CA1160772A - Conference circuit for time division system - Google Patents

Abstract

-1- Y. Parmar-P. Pau 2-2 Conference Circuit for Time Division System Abstract A conference circuit for a telecommunication system using pulse code modulation techniques. The con-ference circuit acts between time slot interchangers to join the conference parties in a conversation. The speech from each party is sampled and that exhibiting the highest amplitude is transmitted to the remaining parties to the conference. The sample comparison is made using only the most significant bits of the sample to minimize speaker shifts due to minor loudness fluctuations.

Description

sackground of the Invention The present conference circuit has been designed for use ~Yith a system of the type shown in Canadian Patent Application 332,385 filed July 23, 1979 by F. Ahmed. In a system of the type shown, parties (representin~ either sta-tions or trunks) may be added to a conference by depressing a conference button followed by dialing of the number of the station to be added.
I~hen a conference is in process, the speaker selec-tion procedure used is one which includes features similax to ones sho~m by U.S. Patent 4,022,9al to D. McLaughlin et al. In that patent, the conversation speaker is chosen by an amplitude comparison.

l 16077'~

-2- Y. Parmar-P. Pau 2-2 The conference circuit of that patent acauires the PCM code of all conferees each frame during their re-spective time slots from the switching networ~. During each conference time slot, a decision is made as to who the speaker will be for the next frame. The remaining con-ferees will then be sent that PCM code during their respec-tive time slots during the next frame. ~he conference speaker is marked in a memory so that when his time slot occurs he is not sent his own PCM code, instead he is sent a code indicating silence.
The procedure followed by the circuit is to re-ceive coded voice sar,ples of all conferees during every time frame, choose a speaker and then send the speaker's coded voice sample to all conferees e~cept himself during the neY~t frame. During each frame, each conferee (except the loudest) hears the last frame conference speaker, and the loudest speaker is being selected for the following frame. Thus a speaker selection is made each frame.
The method of choosing the speaker in the cited ~5cLaughlin patent is to first clear the PCM buffers. Then speech data from the first conferee is stored in a register.
The contents of the register is compared to that of tempor-ary speaker register. If the conferee code corresponds to a larger pulse amplitude modulated (PA~I) sample, (that is it presents a smaller binary value PCM code), the conferee code is loaded into the temporary speaker register. Each new conferee code is loaded to the conferee register and then compared to the temporary speaker register. If it is larger, it is then transferred and becomes the new temporary l 160772

-3- ~'. Parmar- P. Pau 2-2 spea~er. If not, it is written over when the next con-feree code is loaded in. Finally, the time slot occurs during which the sample is received and the te~porary sp~a~er register is transferred to a conference spea~er register. This sample becomes the instant conference speaker for the next frame and this register contains the PCM code which all conferees except the speaker himself will receive. It will then be updated one frame later durina the next occurrence of time slot.

Summary of the Invention The present invention is directed to a PCM
conference circuit which is capable of connecting two or more parties together in a conversation path. ~ith the circuit shown, a maximum of eight four party conferences may be in process at any one time.
Within the conference circuit, speech samples are received as eight bit words in parallel, and these speech samples are stored in a RP~1, having the capacity for storing samples from all eight conferences. Time slot information concerning the parties to the conference which had been assigned by the system processor, are stored in a separate ~ 5. The speech samples of the parties to the conference are stored as received. A comparison is made between a stored sample and each subsequent sample for loudness. The sample indicated to be louder is transferred to a loudest speaker memory along with a registration of the address of the party having entered this loudest sample.
During the same frame and portion of the next frame, this loudest sample is sent to the parties to the conference l 160772

