Olivia MFSK is an amateur radioteletype protocol, utilizing a pair of frequency-shift keying (MFSK) and designed to work in sophisticated (low signal-to-noise ratio plus multipath propagation) conditions on shortwave bands. The signal might maybe well well moreover moreover be accurately got even when the surrounding noise is 10 dB stronger. It’s some distance in most cases used by amateur radio operators to reliably transmit ASCII characters over noisy channels utilizing the high frequency (3–30 MHz) spectrum. The efficient knowledge price of the Olivia MFSK protocol is 150 characters/minute.

Olivia modes are in most cases continuously known as Olivia X / Y (or, alternatively, Olivia Y / X ), where X refers again to the assorted of diverse audio tones transmitted and Y refers again to the bandwidth in hertz over which these indicators are unfold. Examples of total Olivia modes are 16/500, 32/1000 and 8/250.

## History

The protocol used to be developed at the tip of 2003 by Pawel Jalocha. The first on-the-air assessments had been performed by two radio amateurs, Fred OH/DK4ZC and Les VK2DSG on the Europe-Australia course within the 20-meter amateur band. The assessments proved that the protocol works effectively and can enable usual intercontinental radio contacts with as exiguous as one watt RF energy. Since 2005 Olivia has change true into a old model for digital knowledge transfer under white noise, fading and multipath, flutter (polar course) and auroral conditions.

## Voluntary channelization

Since Olivia indicators might maybe well well moreover moreover be decoded even when got indicators are extraordinarily frail, (signal to noise ratio of –14 dB),[1] indicators exact ample to be decoded are now and again under the noise flooring and subsequently no longer capacity to peek manually. As a result, amateur radio operators trust voluntarily decided upon channelization for this mode. This channelization enables even imperceptibly frail indicators to be properly tuned for reception and decoding. By total conference amateur stations provoke contacts utilizing both the 16/500 or 32/1000 modes after which swap to diverse modes to continue the dialog. The following desk lists the total heart frequencies utilized within the amateur radio bands.

Olivia Middle Frequencies [1]
Band 16/500 Frequencies (kHz) 32/1000 Frequencies (kHz)
160 meters 1808.75, 1809.25, 1840.75, 1841.25 N/A
80 meters 3577.75, 3583.25, 3522.75 3578.0, 3616.0, 3523.0, 3621.0
60 meters 5.3665 N.A
40 meters 7026.25, 7043.25, 7073.25, 7076.75 N/A
30 meters 10139.25, 10142.25, 10143.25 N/A
20 meters 14076.4, 14075.4, 14078.4 14106.5, 14107.5, 14108.5
17 meters 18103.4, 18104.4 N/A
15 meters 21087.25, 21087.75, 21130.25 21153.5, 21154.5
12 meters 24922.25 N/A
10 meters 28076.75, 28077.25 N/A
6 meters 50087.25, 50287.25, 50292.25 N/A
2 meters 144136.25 N/A

Following the introduction of new amateur radio digital modes and the following overlapping with the Olivia calling frequencies on some amateur radio bands, US radio amateur Tomas Hood NW7US consulted the members subscribed to an Olivia MFSK customers newsgroup and created a brand new list of calling frequencies in November 2017.

Olivia MFSK mode calling frequencies (Nov 2017)
Band Dial Frequency

(KHz in USB mode)

Exact Centre

Frequency (KHz)

New Olivia

Tones/Bandwidth

160 metres 1825.40 1826.90 Olivia 8/250
80 metres 3571.40 3572.90 Olivia 8/250
40 metres 7071.40 7072.90 Olivia 8/250
30 metres 10141.40 10142.90 Olivia 8/250
20 metres 14071.40 14072.90 Olivia 8/250
17 metres 18101.40 18102.90 Olivia 8/250
15 metres 21071.40 21072.90 Olivia 8/250
12 metres 24921.40 24922.90 Olivia 8/250
10 metres 28121.40 28122.90 Olivia 8/250

