DVB Basics

[Doom9]

When I decided to get started with DVB, I was in for a rough ride. I couldn't find a site that told me everything I really needed to know, and so even though I already has a basic knowledge when I went to a store for the first time, I was still at the mercy of the salespeople. This is an attempt to give you some of the information I think might be useful to you, and which helps you to become a better informed customer.

The DVB standard currently consists of 3 major parts:

DVB-S
DVB-C
DVB-T

The last letter indicates the way the signal is transmitted. DVB-S is based on satellites, DVB-C is based on the cable network in your house, and DVB-T is based on terrestrial transmission. All technologies supercede existing analogue TV via satellite, cable and terrestrial antenna. But the DVB standard is broader than just -S/-C/-T. DVB-H (http://www.dvb.org/documents/white-p...VB-H.final.pdf) is an upcoming standard to broadcast TV content to mobile devices like PDAs or mobile phones. DVB-S2 (http://www.dvb.org/documents/white-p...B-S2.final.pdf) is an extension of the DVB-S standard which is both more efficient than the existing DVB-S standard, and offers more services. It is also geared towards the transmission of HDTV content.

Before we get started a bit more info about satellite reception, which is currently the most commonly used way to consume DVB content. You need a satellite dish that is pointed into a fixed direction (unless you use a system with a motor.. in this case your dish has to be repositioned to receive from a different satellite). Yet, it is still possible to receive from more than one satellite using just one dish. All you need is satellites that are not too far apart (satellite positions are usually indicated in degrees East / West - those correspond to the geographical of a satellite. Here's a list of satellites: http://www.lyngsat.com/), and an LNB (low noise blocker, also known as LNC - low noise converter) that receives the satellite signal from the dish (the signal is reflected from the dish, for a graphical view check this guide: http://www.doom9.org/DigiTV/dbox-positioning.htm). An LNB can receive from exactly one signal, so you need an LNB per satellite if you have a fixed dish (or one for a mobile one). Note that while a mobile dish sounds like a good idea at first, it's much harder to setup and generally a bit more problematic from an operational standpoint.

Satellites can transmit in two bands: low band (10.6-11.6GHz) and high band (11.6-12.7GHz) using horizontal or vertical polarization. The LNB maps the signal down to 950-2050MHz for cable distribution, lowering the frequency by the LOF (local oscillator frequency) which is usally 9.75GHz for low band and 10.6GHz for high band. An RF cable can only transmit one of the 4 possible combinations (low/vertical, low/horizontal, high/vertical, high/horizontal) and the signal is selected by LNB power supply level (14/18V) for H/V switching and a modulated 22KHz tone for selecting the high band. You can have three types of LNB: Single output LNBs can serve only a single receiver. Then there are two types of multiple-output LNBs: LNBs with a built-in switch have 1, 2, 4 or 8 outputs, and can serve the same number of receivers as they have outputs. Other LNBs output either each of the 4 possible signal types separately (quad output) or offer only low band, separated into vertical and horizontal (2 outputs). Those require a DiSEqC multiswitch, allowing one or multiple receivers to get the proper signal.

If you have a single output LNB, you can connect one DVB-S receiver. Many receivers loop the signal to another RF output, allowing you to connect multiple receivers in serie. However, if you have a multi satellite setup, looping the LNB means that each receiver can only receive from the same satellite (different channels are still possible). If you want each receiver to be able to chose a satellite independently of the other receivers, you need a DiSEqC switch, which allows each receiver to select the proper input signal. A receiver sends a specific 12 KHz signal to the switch, telling him from which input it wants to receive the signal from. Depending on the number of outputs of the switch, you can connect a certain number of receivers independently of each other.



Besides the means of transmission, what are the main differences between the 3 DVB standards? Channel bandwidth and signal modulation.

On a satellite, we have a number of so-called transponders. Those are circuits that receive the signal from the earth, modulate, amplify and re-transmit it back to the earth. There are about 20 - 30 transponders on a single satellite, and each has a bandwidth between 27 and 72 MHz (most transponders use 36 MHz, which offers about 38 Mbit/s). A TV channel occupies only a fraction of this bandwidth (how much?) and bitrate (many DVB channels use a lower bitrate than DVD, so the average is below 5 MBit/s), which allows for a number of channels on a single transponder. Most satellites also carry digital radio, which of course uses a lot less bandwidth. The signal sent back to earth using QPSK modulation. This means, a single transmitted symbol contains two bits, and every symbol has the same amplitued and is 90° apart from the next.