-4- Y. Parmar-P. Pau 2-2 other than the party generating this loudest sample.
It is therefore an object of the invention to provide a time division conference circuit capable of handling a multiplicity of multi-port conferences.
It is further object of the invention to provide a conference circuit which uses loudness comparison of time division voice samples to determine the conference speaker and which outputs its speaker samples in eight bit parallel PCM encoding format to time slot interchangers for the system.
As another object of the invention, a conference circuit is provided to act between time slot interchangers to control the transmission of conference speech data be-tween conferring parties. The transmission of speech is based on the comarison of speech amplitudes from the con-ferring parties to transmit the speech of the loudest sample in digital form to the other parties to the confer-ence. The speech amplitude comparison is effected by com-paring the four most significant bits of an amplitude sample from successive conferring parties with like bits of stored samples from the previously stored loudest sample for that time di~ision frame to determine the frame the speech sample of the loudest conferee to be transmitted during the next frame.
Other features, objects and advantages Jill be-come apparent from the followin~ specification viewed in conjunction with the drawings to be described briefly here-after.

l 160772

-5- Y. ~'armar-P. Pau 2--2 Brief Description of ~he Drawings Figure 1 is a block diagram showing the posi-tion of the conference circuit of the present invention within a representative system and Figure 2 is a schema-tic block diagram of the conference circuit of the pre-sent invention.

Detailed Description of the Drawings In Figure 1, we show a portion of the block diagram shown in Canadian Application SN 332,385 filed July 23, 1979. In that system, there are groups of up to twenty-four stations or trunks having access to the system. Each group has its own group bus over which the parties, stations (or trunks) access the group, the group bus being operative to feed information between the sta-tions (or trunks) and the group control elements such as the group processor and the group Time Slot Interchanger (TSI). Time Slot Interchangers are wellknown in the art as for example beina shown in U.S. Patent 4,16~,627 issued August 14, 1979 to M. Langenhach-Belz et. al. Such cir-cuits act to feed information from a time slot in one time reference system or highway to one or more time slots with-in either the same or another highway. In this way coded speech samples may be transferred, t~iereby associating two or more stations or trunks in a time division conversation.
~hen a normal conversation is set up between parties (the term being used to cover either stations or trunks), two time slots are assigned to each party, one for receiving and one for transmitting. In a two-party conver-1 ~6077~

-6- Y. Parmar-P. Pau 2~2 sation, the receiving time slot of party A is the trans-mitting time slot of party B and vice versa, and hence a total of two time slots is required. ~lo~ever, in a multi-(three or more) party conversation such as a conference, two separate time slots are required for each party and thus the total number of time slots required for a con-ference is equal to t~70 times the number of parties in the conference.

l 160772

-7- Y. Parrnar- P. Pau 2-2 In Figure 1, we show a simplified block diagram of the system disclosed in the previously cited Ahmed application. In the system stations and trunks are group-ed into groups of up to twenty four stations or trunks pergroup. The stations or trunks have access to their respec-tive group over individual four conductor connections, each connection including two data conductors and two speech con-ductors. The stations connections terminate in a station or trunk circuit which has access to the group bus using time division principle.
Each group bus feeds data bidirectionally between a group controller including a time slot interchanger (TSI) and the individual station and/or trunk circuits. In gen-eral a TSI of the type shown herein is disclosed in U.S.Patent ~,16~,627 issued August 14, 1979 to M. Langenbach-Belz et. al.
The group controller through its TSI, feeds speech data over a PCM highway having 193 time slots per frame. Each frame is of 125 microseconds duration and in-cludes 192 usable time slots and one time slot used for synchronization. This PCM highway feeds speech data from a station or trunk in one group to a station or trunk in another group for completion of a call between the stations and/or trunks.
The conference circuit of the present invention accesses this bus to interchange speech data in PCM form in the stations and/or trunks through the group TSI's.

