Audio/video communications processor

Field of the Invention

The invention relates to an audio / video communication processor for the fields of telecommunications and computers and, in particular, to systems which serve a group of users at a workstation as means for acoustic and visual communication.

glossary

- RATING - A method by which the software identifies a group of binary numbers as numbers with a specific location in a fixed binary number scheme, such as the hundreds and thousands in the decimal system.

- COMPRESSION - A method of using software or hardware to generate codes that control the replacement of long chains of binary data with replacement codes that are one or more orders of magnitude smaller. The code can be in modulation format or in binary data format.

- PEL - A term used to describe a digital image data element. A binary number that represents the range of luminosity, chrominance and tint in a digital video system. The smallest independent unit of visual acuity, represented by a binary number on a screen, is a concatenation of picture elements, from which the term PEL is derived.

- PIXEL - Another word for PEL.

- COMMUNICATION PROCESSOR - This term describes a new class of communications equipment that connects users at a workstation to the resources for a video conference over the telecommunications network. The communication processor consists of several sub-processors, which together form the communication processor. Linked form; Communication processor.

- PACKAGE - A collection of binary data on the network side of the communications processor that contains control information for the network and the communications processor, video data and audio data.

- DACS (Digital Access Cross Connect) - digital cross-connect switching device, a machine for exchanging 64,000 bps time slots, either in DS or in another format, between several T1 carriers, regardless of the signaling codes in the DS or T1 - porters. DS refers to the digital phone channel encoding formats DS1, DS2 or DS3. The base rate of 64,000 bps can be a channel that is not a telephone channel and that could contain any type of binary data. T1 refers to the digital multiplex base rate of 1,544 Mbps, which can consist of 24 digital telephone channels of the type DS.

- TASI - Time Assigned Speech Interpolation, refers to speech compression. Speech, or rather, digital speech, is processed according to either the u-law or the A-law evaluation. One is a North American standard, the other a European standard. In other countries, either the North American or the European or the CCITT standard is applied. International piping systems have to recode between the two as they are not compatible. TASI algorithms read the data and hide the speech gaps so that only the actual speech data is transmitted. In addition, TASI can insert codes for certain known speech data patterns, which achieves further compression. The bandwidth saved in this way is used for other purposes.

- Artificial Intelligence (AI) - A software program that, based on a rule group, reacts to events and causes events. The rules are not strict and quick. They are generally soft rules, comparable to commands given to a neophyte. A critical feature of these rules is that they cover all possible events, including events with a low probability. They cover the permutation of events in every possible combination. The aim of the formulation of these rules is to not allow a situation with insufficient or impossible reactions to the external stimuli.

- Plesiochron - A property of two or more clock circuits so that when a synchronization process is performed and canceled, they remain in sync for long periods of time. A day would be a minimally acceptable period of time.

State of the art

As prior art for the invention, some patents are mentioned here as references; afterwards other works are mentioned; these relate to products that contain some elements relevant to a product made using the invention. In on August 14, 1990 to LUMELSKY and on International Business Machines Corp. United States Patent US-A-4,949,169 describes an interface architecture for connecting a number of image display devices over a high-speed, limited bandwidth digital communication link. With the interface architecture at each screen node, sequential data can be transmitted from pixels, which are composed of separate Y and C fields from a digital TV source at each node, and which represent a scaled video window. The audio information is transmitted with the video on sections of the network bandwidth that are not used by the video. It is an object of the present invention to provide a hardware system which enables the use of the existing hardware in the various image display devices and the associated communication adapters, so that a minimum of additional control hardware and software is required.

U.S. Patent 4,780,761, issued October 25, 1988 to DALY et al and assigned to Eastman Kodak Company, relates to a device that takes into account that the human eye is less sensitive to diagonally oriented spatial frequencies than horizontally or vertically oriented , The transceiver can quantify the transformation coefficients according to the model of the human visual apparatus. This system was not designed for video conferencing between workstations over the telecommunications network. It does not work with a video subprocessor as part of an audio / video communication processor and is not suitable for network control, so it cannot be used in the network.

US Patent No. 4,494,144, issued to BROWN on January 15, 1985 and assigned to AT&T, describes video transmission with reduced bandwidth and good video presence. The bandwidth reduction is achieved by dividing the camera video image into segments, by determining the activity level in each segment, and by transmitting the signal of each segment with a resolution that is related to the activity level within the segment. The most active segment is transmitted with the highest resolution, while other segments are transmitted with lower resolutions. This system was not designed for video conferencing between workstations over the telecommunications network. There is no timeshare with a video subprocessor as part of an audio / video communication processor, which means that use in the network is not possible.

