WO2005088998A1

WO2005088998A1 - Transmission of data over a radio frequency voice communication channel

Info

Publication number: WO2005088998A1
Application number: PCT/ZA2004/000090
Authority: WO
Inventors: Russel Ian Hawkins; Donovan Martin; Deon Rothmann Toerien
Original assignee: Ip Works (Proprietary) Limited
Priority date: 2004-03-18
Filing date: 2004-08-03
Publication date: 2005-09-22

Abstract

Techniques for transmitting digital data over radio frequency voice transmission channels are disclosed, as well as techniques for processing the received radio frequency signal. In one embodiment the data is frequency shift keyed after being subjected to forward error correction. After filtering and impedance matching it is transmitted. The received signal is filtered, subjected to full wave rectification, enveloped detection and ADC sampling. In another embodiment the digital signal is subjected to forward error correction and then audio data packing before being transmitted. In this embodiment the one binary digit is represented by a tone of predetermined frequency, and the other binary digit by white noise.

Description

TRANSMISSION OF DATA OVER A RADIO FREQUENCY VOICE COMMUNICATION CHANNEL

FIELD OF THE INVENTION THIS INVENTION relates to the transmission of data over a radio frequency voice communication channel.

BACKGROUND TO THE INVENTION The voice communication channels of mobile (cellular) networks have limited band width. Band width is an indication of the frequencies that the channel can transmit. The range of the human voice is from about 1 kHz to about 4kHz. Voice communication channels have a band width of from about 800 Hz to about 4kHz and can thus cope with normal speech.

A GSM module is the major component of a cellular telephone. It has two main functions. It processes speech, which is in the form of an analogue, that is, wave form signal, picked-up by the cellular telephone's microphone and converts it into a digital signal which is then transmitted at radio frequency. Conversely it processes incoming digital signals at radio frequency and converts the digital signal back to analogue form. It is this signal which is used to drive the speaker of the cellular phone.

The digital signal sent at radio frequency by a cellular telephone is not transmitted as a continuous stream. Using a combination of known modulation techniques, the telephone breaks the digital signal into a multitude of "packets" of information which are then transmitted. The conventional GSM modem can transmit at different frequencies and on up to eight different time slots. The packets of information are scattered over different channels and interspersed with packets of information from other conversations. Thus many conversations can take place simultaneously on the same channel. This is referred to as "TDMA" - time division multiple access. When a call is instituted a frequency is allocated and also a time slot. The allocated time slot repeats at very short intervals. The speech transmitted is that which is "heard" by the cellular telephone during each time slot.

The packets of information transmitted in respect of one conversation must, at the receiver, be separated from packets of information pertaining to other conversations. This is achieved by preceding each packet of information with a distinguishing header that can be detected. The packets are then reassembled electronically into approximately the form of the original digital signal, converted to analogue form and fed to the speaker.

Some degeneration of speech signals inevitably occurs due partly to the fact that only short bursts of the conversation are transmitted. Various techniques are used to reduce what is known as the Bit Rate Error (BER). However, even with this correction the received speech is not of the quality of the original. However, for speech transmission purposes the degradation is not such as to cause a significant problem. The so-called VOCODER (voice encoder) module of the telephone checks the transmission, predicts what is missing from a transmission, and attempts to correct the received signal.

Dedicated radio frequency data transmission channels exist. They have a band width of up to about 13kHz. There is no conversion from analogue to digital form and no reconstruction of the received signal. The signal is transmitted as a bit stream. However, using such dedicated channels is far more expensive than using voice communication channels.

Voice communication channels are deemed unsuitable for data transmission for two main reasons. These are:

1. The signal processing which occurs means that the signal received differs from the signal sent. As explained this is not significant in voice communication but is significant in data communication as it means that data is lost or corrupted. The human voice is by nature slow changing which means that certain predictive and corrective algorithms can be employed to reconstruct "lost speech". It is also possible to filter out background noise. It is the way the CODEC (encoder / decoder) and the VOCODER of the GSM module handle information and performs their signal correction function that results in data being lost or corrupted.

2. The narrow band width inhibits transmission of data at high speed. United States Patents 6,144,336 and 6,493,338 disclose prior proposals for improving the transmission of data over wireless communication channels.

The present invention seeks to provide a method of, and system for, enabling data transmission of acceptable quality to take place over a voice communication channel.

BRIEF DESCRIPTION OF THE INVENTION According to one aspect of the present invention there is provided a method of transmitting digital data over a radio frequency voice communication channel which comprises employing a tone of frequency f1 to represent either digital "1" or digital "0", and white noise to represent the binary digit not represented by the tone of frequency f 1.