## Tones and bandwidth combos

Conversations utilizing Olivia are by conference initiated utilizing both Olivia 16/500 (16 tones over a 500 Hz bandwidth) or Olivia 32/1000 (32 tones over a 1000 Hz bandwidth).[1] As soon as communications trust been established, the communicating parties mutually opt if any other mode would greater suit the present propagation conditions. The which that it is probably you’ll have in mind more than just a few of tones that will moreover moreover be chosen are 2, 4, 8, 16, 32, 64, 128, and 256 with better numbers of tones giving more knowledge redundancy but slower throughput and lower numbers of tones giving faster throughput at the associated price of less redundancy. Available bandwidths for Olivia are 125 Hz, 256 Hz, 512 Hz, 500 Hz, 1000 Hz, and 2000 Hz with wider bandwidths giving faster throughput and narrower bandwidths giving slower throughput. The most continuously used combos are 4/125, 8/250, 8/500, 16/500, 16/1000, and 32/1000.

## Technical details

Being a digital protocol, Olivia transmits a lag of ASCII (7-bit) characters. The characters are sent in blocks of 5. Every block takes 2 seconds to transmit, thus the efficient knowledge price is 2.5 character/second or 150 characters/minute. The most accepted transmission bandwidth is 1000 Hz and the baud price is 31.25 MFSK tones/second. To accommodate for diverse conditions and for the motive of experimentation the bandwidth and the baud price might maybe well well moreover moreover be changed.

The Olivia transmission machine is constructed of two layers: the lower, modulation and forward error correcting (FEC) code layer is a classical a pair of frequency-shift keying (MFSK) whereas the easier layer is a forward error correcting code basically based mostly on Walsh functions.

Each layers are of the same nature: they constitute a “1-out-of-N” FEC code. For the principle layer the orthogonal functions are (co)sine functions, with 32 diverse frequencies (tones). At a given time ideally suited a kind of 32 tones is being sent. The demodulator measures the amplitudes of the entire 32 which that it is probably you’ll have in mind tones (utilizing a Fourier transform) and (vibrant that ideally suited one of those 32 might maybe well well moreover trust been sent) picks up the tone with the ideally suited amplitude.[2]

For the second FEC layer: every ASCII character is encoded as for sure one of 64 which that it is probably you’ll have in mind Walsh functions (or vectors of a Hadamard matrix). The receiver over again measures the amplitudes for all 64 vectors (here comes the Hadamard Changed into) and chooses the finest.[3]

For optimum performance the categorical demodulators work with soft decisions and the closing (mighty) map to decode a character is taken ideally suited at the second layer. Thus the principle layer demodulator indubitably produces soft decisions for every of the 5 bits associated to an MFSK tone as a replacement of merely picking up the ideally suited tone to kind mighty decisions for those 5 bits.

In direct to exercise some distance from easy transmitted patterns (bask in a fixed tone) and to sever support the chance for a faux lock at the synchronizer the characters encoded into the Walsh feature lumber through a scrambler and interleaver. This stage merely shifts and XORs bits with predefined scrambling vectors and so it doesn’t enhance the performance where the white (uncorrelated) noise is anxious, however the following pattern positive aspects sure positive traits which are of sizable wait on to the synchronizer.

The receiver synchronizes robotically by shopping through which that it is probably you’ll have in mind time and frequency offsets for an identical pattern. The frequency search differ is on the total +/- 100 Hz but might maybe well well moreover moreover be as high as +/- 500 Hz if the user needs so.

### The MFSK layer

The default mode sends 32 tones interior the 1000 Hz audio bandwidth and the tones are spaced by 1000 Hz/32 = 31.25 Hz. The tones are formed to sever support the amount of vitality sent outside the nominal bandwidth.

The right kind form formula is:

${displaystyle +1.0000000000+1.1913785723cos(x)-0.0793018558cos(2x)-0.2171442026cos(3x)-0.0014526076cos(4x)}$

where x ranges from – π to π.

A net site of the window (“form formula”)

The coefficients articulate the image form within the frequency domain and had been calculated by a minimization direction of which sought to kind the smallest crosstalk and the smallest frequency spillover.

The tones are sent at 31.25 baud or every 32 milliseconds. The allotment is no longer preserved from one tone to the subsequent: as a replacement a random shift of ±90 levels is launched in direct no longer to transmit a pure tone when the the same image is many cases sent. Because the symbols are smoothly formed there might maybe be no must always exercise the allotment fixed, which on the total is the case when no (e.g. sq.) shaping is used.