On a cable network, the available channel bandwidth is smaller (8 MHz), and the modulation scheme used is 64QAM. As one symbol can have 64 values rather than only 4 as in DVB-S, more information can be transported per symbol, which leads to the same available bitrate per channel (38MBit/s). 64QAM is more complex to decode and more error-prone, but since the signal quality is better than when receiving from a satellite (keep in mind that the distance from the satellite to your dish is huge), you're still likely to get a less distorted signal.

The youngest deployed DVB standard, DVB-T uses regular house antennas. Channel bandwidth is 8 MHz just as in DVB-C and the modulation scheme used is COFMD (Coded Orthogonal Frequency division multiplex) - the same as for terrestrial digital audio broadcast (DAB).

Besides the modulation coding, DVB also uses convolutional coding, and forward error correction (adding additional information to the transmitted signal, which permits its reconstruction if a part of the signal gets lost / is corrupted), but a full understanding of this requires a college degree, and not everybody likes signal processing (I've had it and didn't particularly enjoy it).

So, now we know why a DVB-S receiver can't handle DVB-C/-T broadcasts and the other way round.

It is important to understand that so far, the term channel wasn't used for a single TV channel, rather it applies to a transmission channel, which can contain multiple radio and TV channels. However, the term channel still makes sense, because what is being transmitted is essentially one channel, or using the appropriate technical term: a transport stream:

DVB uses MPEG-2 compression for video, and either MP2 (MPEG-1 audio layer 2) or AC3 (Dolby Digital 2.0 or 5.1) for audio. Audio bitrates used are usually in the 192 - 256kbit/s range for MP2, and 192 - 448 kbit/s for AC3. If you've ever used an MPEG-2 encoder, you may have noted that you have two output options: elementary streams and program (system) streams. The former will give you one audio (.mp2) and one video (.m2v or .mpv) file, whereas the latter will give you one single file containing both audio and video (usually an .mpg file). When you create the latter format, your encoder, divides audio and video into packets of a common size (the size can vary). Each packet of such a stream (known as PES: Packetized Elementary Stream) has a 8 byte header which besides a 3 byte startcode, contains 1 byte for the stream ID, a 2 byte to indicate the length of the packet and two timestamps: the DTS (decoding timestamp) and the PTS (presentation timestamp). The former indicates when a packet has to be decoded, and the latter when the decoded packet has to be send to the decoder output. Why do we have two timestamps? Because MPEG-2 allows for bidirectional encoding (b-frames), which requires certain frames to be decoded out of order (as an example b frames reference previous and future frames, and in order to decode, both referenced frames have to be available. So if frame N references frames N-1 and N+1, and N is a b-frame, the decoder has to decode the frames in the following order: N-1, N+1, N, and sends them to the output in the following order: N-1, N, N+1).

On a DVD, we have a program stream, which is a multiplexed collection of PES packets, all coded according to the same reference clock. The video stream is divided into access units, each containing one frame and an identifcation number corresponding to the order in which the frame has to be displayed. Audio is also grouped into packets of the same size (each audio frame has the same size). Both video and audio stream(s) have their own reference clock so they can be synchronized.

To transmit digital video, another structure is used: Transport streams. A transport stream can in fact contain both multiple video and audio channels (in contrast: a program stream allows only for one video stream), thus, a transport stream can contain multiple TV chains, each encoded at different bitrates and having different timestamps. A TS stream is also based on PES packets, but in order to identify which packets belong to which station, additional information is needed: the PSI (Program Specific Information) is used to tell a decoder which packets belong together (video, audio and additional data like subtitles, teletext, etc.).

The PES packets in a TS stream are all 188 bytes large. The first 4 bytes is used by a header, which contains a transport error indicator, a packet identificator (more about this later), some scrambling info (for scrambled TV channels) a continuity counter (which allows the decoder to determine if a packet has been omitted, repeated or been transmitted out of sequence), and some more fields that shouldn't concern you unless you write an application that should handle TS streams. To have a common clock (ticking at 27 MHz), the adaptation field is periodically used to insert a global timestamp (PCR: Program Clock Reference)

Based on the PSI, a decoder can extract the packet identifiers (PIDs) that belong to a certain TV channel, and decode only those if you're watching a certain channel. If you've ever seen a program capable of recording multiple audio streams at once, or even multiple channels at once, you notice that for this to work you have to indicate the proper video and audio PIDs so that the decoder can extract the proper data. Alternatively, you can record the entire TS with all the channels it contains, and later on extract whichever content you're actually interested in. Note that this only works for TV channels on the same transmission channel, unless you have a device that can tune into different channels at the same time.