According to a further aspect of the present invention there is provided a method of transmitting digital data over a radio frequency voice communication channel which comprises employing the technique of audio data packing thereby to obtain a service of discrete values which represent the analogue wave form and transmitting each value as a frame of digital data.

The received radio frequency signal transmitted as defined in the preceding paragraph is filtered, subjected to spectral power measurement to determine the amplitude shift keying spectral power and recovering the digital data by using the method known as early, late gate timing data recovery.

According to another aspect of the present invention there is provided a method of transmitting digital data over a radio frequency voice communication channel which comprises encoding the data to be transmitted with correction parity bits, subjecting the encoded data to frequency shift keying to modulate the data onto to the GSM voice communication channel, filtering the signal so that the envelope of the signal is of cosine shape, matching the impedance of the data signal to that of the GSM audio signal channel, and transmitting the data carrying signal over the voice communication channel.

The received radio frequency signal transmitted as defined in the preceding paragraphs is filtered to remove signals other than the amplitude shift keying carrier, full wave rectifying the signal to provide a dc signal, subjecting the rectified signal to envelope detection and subjecting the signal to ADC sampling thereby to digitize the received signal.

According to yet another aspect of the present invention there is provided a method of manipulating a stream of raised root cosine pulses that have been transmitted over a cellular network voice communication channel, the method comprising filtering the pulse stream to attenuate the noise component of the pulse stream, subjecting the pulse stream to full wave rectification, feeding the rectified signal to an envelope detector to create a near square wave pulse stream, and feeding the near square wave pulse stream to a threshold detector to shape the pulses to square form.

According to a still further aspect of the present invention there is provided apparatus for transmitting data over a voice communication channel, the apparatus comprising an ASK modulator for modulating a digital bit stream to produce a pulse stream consisting of raised root cosine pulses and a GSM module to which the pulse stream is fed for transmission over a voice channel of a cellular network.

BRIEF DESCRIPTION OF THE DRAWINGS For a better understanding of the present invention, and to show how the same may be carried into effect, reference will now be made, by way of example, to the accompanying drawings in which:-

Figure 1 diagrammatically illustrates data to be transmitted; Figure 2 is a block diagram of a master / slave processor arrangement; Figure 3 is a block diagram of a DSP processor arrangement; Figure 4 is a block diagram illustrating the processing stages to which the binary input signal is subjected prior to transmission; Figure 5 illustrates the effect of adaptive filtering on FSK tones; Figure 6 illustrates a wave form before and after matched filtering; Figure 7 illustrates discrete data values representing an ASK (amplitude shift keying) carrier; Figure 8 illustrates the demodulation of the signal after transmission at radio frequency; and Figures 9 to 20 illustrate waveforms at various stages of modulation and demodulation.

DETAILED DESCRIPTION OF THE DRAWINGS In accordance with the ASCII system the numeral 1 in binary code is represented by: 00110001

In electronic form a "zero" is represented by a signal of a specific value and a "one" by a signal of a different value. Thus in the upper half of Figure 1 , zeros are shown as low value signals and ones as higher value signals. The start point of the transmission of this signal is indicated at S and the end point at P.

Each signal is of a predetermined duration measured on the horizontal x axis, of predetermined amplitude as measured on the vertical y axis, and of preselected frequency.

There are various modulation techniques available which convert the binary signal to a modulated radio frequency signal that can be transmitted. Some techniques are Frequency Shift Keying (FSK), Amplitude Shift Keying (ASK) and Pulse Code Modulation (PCM).

For the purposes of the description of the upper row of blocks in Figure 4, Amplitude Shift Keying will be described. Using this system a binary 1 is represented by a specific frequency. To represent a binary "0" the carrier is turned off or drastically attenuated to accommodate the PDC (Linear predictive coding) algorithms in the VOCODER (voice encoder) module.

The lower portion of Figure 1 diagrammatically illustrates the representation of the digit "1" in binary form after amplitude shift keying has taken place. The frequency chosen is one which lies in the human speech spectrum at a point where human speech is dominant. Experimental work has shown that 2 to 3 kHz is a suitable frequency range. In this frequency range the effect of the VOCODER is less than at other frequencies.

Referring now to Figure 2, the reference numeral 10 designates the user terminal which is a computer or other data storage device. It is digital data on this which is to be transmitted to a remote computer or data storage device.

