jvankatwijk / drm-receiver Goto Github PK

DRM (Digital Radio Mondiale) is - as the name suggests - a form of digital radio. In Europe DRM is not very popular, most stations that started a DRM transmission in the first decade of the century already stopped. However, Rumenia and Kuwait still have a regular DRM transmission that I can receive, the website "http://www.hfcc.org/drm/" gives an overview.

The DRMm receiver is an experimental tool, a heavily reduced swradio, with a single decoder, just for DRM. The decoder uses the FDK_AAC library for AAC decoding, meaning that xHE-AAC can be decoded next to AAC.

The decoder takes a samplerate of 12000 Samples/second as input.

The incoming sample stream is best seen as a sequence of words, for mode B, such a word contains just over 300 samples.

Apaprt from the time synchronization, i.e. finding in the input stream where the words start, all processing is done in the frequency domain. The incoming words are processed by an FFT processor, and each incoming word is eventually translated into a word containing just over 200 carriers (The FFT is done on segments of 256 samples). These translated words are grouped into frames (for mode B a frame contains 15 words) and each group of 3 frames is grouped into a superframe.

The layout of frames and superframes is defined in the standards, i.e. it is defined which carriers are used for synchonization, which carriers belong to the FAC (Fast Access Channel), which carriers belong to the SDR (Service Description Channel) and which carriers contain the payload.

The decoder gives quite some information on the state of decoding. The 4 boxes top right tell about the synchronization reached, from top to bottom

• time sync tells whether or not the software thinks the input blocks in the input sample stream are identified;
• FAC sync tells whether or not decoding the FAC is successfull. The FAC contains specified carriers, with data encoded as QAM4, and it gives general information on the structure of the transmission;
• SDC sync tells whether or not decoding the SDC is successfull. The SDC contains - as the name suggests - data telling where to find the sevice data and how to interpret it. SDC data is encoded as QAM4 or QAM16; *aacSync tells about the decoding of the selected service. While the previous ones dealt with the whole content of the transmission, the selected service may be just one of 4. The payload of the transmision is encoded in QAM16 or QAM64 format. The audio services are encoded as AAC streams or xHE-AAC streams, that is why this software uses the FDK_AAC library, since that is able to handle both formats.

The boxes, jer elabeled "3" and "c" tell that in the transmission the mode "c" was detected and the spectrum layout "3" (the latter just indicating a 10 KHz wide signal). The box with "QAM16" tells that the audio content is encoded in QAM16, i.e each complex carrier takes one of 16 positions (which can be decoded into 4 bits), and the box with "AAC" tells that the audio was encoded in AAC format (the other formak is xHE-AAC).

The 3 numbers in the middle of the widget give an indication of the quality of the decoding.

Top left of the widget the frequency offset is indicated, the total offset can be obtained by adding the two numbers.

The label tells that the picture is taken from a (recorded) transmission of "the voice of Nigeria"

The "scope" with the yellow arrows gives an impression on the amount with which the signal has to be corrected. In the frequency domain signal, some carriers are known to have a predefined phase and amplitude, i.e. pilots. The scope tells the correction to be applied on these pilots, information that can be used to transform all carriers.

The second picture shows - from a different recorded transmission - the so-called channel. On the way from transmitter to receiver the signal is malformed, in the picture the impulse response of the channel, i.e. the way the channel filters the transmission , is shown.

The small black scope shows the constellation of the decoded data. In the first pisture the signal was QAM16, in the last two pictures a QAM64 encoding is used. The last picture shows an almost ideal signal, all 64 positions the carriers may take are clearly identifyable.

For building an executable the - more or less - standard libraries are supposed to be installed. Qt, Qwt, fftw3, libsamplerate and libsndfile and of course, the support libraries for the devices one chooses.

The estimator computes the difference between the pilot values as seen in the data and the pilot values as they should be.

F_p is initialized with the precomputed values and is
matrix filled with the (pilot, tap) combinations, where for the
pilots, their carrier values (relative to 0) are relevant
Basically the FFT transformer, with positions related to
the positions of the taps

The approach for channel estimation is as follows:
we call the channel in the time domain h_td,
in the frequency domain that will then be h_fd = F x h_td
we observe the values of the pilots at the recoever
side, X_p, It is known that they were transmittes as s_p
We know then that
X_p = diag (s_p) * f, where f = F_p * h_td
so, X_p = S_p * F_p * h_td
with S_p is diag (s_p), and F_p is the FFT transform of the channel
S_p en F_p can be computed, since they are known, regardless
A_p = S_p * F_p
the equation then is X_p = A_p * h_td
we solve that as h_td = A_p ^ -1 x X_p
recall that h_td is the chaannel in the time domain, so
h_fd = F_p * h_td is the solution

Estimator_1 is the simplest one, it works though. Estimator_3 uses the "eigen" template library for the matric operations.

The equalizer itself is indirectly derived from the one in diorama, which was aMatlab implementation, and was translated (transltterated) by M. Bos in his package RXADADRM.

For Windows an installer is available, for Linux an AppImage. The AAC decoder for the Windows version is the fdk-aac library, for the Linux one needs to install the libfdk-aac library is used.

Of course, one may create an executable. Note that to keep things simple the only path is using qmake/make.

By setting some parameters in the ".pro" file, one may choose between aac decoding using the "faad_drm" library or the "fdk-aac" library.

The latter has the (potential) advantage of being able to decode xHE-AAC as well

Since DRM is (mainly) transmitted on shortwave, there is support for devices with which shortwave reception is possible

• the SDRplay, in the drm-receiver.pro file one may choose between using the "old" 2.13 interface library, or the 3.0X library;

• the Hackrf.

• the DABsticks, for RT820 based sticks with a driver that allows selecting frequencies as low as 13 MHz. Note that direct sampling gives real rather than complex samples and is not useable.

Copyright

    Copyright (C)  2019, 2023
    Jan van Katwijk ([email protected])
    Lazy Chair Computing

    The drm receiver software is made available under the GPL-2.0. and
    is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU General Public License for more details.