Time slot addresses and other control data is fed to the l 160772

-8- Y.Par~ar-P.Pau 2-2 conference circuit from the system control circuit or controller.
In the system as sho~m by the Ahmed application previously referred to, when a conversation is set u~ be-tween parties, two time slots are assigned to each party,one for receiving and one for transmitting. In a two-party conversation, the receivins time slot of party ~ is the transmitting time slot of party ~ and vice versa, and hence a total of two time slots is required. Ho~ever, in a multi-(three or more) party conversation (a conference), twoseparate time slots are requlred for each party and thus the total number of time slots required for a conference is equal two times the number of parties in a conference.
To set up a conference, the system control has to ~ass to the conference circuit the information concerning trans~ittin~ and receiving time slot assignmen~s of each party to the conference. This information is stored in the RX TIr~ SLOT ASSIGN~ T RA~'l (Figure 2) and TX TI~
SLOT ASSIG~ T RA~ of the conference circuit resectively.
Note that the transmitting time slot assignment information of a party is stored in the RX TIr~ SLOT ASSIGi~M~T RA~ in the conference circuit, not the TX TIME SLOT ASSIG~ E~T RA~1.
The conference circuit picks up the speech sample words (in PCM encoded format) of each party from the PCM HIGH~IAY every frame at the party's transmitting time slots. It makes a comparison on the speech sample ~ords and selects the loudest speaker. The speech sample word of the loudest speaker is then transmitted from the

-9- Y, Parmar-P. Pau 2-2 conference circuit to all other parties at their receiving time slots. A sample word representing a silent period is sent to the loudest speaker.
The conference circuit in one acceptable size, is capable of handling up to eight con~erences of maximum four party each at one time In explanation of Figure 2, ~e note that assign-ments of ~hat time slots are assigned to a conference are stored in the RX TIME SLOT ASSIGN~NT RAM. The RX TI~
SLOT ASSIGNMENT R~M i5 divided into eight separate blocks, one for each conference. Each block contains four storage locations, each location capable of storing the time slot allocation information of one conferee to that conference.
Under the control of the TIMING AND LOGIC cir-cuit, speech sample words in 8-bit PCM encoding format are transmitted from the PCM HIGHWAY and read into the RX RAM
in a location provided by the TIr~E SLOT COUNTER through the multiplexer t~UX 1).
There are 192 locations in the RX RAM, each of these locations representing an available time slot.
Sample words derived from a time slot are stored in the corresponding location of the R~Y R~M and the RAM data, and is updated in every frame.
The RX TIME SLOT ASSIGN~NT RA~ passes the time slot assignment information to the RX RAM through the multiplexer (MUX 1). The sample word of the corresponding time slot is then output from the RX R~M to the TE~IPORARY

SPEAXER REGISTER and or the LOUDER SPEAKER COMPARATOR~

_g_ l 16077~

-10- Y. Parmar-P. Pau 2-2 As soon as a conference is set, the FIRST
conferee is assumed to be the loudest speaker. Its sam-ple word is read out of the R~ RAM and is loaded into the TE~lPORARY SPEAKER REGISTER (LATCH A). This sample word is then fed to the input A of the LOUDER SPEAKER CO~ARATOR.
The sample word of the second conferee is read out of the RX R~ at the second time slot period and is fed to the inp~t B of the LOUDER SPEAKER COMPA~TOR. Only the most significant four bits OL each of the sample words are taken for comparison. If the comparison shows that the first conferee is the louder spea~er, its sa~ple word will remain in the TE~ORARY SPEAKER REGISTER and the sample word of the second conferee will be ianored. However, if the co~-parison shows that the second conferee is a louder speaker, the LOUDER SPEAKER COMPARATOR will generate a pulse to the TIr~lING AND LOGIC circuit which will then generate a signal to strobe in the sample word of the second conferee from the output of the RX RAM to the TEMPORARY SPEAXER REGISTER.
At the third time slot period, the sample word of the third conferee is read out of the RX RAM and is fed to input B of the LOUDER SPEA~ER COMPARATOR to compare with the sample word stored in the TE~PORARY SPEAKER REGISTER. Again, if not the third conferee is/a louder speaker, its sample word will be ignored. At the fourthtime slot period, the sample word of the fourth or the last conferee is compared to the sample word stored in the TEMPORARY SPEAKER REGISTER in the same manner described for the 2nd and 3rd conferee. Before the sample word of the first conferee of the next conference l 160772 -ll- Y. Parmar-P. Pau 2-2 is loaded into the TEMPOR~RY SPE~ER REGISTER, the sample ~ord of the loudest speaker of the present con-ference is loaded into the LOUDEST SPEAKER SAMPLE WORD
STORAGE (LATCH B). At the same time, the information as to which conferee is the loudest spea~er is stored in the LOUDEST SPEAKER I.D. LATCH.
A four time slot period is required to select the loudest speaker of a conference. During the next four-time slot period, while the conference circuit is selecting the loudest speaker for the second conference, the sample word of the loudest speaker of the first con-ference is loaded from the LOUDEST SPEAKER SAr~PLE ~ORD
STORAGE (LATCH B) through the MASK circuit to the OUTPUT
RA~I at some locations dedicated to the conferee of that conference. The LOUDEST SPEAKER I.D. LATCH AND LOGIC cir-cuit provides to the MASK circuit the information as to which conferee is the loudest speaker. The ~SASK circuit upon receiving this information, mas~s the sample word into a sample word comprised of a silent period when it is loaded to the location dedicated to the loudest speaker in the OUTPUT RP~S.
When the OUTPUT RAM is in its input mode, it receives data tsample words) from the r~ASK circuit and addresses from the time slot counter through a multiplexer (I~UX 2). ~hen the OUTPUT RA~S is in its output mode, it receives address information from the TX TII~SE SLOT ASSIGN-MENT ~ S through MUX 2 and its contents (sample words) are read out and transmitted to the OUTPUT BUFFER AND LATCH at a time slot requested by the TX TIr`SE SLOT ASSIGN~ENT RP~1.