The user terminal 10 is connected via an industry standard RS232 link to a master processor 12 (Figure 2). The data, after processing by the processor 14 via an SPI (serial peripheral interface) bus. The slave processor performs the FEC (forward error correction) and modulation steps. Two processors are used to obtain the requisite processing power. The block 16 represents modulation circuitry and 18 represents the GSM module (global system for mobile communications module). The TTL (Transistor Transistor Logic) link indicates that in the processors the voltage levels are in the range of 0 to 5 volts or 0 to 3 volts.

In the alternative embodiment of Figure 3, the user terminal 10 is connected to a digital signal processor 22 in which signal processing, that is, modulation, demodulation and FEC are performed by the DSP (digital signal processing) software running on the digital signal processor.

Figure 2 relates to the top line of blocks in Figure 4 and Figure 3 relates to the lower line of blocks in Figure 4.

In the upper line of blocks of Figure 4, the precessing of a digital input signal from the user terminal 10 to a GSM module is illustrated and the captions indicate the stages of processing. In the lower line of blocks of Figure 4 and alternative method of processing the digital data is illustrated. The various processing stages will now be described with reference to the upper line of blocks.

User serial data in RS232 format is received in the master processor. The data is encoded with correction parity bits. The techniques for forward error correction (FEC) are known. Two techniques which can be used are the Verterbi and Reed Solomon techniques.

Once the data has been encoded it is subjected to amplitude shift keying (ASK) which modulates the data onto the GSM voice communication channel. ASK is preferred to the more commonly used FSK (frequency shift keying) because of the nature of the GSM voice compression algorithms. One of these algorithms is the one which performs adaptive filtering and which constantly adapts itself to the incoming frequencies. If the FSK technique is used and a constant tone f1 representing a digital "1" is used, then the adaptive filter adapts its filtering parameters to this tone and the tone is passed and not filtered out. This is illustrated in Figure 5 where the characteristics of the tone f1 are shown by the full lines and the envelope of frequencies that the filter passes by the dotted line. If a second tone f2 representing digital "0" is fed to the filter it responds slowly to the new tone as it is outside the bell-shaped filter curve shown in dotted lines. This results in the second tone f2 being filtered out and hence lost.

If a tone at frequence f1 is used for digital "1" and white noise for digital "0" then the effect of the adaptive filter can be overcome. At 1 ahead of the channel coding, the signal is as shown in the upper half of Figure 1. At 3, after ASK modulation, it is as shown in the lower half of Figure 1.

After ASK processing the tone signal at frequency f1 it is shown on the left hand side of Figure 6. By passing it through a matched filter the frequency envelope is set to predetermined parameters and shaped so that it has the configuration shown on the right of Figure 6. The envelope is of cosine shape which conforms to commonly used matched filter designs. Such filters are used on both the transmission and receiver side. The shaping reduces the bandwidth requirements of the transmission channel. The next stage is impedance matching which ensures that the impedance of the data signal to be transmitted matches that of the GSM audio channel input thereby achieving maximum power transfer of the modulated audio signal by the GSM module.

Input data in RS232 digital form, in accordance with the embodiment of the invention illustrated in the lower part of Figure 4, is processed as follows. The processor (Figure 3) now handles all the functions handled by the slave processor of Figure 2. The first stage, as described above, is FEC using, for example, the Verterbi or Reed Solomon techniques.

In the embodiment being described, the technique known as audio data packing is employed. The wave form (the left hand side of Figure 6) representing digital "1" is represented by a series of data values which are derived from the wave form as illustrated in Figure 7. These values are created at regular intervals and simulate the analogue signal. It ensures that the frequency of interest is retained and unwanted harmonics are eliminated. This is illustrated in Figure 7 where the instantaneous values marked t1 to t6 represent the wave form illustrated. Each discrete value is transmitted in the form of a frame of digital information with headers etc. which enable the frame to be identified.

The algorithm which runs in the GSM module was written on the basis that audio data will reach it continuously, or at least at regular intervals. The effect of this algorithm is known as regular pulse exited long term prediction (RPE-LTP). Only in these conditions does the algorithm function correctly. When data is being sent in ASK format, there are signal bursts each representing digital "0" and periods of "silence" representing digital "1". By generating "white" noise in the block designated "zero mean PCM data (WGN)" and adding this onto the carrier signal, the algorithm detects the continuous input of audio data that it is written to handle and thus continues to function.

Turning now to the top line of blocks in Figure 8, recovery of the transmitted data received by the remote GSM module will now be described. In Figure 9 the data carrying signal transmitted by the system represented by the top line of blocks in Figure 4 is illustrated after processing in the GSM module.

The filters of the block to which the incoming signal of Figure 9 is fed comprise matched band pass filters which permit the ASK carrier to be received, and passed for further processing whilst minimising GSM noise and other signal induced noise. The block comprises low noise operational amplifiers in a multi-feed back filter arrangement.

