[PATCH] Re: [linux-dvb] [PATCH] Multi protocol support (stage #1)

Manu Abraham abraham.manu at gmail.com
Fri Jun 2 16:22:05 CEST 2006

Johannes Stezenbach wrote:
> On Fri, Jun 02, 2006, Manu Abraham wrote:
>> Johannes Stezenbach wrote:
>>> So, it seems that for hierarchical BC mode PLHEADER does
>>> not use pi/2-BPSK. Now, I don't know if and how the
>>> receiver could figure out that hierarchical modulation
>>> is used, however, once it knows that, it can get the
>>> other transmission parameters from the PLHEADER.
>> Well, in BC-BS mode we have the S2 satellite delivery system descriptor
>> I will try to explain what i understood.
>> According to the patent
>> United States Patent Application: 20050089068
> [snip]
>> An approach is provided for supporting signal acquisition and frame 
>> synchronization in a digital broadcast system utilizing Low Density 
>> Parity Check (LDPC) codes. Hierarchical modulation is utilized to 
>> provide backward compatibility, whereby the lower layer signal is 
>> encoded using LDPC coding. A signal is received, whereby the signal is 
>> modulated according to the hierarchical modulation scheme including an 
>> upper layer and a lower layer. The signal includes a data pattern and a 
>> coded frame. The dependency of the received signal on the upper layer 
>> modulation is removed. The modulation removed signal is correlated with 
>> multiple predetermined data patterns to determine the data pattern of 
>> the signal. The code rate of the coded frame is derived based on the 
>> determined data pattern. The above arrangement is particularly suited to 
>> a digital satellite broadcast system.
> [snip]
>>> While STB0899 does not seem to have a register for
>>> enabling hierarchical BC mode, it still has a (read-only)
>>> "low priority stream detection" bit in register TSULSTKM.)
>>> Maybe this is a hint that it can detect hierarchical
>>> BC mode by itself.
> Well, the question is not how hierarchical modulation works,
> the question is if and how the demod can detect this
> automatically. The text from the patent you quoted doesn't
> shed any light on this, it says (if I understand it right)
> that one has to know that hierarchical modulation is used,
> then remove the HP part and then "correlate with multiple
> predetermined data patterns" (the ones given in EN 302 307
> annex F) to find the code rate.

The patent elsewhere says like this..

[0036] In addition, a lower layer transmits another data stream 
independent of the upper layer, whereby the lower layer can be Binary 
Phase Shift Keying (BPSK) modulated. The BPSK modulated data is 
converted to phase information associated with a predetermined angle 
.theta. by a converter 303. The angle .theta. can be determined based on 
the available power. The modulated upper and lower layers are then 
multiplied by a multiplier 305. According to one embodiment of the 
present invention, the multiplied signal is further pulse-shaped.

[0037] Signal acquisition involves determining whether a lower layer 
signal exists. If the lower layer signal is provided, the module 301 
then determines .theta.. According to an embodiment of the present 
invention, the lower layer is coded with Low Density Parity Check (LDPC) 
code. LDPC codes are typically rather long, for instance, of length 
64,800 bits per LDPC coded frame. In order to properly decode the LDPC 
code, the receiver needs to first achieve frame synchronization.

[0038] FIG. 4 is a diagram of an exemplary signal constellation used by 
the hierarchical constellation mapper of FIG. 2. As seen in FIG. 4, the 
two bits that are underscored represent the upper layer. The 
constellation 400 chooses two bits from the upper layer and one bit from 
the lower layer to select one of the points of the hierarchical 
constellation point. It is recognized that other equivalent labeling of 
the constellation can be utilized; for instance, the constellation can 
be labeled based on Gray labeling. In any configuration, to maintain the 
backward compatibility, the upper layer is mapped in the manner defined 
by the particular legacy system (e.g., DVB-S system).

[0039] Under the above arrangement, the upper layer can be received by 
any legacy receiver employing traditional digital receiver design 
technologies. In fact, the upper layer receiver does not need to be 
aware of the existence of the lower layer. The received signal is timing 
and carrier phase synchronized for the upper layer before it is passed 
to the upper layer forward error correction coding. Since the lower 
layer is subjected to the same timing and carrier phase impairments, the 
lower layer has the benefit of the achieved timing and carrier phase 
synchronization of the upper layer.

[0040] Before the lower layer LDPC decoder can function, the framing 
information of the LDPC code, which includes the starting point of a 
LDPC coded frame and the code rate of the LDPC code, has to be acquired. 
This aspect significantly differs from the upper layer processing. 
Firstly, the lower layer operates at low signal-to-noise ratios, 
typically, at Es/No below 0 dB. Secondly, the lower layer needs to 
recover the framing information to proceed to LDPC decoding, whereas the 
upper layer can be first decoded, as convolutional coding can be self 
synchronized without using any framing information. This suggests that 
upper layer framing recovery can benefit from the coding gain of the 
forward error correction coding. In contrast, the lower layer cannot 
benefit from the coding gain. Furthermore, in order to maintain backward 
compatibility, any framing information of the lower layer has to be 
carried strictly in the lower layer. At any given transmission epoch, 
both upper and lower layers are used to determine the signal point to be 
transmitted; thus, the transmission of known signal points for training 
the lower layer cannot be used.

> This would match my understanding as decribed in my previous
> mail, that the PLHEADER is not pi/2-BPSK modulated but part
> of the LP bitsream in hierarchical mode.


> (This is consistent with annex F, the input to the
> hierarchical mapper is a *bitstream*, not a pre-modulated
> signal. 

Yes, only the hierarchial mapper plays around with it. I made a nice 
diagram which makes it a bit more clear. The figure in  Annex F is a bit 
confusing (not saying that it is wrong, but it is confusing)

in pdf format: http://www.thadathil.net/dvb/hmod/illustration.pdf
in compressed SVG format: 
If someone has issues with mime types, the directory view 

> Also, I guess pi/2-BPSK PLHEADER would break
> DVB-S HP receiption.)

I don't understand why. maybe i didn't follow it clearly, maybe you can 
explain, why you felt that way.

>> For the STB0899 the information might not be needed, but could be needed 
>> by another demod some time later.
> possible
> Currently I think the demod could just try to find SOF pattern
> in pi/2-BPSK demodulated bitstream, and if that fails try
> to find one of the BC patterns in hierarchical demodulated
> LP-bitstream. If a particular demod cannot do this in firmware, but
> has a register to explicitely choose hierarchical BC mode, we
> could emulate it in dvb-core (similar to INVERSION_AUTO emulation).


>> I think for the time being we can go with the minimalist approach as it 
>> is in multiproto7
> I would do it that way, yes.



More information about the linux-dvb mailing list