The on 29 U.S. Patent No. 4,862,264, issued to WELLS in August 1989 and assigned to British Broadcasting Corporation, describes a method of encoding a video signal for transmission in a restricted bandwidth by dividing a frame of picture information into a group of individual blocks, determining the scope of the Image activity in each block, sampling the information in each block, at a sampling rate related to the image activity in that block, and adding an additional signal indicating the sampling rate used for this block to the encoded block. For each block it is decided whether it is transmitted with full accuracy or whether it is reconstructed from the previous frame. In fact, each block is scanned twice at the same time. The first scan is done with a subrate and the second scan with the Nyquist rate. A block activity generator and a motion activity generator make decisions about whether to transmit with high accuracy or low accuracy. The samples can be transmitted analog or digital.

U.S. Patent No. 4,654,484, issued to REIFFEL on March 31, 1987, and assigned to INTERAND CORP, describes an improved apparatus for quickly compressing, decompressing, and displaying broadband information transmitted over a narrowband communication channel. A video image is assembled cyclically from digitized data, which represent a grayscale intensity for individual pixels grouped in pixel groups, in phases with low resolution and phases with high resolution. In the initial cycle of the low resolution phase, a representative sample of the cell intensity values ​​is transmitted from a sending station to a receiving station according to a video compression routine. A video decompression routine was then used at the receiving station to calculate an intensity value for those pixels whose intensity values ​​are not transmitted, and an initial image was displayed.

U.S. Patent No. 4,682,225, issued to GRAHM on July 21, 1987 and assigned to NASA, describes a method and apparatus for telemetric adaptive bandwidth compression. Adaptive sampling from a video signal creates a sequence of sampled fields. Each field and its area rate information are sequentially transferred to a storage means with a plurality of adaptive fields and stored there. The patented device can be used in spacecraft docking systems where large amounts of image information and data have to be transmitted with a limited finite bandwidth. This invention is suitable for communication systems in space travel, for the transmission of both video and data signals from a spacecraft. In particular, a manual signal can control parameters such as area rate, sample ratio, number of low resolution video frames that are displayed simultaneously, or the division of the downlink communication bandwidth between the data and the video information.

While this patent has little relation to the preferred application of the invention, this sophisticated system, as will be seen, can be significantly improved by the invention. During a video conference, the user can request reinitialization at any time. The communication processor also detects data from poor transmission and automatically reinitializes. The communication processor controls the reinitialization in such a way that the bandwidth allocation between the users is not disturbed.

The US patent US-A-4,739,413, issued to MEYER on April 19, 1988 and transferred to LUMA TELECOM, describes a modulator / demodulator optimized for video, in which the adjacent modulation amplifiers are matched to the gray values ​​of adjacent pixels. Each modulation symbol corresponds to a certain brightness value of the pixel in a ratio of one to one. Issued on March 5, 1974 to GOLDING and on COMSAT CORP. US Patent Specification US-A-3,795,763 describes a digital television transmission system for transmission with a significantly reduced bit rate and bandwidth. The sampling rate is reduced by frequency gearing techniques, the number of bits per sample is reduced by digital differential PCM with edge recoding. A further reduction in the bit rate can be achieved by hiding approximately half of the color data and all synchronization pulses from the transmitted signal. So that the synchronization information can be reconstructed, synchronization words are transmitted periodically. The transmitted bits are multiplexed in accordance with a specific format that ensures correct alignment of the luminance and chrominance lines at the receiver. The Y&C are separated and sampled at a speed below the Nyquistrate. The samples are quantified and converted into difference samples with a further bit reduction. The audio signals are sampled at the horizontal sampling rate and the digital equivalents of the audio and video signals are serially multiplexed into an output stream. Every second pair of C is completely hidden from the multiplexed serial bit stream, but is reconstructed at the receiver from neighboring C information.

U.S. Patent No. 5,043,810, issued to VREESWIJK on August 27, 1991 and assigned to US PHILIPS, describes a method for improved transmission or recording and an improved device for video signal processing and an improved receiver for a television signal. The processing method is a spatial and / or temporal consistency check of a selection that relates to the spatially and / or temporally adjacent parts of the image. The decision-making process includes adjacent parts of the image that can affect the part to be processed. A block of pixels forms part of the image and is scanned in accordance with a scan pattern which does not correspond to this operation, said block adjoining a block which is scanned with a scan pattern which corresponds to this operation, the corresponding scan pattern.