The modulator makes spend of the Gray code to encode 5-bit symbols into the tone numbers.

The waveform generator is basically based mostly on the 8000 Hz sampling price. The tones are spaced by 256 samples in time and the window that shapes them is 512 samples lengthy. The demodulator is basically based mostly on the FFT with the size of 512 functions. The tone spacing in frequency is 8000 Hz/256 = 31.25 Hz and the demodulator FFT has the resolution of 8000 Hz/512 = 15.625 Hz thus half of of the tone separation.

To adapt the machine to diverse propagation conditions, the assorted of tones and the bandwidth might maybe well well moreover moreover be changed and the time and frequency parameters are proportionally scaled. The more than just a few of tones might maybe well well moreover moreover be 2, 4, 8, 16, 32, 64, 128 or 256. The bandwidth might maybe well well moreover moreover be 125, 250, 500, 1000 or 2000 Hz.

### The Walsh functions FEC layer

The modulation layer of the Olivia transmission machine within the default mode sends for sure one of 32 tones at a time. Every tone thus constitutes a image that carries 5 bits of recordsdata. For the FEC code, 64 symbols are taken to salvage a block. Inside of every block one bit out of each image is taken and it forms a 64-bit vector coded as a Walsh feature. Every 64-bit vector represents a 7-bit ASCII character, thus every block represents 5 ASCII characters.

This map, if one image (tone) becomes corrupted by the noise, ideally suited one bit of each 64-bit vector becomes inferior, thus the transmission errors are unfold uniformly across the characters interior a block.

The two layers (MFSK+Walsh feature) of the FEC code might maybe well well moreover moreover be handled as a two dimensional code: the principle dimension is formed along the frequency axis by the MFSK itself whereas the second dimension is formed along the time axis by the Walsh functions. The two dimensional affiliation used to be made with the foundation in thoughts to resolve such organized FEC code with an iterative algorithm, nonetheless, no such algorithm has yet been established.

The scrambling and straightforward bit interleaving is utilized to kind the generated image patterns appear more random and with minimal self-correlation. This avoids faux locks at the receiver.

Bit interleaving: The Walsh feature for the principle character in a block is constituted of the first bit of the first image, the 2nd bit of the 2nd image, and so on. The 2nd Walsh feature is constituted of the 2nd bit of the first image, the Third bit of the 2nd image, and so on.

Scrambling: The Walsh functions are scrambled with a pseudo-random sequence 0xE257E6D0291574EC. The Walsh feature for the first character in a block is scrambled with the scrambling sequence, the 2nd Walsh feature is scrambled with the sequence rotated true by 13 bits, the Third with the sequence rotated by 26 bits, and so on.

## Samples

The listed audio recordsdata both are encoded with the message: “Welcome to Wikipedia, the free encyclopedia that someone can edit.”

## Contestia

Contestia is a digital mode that is right away derived from Olivia, which is no longer as tough. It’s some distance more of a compromise between performance and creep. It sounds and looks nearly the same to Olivia, and might maybe well well moreover moreover be configured in as many ways, but has twice the price. The mode used to be developed by Prick Fedoseev, UT2UZ, in 2005.

The Contestia mode has 40 codecs factual bask in Olivia – just a few of which are belief about frequent and so that they all trust diverse traits. The codecs trust a variation in bandwidth (125, 250, 500, 1000, and 2000 Hz) and more than just a few of tones used (2, 4, 8, 16, 32, 64, 128, or 256).

The identical old Contestia codecs (bandwidth/tones) are 125/4, 250/8, 500/16, 1000/32, and 2000/64. The most continuously used codecs are 250/8, 500/16, and 1000/32.[4] The increased creep of Contestia is achieved by utilizing a smaller image block size of (32) slightly than Olivia (64) and by utilizing a 6-bit decimal character location slightly than 7-bit ASCII location which Olivia makes spend of.

This diminished character location doesn’t print out in both upper and lower case (such as RTTY). Some traffic nets might maybe well well moreover no longer take grasp of to spend this mode because it doesn’t give a enhance to upper and lower case characters and extended characters came upon in many documents and messages. This doesn’t pose a field for accepted digital chats interior ham communications.[5]