MPEG-2 and H.264 use blocks, macroblocks, and slices. HEVC replaces them all with Coding Units (CU’s) up to 64x64.
Testing and troubleshooting the content from the source to the viewer requires two types of assessments: 1) syntax/semantics to ensure the stream will play, and 2) quality of video and audio to ensure the best possible viewer experience.
The Society of Motion Picture and Television Engineers (SMPTE) specifies UHDTV1 as the 4K format, and UHDTV2 as the 8K format (double 4K format). This story will focus on 4K, and is based on the Tektronix eBook “A Brief Guide to Ensuring Quality & Compliance While Delivering 4K/UHD.”
Most UHDTV1 premium content is encoded with H.265/HEVC for linear broadcast and ABR delivery at multiple bit rates. The HEVC standard (ITU-T H.265), is presently the most efficient way to deliver 4K content to viewers and subscribers. An HEVC encoder does not need to use every HEVC command, but the ones that it does use must follow the approved standard from the H.265 standard.
The HEVC standard includes many levels from level 6.2 for 4k (3840x2160) at 300 Hz down to level 1 for 128x96 at 15 Hz. Each of these levels defines a maximum bit rate and maximum luma sample rate. Figure 1 shows each of the many possible levels for an HEVC codec.
HEVC is about twice as efficient as the ITU-T H.264 video codec standard, and one of the key bandwidth-saving features is its redefinition of blocks. Where MPEG-2 and H.264 used blocks, macroblocks, and slices, HEVC has eliminated all three of these in favor of new Coding Units (CU’s). The largest CU size is the 64x64.
For pictures or video frames with large flat areas, the new CU feature allows for a large reduction in bandwidth. HEVC also includes more accurate motion vectors. The new 64x64 CU’s probably make the biggest impact. The lead photo in this story shows the block and CU size difference on the same video clip.
Transcoded video is often used to scale or reduce bandwidth. In the case of OTT and ABR (discussed later), one video program will be transcoded into many video outputs. Each video will be at a different bit-rate, format, or both allowing remote clients to seamlessly switch between rates. This feature requires accurate timing alignment of the Instantaneous Decoder Refresh (IDR) and Encoder Boundary Point (EBP) frames.
Monitoring and Troubleshooting
Monitoring and troubleshooting often requires a deeper look into the compressed video or audio stream because of detected errors or QoS/QoE issues found by the provider, viewer or subscriber. It is best to automatically record a transport stream when such errors or issues are detected. The recorded transport stream file can then be analyzed in greater detail by a codec analysis tool.
At each point in the system where the stream is manipulated, such as transcoding and re-multiplexing, it should be verified again to assure interoperability. If tests indicate failure on QoE (syntax) or exceeded thresholds, the transcoder vendor should be contacted immediately to bring the equipment back into an interoperability state since failed content causes viewer complaints. All tests must indicate compliance before content continues to IP or RF modulation for delivery to viewers.
Figure 3. The inverse relationship between EVM and MER is how an increase in noise or decrease in signal quality causes a digital signal to approach the “digital cliff". Click to enlarge.
Digital Video over RF
The main RF formats are terrestrial, cable, and satellite. For each of these RF formats, it is important to maintain a high signal to noise ratio (SNR) to ensure reliable delivery of the digital signal. As the signal level lowers or degrades, noise becomes a greater part of the overall signal. At some point, this ratio approaches what is called the “digital cliff.” This is the point where the built-in redundancy from forward error correction (FEC) and Reed/Solomon fails to preserve the compressed signal, and random values begin arriving within the decoder. At the same time, the display shows tiling, slices errors, freeze frames, and then a complete loss of the program.
With digital RF transmission, one of the modulation error ratio (MER) tests is called Error Vector Magnitude (EVM) and is measured in percent. An EVM of zero means the received symbol landed directly center of the expected target location. An EVM of 100% is bad and means that the received symbol landed outside the expected symbols’ target range, and usually in one of the other targets. The lower the EVM value, the better the signal quality. The higher the EVM value, the more that the receiver relies upon the Forward Error Correction (FEC) to correct any problematic symbols.
An example of a DVB-S2 8PSK signal is shown in Figure 4. The MER in this signal is 16 dB which is not great due to the significant amount of noise. The figure clearly shows that some of the received samples end up landing in unexpected target locations (symbol errors). This is also noted by the Inner FEC bit error ratio (BER) showing 2.35E-3. This means that about 2 bits out of every 1000 are wrong, so the Reed/Solomon FEC is required to correct these bad TS packets. So long as the rate stays better than 5E-3, we can declare this to be a quasi-error-free signal. Any worse, and some errors over time will leak into the video and audio elements and become noticeable.
If you see poor FEC BER, or if the TR 101 290 Transport Error Indicator (Priority 2.1) is red as in Figure 5, there is an RF transmission or reception problem. This could be due to rain fade, multipath, terrestrial interference, or any number of other RF impairments. Detecting the problem is step one. Eliminating the problem requires much more work. First, check your signal levels to ensure enough power. Second, ensure cable/coax integrity. For terrestrial interference (TI), a real-time spectrum analyzer will be required to pinpoint the exact location of the incoming TI.
Monitoring of the RF signal to be delivered to the subscribers is required to ensure that the RF modulation, up-conversion, combiner, and amplifier are all running according to levels and formats specifications. The key tests here are RF power level, signal-to-noise ratio (MER), and ETSI EN TR 101 290 for transport stream compliance.
Digital Video Over IP
Delivering transport streams over IP with User Datagram Protocol (UDP) has almost completely replaced all previous forms of sending transports between equipment, buildings, and even around the country. Monitoring local or distant UDP sessions can be beneficial when trying to pinpoint exactly where the IP error has been introduced. Standard browser IP traffic is lossless though Transmission Control Protocol (TCP/IP) where each packet is checked, and then resent if not 100% correct.
In the case of video over TS, we use UDP which is more like “send and forget.” If a packet becomes corrupt or lost, another copy is not sent. It is critical to ensure that 100% of the UDP traffic is maintained and error free. What happens when you lose just one UDP packet over IP depends on several factors. What type of codec is being used? How low is the bandwidth? Is the missing packet from part of an I-frame?
A loss of one UDP packet translates to seven TS packets. If the loss happens to be within an I-frame, the impact is at its greatest. The resulting video display will be heavily impacted by the decoder itself. Some decoders will black out the bottom half of the bad frame, others will create numerous oddly misplaced blocks, while others simply freeze the video until a new I-frame comes along.
ABR and OTT
With ABR and Over-The-Top (OTT) video, we have an entirely new method of delivering video to the subscriber. The key benefit with ABR is the ability to dynamically change the video bit rate (up or down) to prevent the “Buffering …” message displayed in the middle of a program.
ABR uses a completely different delivery method (TCP/IP) that includes retransmission of any missing or corrupt IP packets. The client-application at the viewer’s location communicates with the server to always maintain the highest possible bit rate. This requires the ABR solution to maintain several instances or video bit rates and formats such as SD, HD or 4K, so that the subscriber does not need to guess which quality or profile will provide the best experience while not having to stop and buffer incoming packets.
Images courtesy Tektronix.
Dennis Kucera is MPEG Applications Engineer at Tektronix.