US Patent No. 4,720,745, issued to DEFOREST on January 19, 1988 and assigned to DIGIVISION Inc., describes a method and an apparatus for improving video displays. An NTSC composite video signal is dematriated and its RGB components are digitized into a pixel array with a frame size of 512 x 512. A frame with high resolution is generated from each input frame. The sub-pixel values ​​for a given pixel are derived from checking the nearest neighboring pixels and using improvement algorithms represented by data in lookup tables. The signal-to-noise ratios are regulated by comparison and a decision to change the value of a pixel based on the value of the nearest neighbors or to replace it with its median and that of its neighbors.

US Patent No. 4,858,026, issued to RICHARDS on August 15, 1989 and assigned to US PHILIPS, describes a method for encoding an image to be displayed. The image is encoded using data compression which consists of first obtaining pixel information as a first, high resolution matrix. Like the first, a second matrix with lower resolution is obtained by low-pass filtering. A third matrix is ​​the difference between these two. A fourth matrix is ​​created by sub-sampling the second matrix (not every pixel is used). The third and fourth matrix are encoded. Complementary decoding is to restore the second matrix and combine the restored second matrix by interpolation filtering with the decoded fourth matrix and by combining the restored second matrix with the decoded third matrix. This method is used, for example, for image coding for compact discs; however, the communications processor does not operate according to the principles described in this patent.

US Patent No. 4,733,299, issued to GLENN on March 22, 1988 and assigned to NYIT, describes a method for converting interlocked sampled video signals into successively sampled video signals. The applicant has learned that motion adaptive processing is not required. The low-resolution information is obtained from the current interlocked field and the remaining detailed information is obtained from a stored signal that contains one or more previous fields. Only the detail signal is obtained from previous fields, and since the human eye cannot perceive movements with high spatial frequencies so quickly, there are few, if any, movement artifacts perceived.

Issued on November 5, 1985 to MATSUDA and on PIONEER ELECTRIC CO. U.S. Patent No. 4,551,755, assigned, describes a bandwidth correction system for a TV tuner. A bandwidth control voltage is applied to a bandwidth adjustment circuit that provides a pass bandwidth that is determined by the relative levels of a video intermediate frequency signal and an audio intermediate frequency signal. This patent appears to imply that the transmission process over-modulates or exceeds the bandwidth allocated for a given moment from time to time, and that a correction signal causes the receiver to recognize and adjust this condition.

The communications processor has a different application, method, and method than that described in US-A-4,551,755. However, the concept of bandwidth correction and bandwidth allocation needs to be differentiated. Bandwidth allocation from video to voice signals or vice versa need not necessarily be done; however, if this happens, it is not for the purpose of correcting a malfunction, such as overmodulation, but rather to manage both types of bandwidth.

Issued to MA on December 20, 1988 and to CAPETRONIC (BSR) Ltd. US Pat. No. 4,792,993, which has been assigned, describes an improved TVRO method (TVRO = TeleVision Receive Only, normally relates to satellite TV). The improvement is an automatic filtering of the audio signals to a frequency range of the band that lies outside the modulated video signal, combining the filtered audio signals and the video signal, and transmitting these signals through the channel with the restricted bandwidth. At the remote station, the signals are broken down into audio and video signals by a reversal process.

The on the 18th U.S. Patent US-A-4,849,811, issued to KLEINERMAN and assigned to KLEINERMAN on July 1989, describes a method for simultaneously sending audio and video signals over standard telephone lines or another channel with limited bandwidth, including fetching a video image, digitizing the image, modulating a signal with the digitized image, fetching audio signals and filtering the audio signals up to a frequency range of the band, which is outside the range of the modulated video signal, combining the filtered audio signals and the video signal, and transmitting such signals over the channel with the restricted bandwidth. At the opposite station, the signals are broken down into audio and video signals by a reversal process. The communication processor has several elements in common with this patent, for example the digitization of the video signals and their digital processing; however, there is an improvement by the invention by combining the audio and video signals in a continuous channel frame that is separated only by the software protocol. Patent 4,849,811 separates the audio and video signals in the restricted bandwidth frame by separately modulating the video and audio signals. The communications processor does not use a telephone channel. Instead, he uses a digital line, the smallest of which is 64,000 serial bits per second. The long-distance carriers can provide this speed as line capacity or dedicated lines, with which at least 64,000 serial bits per second are possible as a video conference service. The best performance is achieved with carriers of networks with high bandwidth.