l 160772 -12- Y. Parmar-P. Pau 2-2 The sample words are then sent out on the Pcrs HIGHWAY.
The whole procedure, from collectina sample words of all conferees selecting the loudest spea~ers, until sendin~ out the sample ~ords to the PCM HIG~IAY, is repeated every frame. Ilowever, in the next Irame, the LOUDEST SPEAKER I.D. LATCH AND LOGIC circuit will provide information to the RX TI~ SLOT ~SSIGNMENT RAM so that the loudest speaker of a conference at the present frame will become the FIRST conferee of that conference in the next frame. Carry over of the loudest speaker to the next frame serves avoid switchin~ spear~ers when the compari-son of the two sample words exhibit equal loudness levels.

Claims (3)

THE EMBODIMENTS OF THE INVENTION IN WHICH AN EXCLUSIVE
PROPERTY OR PRIVILEGE IS CLAIMED ARE DEFINED AS FOLLOWS:

1. A conference circuit for a telecommunications system in which data is transmitted between stations or trunks over a time division highway in multi-bit digital form during time slots allotted to the stations or trunks transmitting and receiving signals in which there are a plurality of time slots in each cyclically recurring time frame, said conference circuit including a speech memory for receiving and storing digital speech data received from said highway for the stations or trunks transmitting over said highway, a time slot assignment memory for controlling the read out of digital speech data representing stations involved in each of a plurality of conferences, a temporary speaker memory for storing speech data representing the loudest speaker from each of said plurality of conferences, comparator means for sequentially comparing speech samples representing the most significant bits of the speech data from stations of each conference in progress against like bits from the loudest speaker as read out of the temporary speaker memory to determine for each conference the loudest speaker for each frame, loudest speaker memory for storing the speech data of the loudest speaker for each conference in progress for output of each loudest speaker speech data of each conference to the remaining stations of the respective conference, means for masking the loudest speaker speech data from reaching the stations having generated the loudest speaker speech data, and in which the loudest speaker sample is retained for each conference in the temporary speaker memory until a higher louder sample is received.

2. A conference circuit as claimed in Claim 1, in which there is an output memory for transferring the speech data from the loudest speaker of one conference back to said highway while said comparator means is comparing speech samples of stations to a second conference.

3. A conference circuit as claimed in Claim 2, in which there is a second time slot assignment memory for controlling the transfer of speech data back to said highway in allotted time slots.