is done on the fly with or without transcoding. A key benefit of the Dilithium Video Transrating is an enhanced customer experience where lip-synch is maintained and the service provider need not to worry about the matching of source and target bitrates.