RTL-SDR for Shortwave Radiogram – Telecommunications, Navigation & Electronics

Introduction

Recently I read several posts on various amateur radio forums regarding the reception of Shortwave Radiogram broadcasts. VOA Radiogram was a Voice of America program that experimented with various digital transmission modes, starting in March 2013. It was superseded by Shortwave Radiogram (Ref.1). I decided I would try reception using the very popular RTL-SDR receiver & Nano up converter. For digital decoding, I used MULTIPSK (Ref.2). These are my results.

Transmission Schedule

Figure 1 shows the transmission schedule taken from the Shortwave Radiogram web site (Ref.3). There are two transmitters, WRMI in Miami Florida and WINB in Red Lion Pennsylvania. Various transmitter powers and antennas are used depending on time & target areas.

Fig.1 Shortwave Radiogram Schedule on WRMI & WINB

MFSK

MFSK or Multi Frequency Shift Keying is a powerful HF High Frequency digital transmission mode developed by Murray Greenman ZL1BPU and Nino Porcino IZ8BLY (Ref.4). It is essentially an extension of RTTY and has its origin in older HF teleprinter systems such as Piccolo and Coquelet. Instead of mark & space tones, there are multiple tones. MFSK has been tested with various numbers of tones, up to 64. The most popular versions are:
MFSK16: SymbolRate=15.625bd/16FSK/BW=316Hz
MFSK32: SymbolRate=31.250bd/16FSK/BW=630Hz
MFSK64: SymbolRate=62.500bd/16FSK/BW=1260Hz

Instead of representing just one data bit, each tone represents one symbol which in turn represents 2^N bits (3,4,5). Tones are separated by the symbol rate, which means they are located at each others’ data nulls. MFSK allows good performance over HF radio paths that have considerable multipath delay or Doppler shift. FEC or Forward Error Correction is added to further improve performance. Since only one tone is on at a time, the transmission has a constant envelope which reduces the need for ultra linear amplifiers.

In simulation experiments between RTTY, BPSK31 and MFSK16, I found the following comparative results for threshold CER Character Error Rates for additive Gaussian Noise (Ref.5). MFSK clearly has a tremendous advantage:
RTTY45: SNR=8dB, CER=0
BPSK31: SNR=2dB, CER=0
MFSK16: SNR=-7dB, CER=0

Figure 2 shows block diagrams of the various MFSK modes. Keyboard data is first encoded from ASCII to varicode. Next FEC bits are added in a R=1/2 convolutional encoder and then interleaved. Finally the bits are Gray encoded into symbols that modulate a phase continuous VCO. For example in MFSK32, varicode data is read out at 62.5bits/sec. In the varicode alphabet, characters can have as few as 4bits “e” or as many as 11bits “EOT”. Let’s say on average 7 bits/char, then the keyboard data is 9char/sec, at 5char/word, or about 110wpm. Figure 3 is a MFSK32 wav file, sounding very much like a Piccolo, hence the name of the 1960’s system.

Fig.3 MFSK32 wav File Sounds Like a Piccolo

RTL-SDR Shortwave Radiogram Reception

Following the schedule in Figure 1, I was able to receive Sunday Feb 28th at 2330UTC/7780KHz and Saturday Mar 6th at 0100UTC/9955KHz. These are my results for Mar 6th. Figure 4 shows the SDR# receive window for the RTL-SDR at 9955KHz. Note the strong signal. Figure 5 shows the Multipsk reception of the program schedule sent in MFSK32. Figure 6 shows the Multipsk reception of the technical article in MFSK64, Figure 7 the Multipsk reception of an HTML doc in MFSK64 and finally Figure 8 the Multipsk reception of a picture.