-
Notifications
You must be signed in to change notification settings - Fork 9
DVB S2 FRAME RATES
DVB-S2 has two frame sizes, 16200 FEC bits and 64800 FEC Bits. These are the bits fed into the LDPC decoder and include the data, the outer BCH coding, and the inner LDPC coding. (FEC Bits are the bits that carry both the data being sent and the forward error correction code bits used to correct the data for the errors that occur during transmission)
P4B is planning on using VCM (variable coding and modulation). This means that we can change both the modulation and coding on every frame. This allows lots of different stuff to be multiplexed onto the single downlink. HIgh performance stations can utilize higher order modulations and lighter coding to achieve greater data rates. Lower performance stations can operate using only the most robust codes. All stations receive and track a single carrier.
In DVB-S2 the symbol rate is always a constant. A symbol in DVB-S2 carries 2 to 5 bits. Receivers have to track the carrier frequency and locate and maintain the symbol alignment. This usually requires course acquisition of frequency, fine acquisition of frequency, and then the tracking of frequency and phase with every symbol. Once the receiver has acquired the symbol stream, it searches for a 90 symbol block called the PLHEADER. The PLHEADER contains a special start of frame pattern called the SOF or start of frame pattern which gives it frame alignment. Continuous back to back frames are always transmitted. If there is nothing to send, a dummy frame is sent instead. The dummy frame consist only of pilot symbols. The dummy frame has it's own MODCOD so that the receiver will know how long it is.
In addition to the SOF, the PLHEADER contains a PLSCODE that indicates the type of modulation and coding that is in the frame. This code also indicates whether the frame is normal or short and whether or not pilot symbols are present. Together, this information indicates the length of the frame in symbols.
A receiver doesn't have to be able to demodulate and decode every frame. In fact, most users with smaller dishes will not be able to decode a 32APSK 9/10 FEC coded frame. Every receiver, however, has to be able to decode a QPSK rate 1/4 frame (to be useful) and every receiver has to be able to maintain frame alignment and decode every PLHEADER.
Currently, the maximum bandwidth envisioned for the satellites is 10MHz with a maximum symbol rate of 8Mbaud. This compares with the standard commercial broadcast transponders which typically use 36MHz wide carries transmitting at 30Mbaud. Commercial transponders are normally allocated in 40MHz channels. There is at present no standard way of allocating Amateur Radio Service wideband channels (but we might want to start thinking about this).
The demodulator takes the incoming digitized ADC samples, applies various filtering, finds the frequency and symbol alignment, and demodulates the symbols to an IQ constellation. All of this is typically done using a fixed sampling rate and all digital processing. The recovered signal constellation for a symbol may have 4, 8, 16, or 32 points depending on the modcod. At the output of the demodulator proper, a "mapper" function translates a single constellation point to 2, 3, 4, or 5 bits, again depending on the modcod. These bits are NOT described by a simple one or zero. Rather, they are transformed to an LLR (discussed further in another article) that encodes the probability that each individual bit encoded in the symbol is a one or a zero. These LLR's are then fed to the FEC decoder. This is process is called soft decision decoding. It allows the forward error correction to utilize more information about the signal, basically how reliably it is a one or a zero. This is particularly important at very low symbol to noise ratios where the noise and the signal are approximately equal.
Frames are organized into groups of 90 symbol that are called slots. Since each symbol carries 2, 3, 4 or 5 bits, there are 180, 270, 360, 450 slots bits in a slot. Each of these is evenly divisible by the number of frame size in FEC bits. So there are 90, 60, 45, or 36 slots in a short frame. Since long frames 4 times as large, the number of slots in these frames are 4 times larger.
If there are no pilot blocks, a short frame may range from 36 to 90 slots of 90 symbols each, i.e. 3240 to 8100 symbols. At the 8Msymbol rate planned for 4B, a short non-pilot frame is thus roughly 1 to 2.5ms long. This is a frame rate of roughly 400 to 1000 frames per second.
Even for a short frame 8100 symbols, this is a relatively long time to maintain precise timing synchronization, especially for when the C/N ratio at the receiver is low. Maintaining synchronization and symbol alignment is easier when so-called pilot blocks are inserted periodically. Since the data in the pilot block is known, the receiver can adjust it's timing and symbol alignment to the pilot block. The receiver may also use block oriented techniques that use the information from two successive pilot blocks to extrapolate the timing between the blocks.
In DVB-S2 pilot blocks consists of 36 symbols of carrier. The presence of the pilot blocks in a particular frame are indicated in the PLHEADER. Pilots can be inserted or not on a frame by frame basis. When pilots are present, they are inserted after every 16 slots unless the next slot if the PLHEADER of the next frame (which serves the same purpose as a pilot). Pilots add about 2.5% overhead.
Each frame carries DATA and two layers of FEC coding. DVB-S2 uses a systematic code. This means that the original data bits are always transmitted and the FEC bits are additional bits. These FEC bits are used to correct the data bits that may have been altered during transmission.
The two FEC codes are called inner and output codes. The transmitter applies the outer code first which in DVB-S2 is a BCH code. Then it applies the inner FEC code. Most of the FEC bits are used for the LDPC (Low Density Parity Code) code. If there are a small number of residual errors after the LDPC corrections are applied, the BCH code in many cases will be able correct them.