GB2157129A - Conference circuit - Google Patents

Abstract

A simple and economic implementation of a conference facility is provided in a distributed telephone exchange, such as an PABX, solely by equipment at the nodes. It includes samples stores (SS1 ... SSN) connected to the transmission medium (1) so that a number of PCM form samples can be received from the other nodes in the conference. These samples pass via a multiplexer (3) to a digital-to-analogue converter (4) and therefrom via a low-pass filter (5) to the handset. Thus the output to the handset is an analogue signal representing the combined (but not summated) effects of the samples from the other conferees. <IMAGE>

Description

SPECIFICATION Conference circuit The present invention relates to the provision of conference facilities in a distributed telephone exchnage, especially (but not only) a PABX.

Conferencing in digital systems, such as PABX's, is often done by a conference bridge which "listens" to all transmitted speech samples from the participating extensions. By some algorithm based on amplitude, one of these samples is selected for sending to the participating extensions. In a distributed system, such a facility can be provided by the use of one or more conference bridge nodes in the system.

An object of the present invention is to provide a system in which such special nodes are not needed.

According to the invention, there is provided a method of providing a conference facility in a telephone system of the decentralised type, in which at a subscriber's node the speech samples from the subscriber's lines involved in the conference other than from the line to which that node belongs are stored temporarily in sample stores, so that for a conference which involves (N + 1) lines N such stores are used, in which the samples are read out from the stores via a multiplexer to a decoder such that the output of the decoder corresponds to the combination of the samples, and in which the result of said combination, which is in analogue form, is supplied to the node's transudcer.- An embodiment of the invention will now be described with reference to the accompanying drawing, which is a highly simplified block diagram of that part of a subscriber's node which is relevant to the invention.

This invention is based on the principle of "time- sharing" the decoder between two or more sets of received speech samples, whereafter the output is filtered in the same way as is normally done. Hence a conference with three participants results in each extension receiving two speech samples per frame.

These speech samples are applied to the decoder within one frame, such that the output of the decoder consists of the decoded value of the first sample part of the time frame, and the decoded value of the second sample for the rest of the time frame. Similarly, for a four-way conference, three speech samples, one from each of the other participants, are applied to the decoder in one time frame, each for its own part of that frame. The resultant output of the decoder is filtered to produce the sum of all the input speech samples.

A disadvantage of such an arrangement is that there is an increase in the out-of-band noise, which may need an improved filter. However, the noise level, as will be seen later is within acceptable levels for conferences of three or four extensions, which are the commonest ones needed.

We first discuss the theory on which are arrangement is based. We consider a rectangular pulsed waveform which consists of a unity amplitude pulse of duration T and repetition period T, and perform a Fourier transform on it, so that it can be described by the following equation:

The output of a decoder receiving speech samples from several sources of which only one is active can be regarded as the pulsed waveform multiplied by the speech waveform. If the modulation function is denoted by fm(t), the expression can be expanded to give:

Since in this case liT = 8KHz (the sampling frequency used), the resultant equation shows an atten- uated version of the baseband signal, plus two double-sideband suppressed carrier modulated signals centred at 8KHz and 16KHz respectively.There are similarly terms at 24KHz and 32KHz, etc., but these are outside the audible range. Thus the ratio of in-band signal power to out-of-band signal power is constant, since fm(t) is a common factor. The out-of-band noise is filtered to leave a predominant base-band sig nal.

If other channels are active at the same node, they produce a similar result, and so produce a linear addition to the baseband signal, and also to the out-of-band noise. This occurs when two or more people start to speak together or one tries to interrupt the other. This has the advantage that all parties speaking can be heard and may even be recognisable, which is useful if a person of greater seniority wishes to speak. Normally, however, only one party speaks at a time.

It can be seen that TiT is inversely proportional to the number of other parties, 1/N-l, where N eq the total of parties involved, so the degradation of the signal-to- noise ratio depends on the value of N.

Table 1 shows the values of the amplitudes of the in-band and out-of-band signals.

TABLE 1 No. of Speech Samples 2 3 4 5 Base-band Amplitude .500 .333 .250 .200 First Harmonic Level .637 .827 .900 .935 Second Harmonic Level 0 .413 .613 .757 To find the degradation in signal-to-noise ratio, the power of the in-band signal is divided by the sum of the powers of the two out-of-band signals, which gives Table 2.

TABLE 2 No. of Speech Samples 2 3 4 5 Reduction in Signal- 2.1 8.8 13 15.6 to-Noise Ratio The CCITT recommended receive filter response shows a 30dB reduction in the out-of-band noise, but with certain integrated circuit units readily available, a typical response has a maximum level of -40dB's.

Thus there is room for a further degradation in signal-to- noise ratio before the CCITT recommendation is exceeded. If the filter which is fed by the decoder can guarantee this -4dB response, then a four-way conversation is possible. In addition, a five-way conversation only exceeds the recommendation by 3dB's, which is probably acceptable to the users.

The accompanying drawing is a block schematic of as much of the circuitry at a subscriber's node as is needed to explain the invention.

The transmission medium 1, which is assumed to by a PCM highway serving a number of subscribers' nodes, is connected to a number N of sample stores SS1 to SSN, the number N depending on how many parties are to be permitted in a conference. In most cases, as indicated above, this would be four or five.

The sample stores are controlled from a timing control circuit 2, which is so controlled as to enable one or more of the stores SS1 to SSN, each at the time slot appropriate to one of the other nodes involved in a connection. The outputs of these stores are read out, also under control of the unit 2 to a multiplexer 3 from which they pass to a decoder 4. This is a digital-tolanalog converter, usually of the low type.

The decoder output goes to a low-pass filter whose output goes to the handset. Note that both the decoder 4 and the filter 5 are controlled from the timing unit 2.

This timing unit is controlled from the system control, either direct or via a microprocessor at the node.

Alternatively, in a fully distributed exchange, the timing unit 2 is controlled solely from the node, which responds to incoming signals, notes the slots at which they arrive, and exercises its control accordingly.

It will be noted that, unlike most conference systems, the generation of the output to a subscriber's handset does not take into account whatever is being sent out from that subscriber's handset.

Claims (4)

1. A method of providing a conference facility in a telephone system of the decentralised type, in which at a subscriber's node the speech samples from the subscriber's lines involved in the conference other than from the line to which that node belongs are stored temporarily in sample stores, so that for a conference which involves (N + 1) lines N such stores are used, in which the samples are read out from the stores via a multiplexer to a decoder such that the output of the decoder corresponds to the combination of the samples, and in which the result of said combination, which is in analogue form, is supplied to the node's transducer.

2. A method as claimed in claim 1, in which the output of the decoder, when N samples are present within one frame contains N successive portions each due to one of said samples.

3. An automatic telephone system, in which subscriber nodes are each connected to a time division multiplex (TDM) highway over which pulse code modulation (PCM) samples are conveyed for speech connects, in which each of the nodes has a plurality of sample stores so that within each TDM frame two or more PCM samples can be received and stored temporarily at the same node to provide a conference facility, in which the samples are read out via a multiplexer to a digital-to-analogue decoder the output of which represents the combination of the samples from the other nodes involved in a conference, and in which the said output includes, for N such samples within a frame of the system; N successive portions each of which is due to one of those samples, whereby the conference facility is provided entirely by equipment located at the subscriber's nodes.

4. An automatic telephone system with a conference facility, substantially as described with reference to the accompanying drawing.