US Patent No. 4,425,642, issued to MOSES on January 10, 1984 and assigned to APP SPEC TECH Inc., describes a second-channel communication system in which a digital data signal is transmitted simultaneously in the signal of a communication medium, for example telephone speech or television picture can. The data signals are converted into very weak multifrequency signals, which consist of fundamental frequencies and harmonic vibrations across the entire communication bandwidth. The data signal is spread-band modulated and its energy content is distributed over the entire band, which results in a slight deterioration in the signal-to-noise ratio, since the audio or video signal is obviously subject to pseudo-noise. Since the data signals generate the pseudo noise in a coherent manner, it is also recognized in a coherent manner and masked out of the audio or video signal at the receiver.

The communications processor does not perform spread band modulation; just as little does it transmit data with voice or video signals in the second channel. Instead, the communications processor uses a protocol to keep audio, video and data in a serial transmission channel:

U.S. Patent No. 3,873,771, issued to KLEINERMAN on March 25, 1975 and assigned to TELSCAN, describes a system for simultaneous transmission of a video and audio signal over the same transmission line using FM narrowband TV while the audio signal is AM -A sideband technology is transmitted. Both the video and the audio signal occupy the channel at the same time in separate frequency ranges. The communications processor does not have to use analog modulation or frequency division multiplexing or narrow band FM TV techniques.

US Patent No. 4,797,750, issued to KARWEIT on January 10, 1989 and transferred to J. HOPKINS University, describes a method and apparatus for transmitting a recorded display generated by a computer simultaneously with the transmission of audio. and / or video signals. A computer generates a series of codes from which an image can be derived, the resolution of which does not depend on the recording transmission medium. These codes are transmitted to a first modem via an RS-232 communication line. The first modem converts these codes into image-bearing audio tones. The audio tones are input into the left audio channel of a VCR. At the same time, the acoustic information is recorded by a microphone and entered into the right audio channel, while video signals are transmitted from a video camera to the video channel of the recorder. During playback, the audio signal in the left channel is decoded by the modem and converted back into the display generated by the computer. The communication processor does not perform spread band modulation, nor is an RS-232 communication line used. Instead, another way of using a modem is described.

US Patent No. 4,736,407, issued April 5, 1988 to DUMAS and assigned to US ARMY, describes an audiographic conference system between two or more users who are either directly connected to each other or by a bridging device via voice lines. Each user has a personal computer, software and an intelligent modem, a cassette player / recorder and a head microphone. The connection between the users is shown in Fig. 1 shown. The intelligent modems wait for a band signal, decode it and pass it on to the computer; The head microphone allows the user to hear while being controlled by the software. The cassette recorder / cassette player runs automatically.

The communication processor differs from this patent in that in the video conference circuit implemented in this patent, computer data and voice are rather passed on to the conference participants.

US Patent No. 4,955,048, issued to IWAMURA on September 4, 1990 and assigned to SHARP Kabushiki Kaisha, describes a method for multiplexing the transmission of audio and video signals, the video signal being divided into a luminance signal (Y) and a chrominance signal (C). is divided. The Y signal is then modulated and the C signal is modulated with a low frequency carrier. The resulting C-modulated signal is converted to a lower frequency. The audio signal, the frequency-modulated Y signal and the frequency-converted C signal are multiplied by a frequency division to be transmitted via a telephone cable.

US Patent No. 4,999,831, issued to GRACE on March 12, 1991 and assigned to UNITED TELCOMM Inc., describes a method for digitally transmitting broadband video, narrowband audio and digital signals over information networks. It describes synchronous quantized subcarrier multiplexing which results in electronic multiplexing of voice, data and multi-channel NTSC full bandwidth video for digital transmission over the communication line and error processing. The channels to be multiplexed must be carefully selected with regard to the frequency content so that they do not overlap one another; The signals are then subjected to low-pass filtering and modulated with local reference signals (Double Side Band Suppressed Carrier, DSBSC) and consequently form baseband, medium-band and high-band channels which are combined with one another and input into a D / A converter. This results in a serial bit stream known as quantized SCM. This patent is different from the present invention.