Figure 10 is representation of the frequency spectra illustrating the predominance of the 2 to 3 kHz frequency selected.

The filtered ASK signal is then full wave rectified by taking the zero mean ASK sinusoid and converting it to a DC component signal. Full wave rectification is used as opposed to half wave rectification because the GSM channel has a bandwidth limited to 4kHz and maximum use has to be made of the number of cycles of the ASK carrier. The signal after processing in the full wave rectifier is as shown in Figure 11.

After rectification the signal is subjected to the procedure known as envelope detection. The waveform created is based on the extremities of the rectified signal and the resultant waveform is an approximation of the original waveform transmitted by the transmitting GSM module. The wave form of Figure 11 after envelope detection is shown in Figure 12.

The wave form produced by the envelope detection procedure is digitized by the procedure known as ADC sampling. The digital signal is produced and then stored in memory for processing by the signal processing and channel decoding circuits. Figure 13 illustrates ADC sampling.

The first operation is to filter the signal to further reduce noise and enhance symbol detection. By nature the GSM audio channel is very dynamic in terms of data delivery after transmission and the signal processing and channel decoding block accommodates this by utilizing a method known as early, late gate symbol timing. The final processing is to decode it using either a Verterbi or Reed Solomon decoder to extract the data. The upper half of Figure 15 illustrates the signal at point 4, that is between envelope detection and ADC sampling. The lower half of Figure 16 represents the signal after final processing, that is, at point 6. Turning finally to the lower line of blocks in Figure 8, the four tap filter is provided to filter out GSM channel noise. As the data was transmitted as a stream of digital data initially the signal to noise ratio is better than the same ratio in the embodiment represented by the upper blocks of Figure 8. A four tap filter is chosen as the length of the filter is a compromise between filter response and data delay. The stream of digital data transmitted represented the discrete value points shown at t1 to t6 in Figure 7.

Once a frame of data has been received, it is subjected to a spectral power measurement to determine the ASK spectral power. This has the benefit that the effects of the RPE-LTP can be detected. The data frame is then adjusted according to a threshold table in memory.

The data frame is now further analysed to look for the characteristic ASK carrier representing the received digital data. These symbols are then extracted as in the first embodiment described by utilizing similar early, late gate timing recovery due to the dynamic nature of the received data (induced by the GSM module).

Once the data is stored in memory it is now ready to be processed. The information is filtered again to increase symbol detection and remove noise. Once the symbols have been extracted, it is then decoded by using either a Verterbi or Reed Solomon decoder to extract the data.

Claims

CLAIMS:

1. A method of transmitting digital data over a radio frequency voice communication channel which comprises employing a tone of frequency f1 to represent either digital "1" or digital "0", and white noise to represent the binary digit not represented by the tone of frequency f1.

2. A method of transmitting digital data over a radio frequency voice communication channel which comprises employing the technique of audio data packing thereby to obtain a series of discrete values which represent the analogue wave form and transmitting each value as a frame of digital data.

3. A method of transmitting digital data over a radio frequency voice communication channel which comprises encoding the data to be transmitted with correction parity bits, subjecting the encoded data to frequency shift keying to modulate the data onto the GSM voice communication channel, filtering the signal so that the envelope of the signal is of cosine shape, matching the impedance of the data signal to that of the GSM audio signal channel, and transmitting the data carrying signal over the voice communication channel.

4. A method according to claim 3, wherein the received signal radio frequency signed is filtered to remove signals other than the amplitude shift keying carrier, full wave rectifying the signal to provide a dc signal, subjecting the rectified signal to envelope detection and subjecting the signal to ADC sampling thereby to digitize the received signal.

5. A method as claimed in claim 2, wherein the received signal is filtered, subjected to spectral power measurement to determine the amplitude shift keying spectral power and recovering the digital data by using the method known as early, late gate timing data recovery.

6. A method of manipulating a stream of raised root cosine pulses that have been transmitted over a cellular network voice communication channel, the method comprising filtering the pulse stream to attenuate the noise component of the pulse stream, subjecting the pulse stream to full wave rectification, feeding the rectified signal to an envelope detector to create a near square wave pulse stream, and feeding the near square wave pulse stream to a threshold detector to shape the pulses to square form.

7. Apparatus for transmitting data over a voice communication channel, the apparatus comprising an ASK modulator for modulating a digital bit stream to produce a pulse stream consisting of raised root cosine pulses and a GSM module to which the pulse stream is fed for transmission over a voice channel of a cellular network.