So now that you know how the signal is transmitted, and what you're receiving, you might be wondering about the possible recording formats. TS seems to be the obvious choice, it's the only format where you can be sure to get everything you want (multiple audio channels if available, and all the timing information allowing your application to create a proper program stream again). However, most DVB softwares can only record in PVA or MPG (program stream) format. So which should you pick?

PVA is the native format of the Technotrend (Skystar1 compatible) cards like the Hauppauge Nexus. It's main advantage is that it allows the storage of multiple streams of various types, which isn't possible in MPG (depending on the type of streams of course). However, if you're using a different card, the additional processing of converting the data into the PVA format could introduce additional errors, thus making the PVA format less efficient. So, if you need subtitles or PCM audio streams, PVA seems like a good choice, but otherwise you can safely use MPG. Of course, TS is still your best option, it offers a demuxing program the best possibilities to correct errors, and it's the only format that allows you to capture any kind of transmitted streams, and not restrict you when it comes to capturing multiple TV channels/ audio channels concurrently.


Resolutions, aspect ratios and framerates:

Even though DVB is mostly used for regular (SD) TV broadcasts at DVD resolution (720x480, 30 fps with drop frame flag, resulting in 29.97 playback fps for NTSC and 720x576 25fps for PAL), the standard also supports HDTV. So far, HDTV over DVB is only frequently used in Australia's DVB-T broadcasts. For SD broadcasts, a variety or resolutions is supported:

720/544/480/352x576 (or x480 for NTSC, for NTSC we also have 640x480), and 352x288/240 (PAL/NTSC respectively. Interestingly enough, one can often find resolutions that are not part of the specs (ETS TR 101 154), like 704x576 or 528x576. All PAL resolutions require 25fps framerate, except for 720x576, where 50 fps is also allowed. All frames can be either progressive or interlaced except for 50fps which has to be progressive.

In the NTSC area, 23.976 and 29.97fps is always supported. Additionally, at 640x480 and 720x480 24fps and 30fps is also allowed and at 720x480 we also have 59.94 and 60 fps (both progressive or interlaced).

So far, all resolutions can be used for either 4:3 or 16:9 content (except for the 640x480 resolution, where only 4:3 is allowed).

For HDTV, DVB supports resolutions of 1920x1080 (at 23.976, 24, 25 and 29.97fps progressive and 29.97/30 fps interlaced) 1920x1035 (25, 29.97 and 30 fps interlaced), 1440x1152 at 25fps interlaced and 1280x720 at 23.976/24/25/29.97/30/50/59.94/60 fps progressive. All those resolutions are only allowed for 16:9 content.

Even though resulutions, framerates, bitrates and audio for SD streams looks like it is DVD compatible, DVB video streams must not be DVD compatible. While bitrate is usually not a problem, the GOPs can be longer than allowed for DVD. While many players can handle that (if you get your authoring program to accept the streams), it could lead to playback problems (stutter, or picture freezing).


Last but not least: DVB offers more services: It allows Internet access (commonly used for satellite, but it seems to be possible using the other DVB standards as well), and to encrypt channels (for Pay TV). DVB content can be encrypted on PES or TS level. The DVB-CA standard (http://www.dvb.org/index.php?id=50) defines how a decoder can decode encrypted channels. A variety of encryption standards is supported, for instance Nagra, Irdeto and Viaccess. In order to decrypt a channel, a CI (common interface) is used. It is based on a standard PCMCIA slot in a DVB receiver. In it, you plug a PCMCIA card which holds a decryption card that looks just like a bank card. You need a different CI module for each encryption standard. Some receivers can also directly read certain descrambling cards, thus saving you from buying a CI module.


Sources:
www.dvb.org
http://www.dvbviewer.com/forum/index...92&hl=dvbmagic
http://iphilgood.chez.tiscali.fr/
http://mpucoder.kewlhair.com/DVD/