Issued on June 25, 1991 to BAJI and on HITACHI Ltd. U.S. Patent No. 5,027,400, assigned, describes a bi-directional multimedia broadcast system. The main control unit receives information from the subscriber stations over a network. The software in the main unit decodes the request from the subscriber station and provides the service by controlling all transmission processes. This service can be a movie or a commercial database. The transmission also includes bandwidth compression for a video signal. This system is described as a broadband ISDN broadcasting system and as a system that also allows a limited number of cable channels to be used with CATV.

The communication processor differs from this patent in that the patent is designed for interactive advertising in ISDN broadband networks, where customers can see video images of the product and have an interactive dialogue with the main station to carry out the transactions. This patent is also applicable to CATV systems where customers can selectively order video programming services rather than having technicians hard-wire the programming services they want. The communication processor is to be used for video conferencing between workstations.

Issued to FISHMAN on September 10, 1985 and owned by JONES FUTURA FOUNDATION Ltd. U.S. Patent No. 4,541,008, which is assigned, describes a system for transmitting TV signals which includes circuitry for processing and encoding a reduced-repeat signal. The system separates the video components, generates sampled digital values ​​of the color and intensity components and stores them in a buffer memory. A data processor compares successive samples of the various video data from which it generates blocks of video data of variable length in order to represent slowly changing signals or rapidly changing signals. One switching arrangement encodes and multiplexes audio signals and synchronizes data into the signal stream, another switching arrangement encodes signal and control data for transmission to a receiver and a circuit on the receiver for reversing this process. The process of reducing repetition involves a variable speed scanning operation using codes that indicate when color, intensity and luminance information is repeating. In the case of repetitions, only every eighth scan is transmitted. The timing information to control the sampling rate is critical. The orderly continuous sequence of line scanning now depends on circuits which provide the identification rate of the sampling rate with interpolation data. On the other hand, the communication processor reduces repetitive video data.

U.S. Patent No. 4,394,774, issued to WIDERGREN on July 19, 1983 and assigned to COMPRESSION LABS Inc., describes a digital video compression and decompression system and methods for compressing and decompressing digital video signals in real time, at speeds that are NTSC color - Broadcast speeds can reach. The system compression device receives digital frames and divides them into subframes; then it performs a single-layer spatial domain to transform the domain transformation in two picture element dimensions. The resulting coefficients are normalized and compressed using a predetermined ratio. An adaptive rate buffer is provided to control the feedback for compression. The compression device adaptively determines the feedback component for the capacity control of the rate buffer in relation to the current data content of the rate buffer in relation to its capacity; it also controls the absolute amount of data from the normalization step so that the buffer is never completely empty or full. In practice, the color image is divided into luminance and I and Q chrominance components. The luminance component is compressed and decompressed using the rate buffer feedback technique with scene-adaptive coding. The I and Q components receive simple spatial low-pass filtering, followed by spatial sub-sampling with dimensional interpolation at the system receiver. The audio signal is filtered and sampled at a fixed rate; it is mixed with bit-screen synchronization codes and transmitted as a serial bit stream.

FURTHER BACKGROUND INFORMATION REGARDING OTHER PRODUCTS

In the following detailed description, the reader is referred to the following works:

1) Article in the April 1, 1991 issue of PC Week Journal, page 43, entitled "Analysts Expect Video Meetings to Boom in 90's" by Michael Zimmerman. Industry experts predict the increase in video conferencing as a form of communication.

2) Stanford Computer Optics Inc., 3530 Sugarberry Lane, PO Box 31266, Walnut Creek Ca 94598, produce a product called "4 Quick" which is an image capture device that uses 30 to 60 frames per second variable delay is detected to produce a residual light image of 512 by 512. It works between the wavelengths 130-920 nm and can create a frame in just 5 ns. This device could be used to produce picture frames for multiple users in a timeshare arrangement.

3) Welch Allen Inspection Systems Division, 4619 Jordan Road, Skaneateles Falls, New York 13153, produces a product called "VP3 Videoendoscope". A small video probe, a miniature video camera that is only slightly larger than a fountain pen, takes the picture and sends it via cable to a device in which the picture is compressed and modulated and then fed into a telephone line at speeds between 14,400 and 2,400 bps. This device may be suitable for integration into the front ring of a screen.

4) Dialogic, 300 Littleton Road, Parsippany NJ, produces an information program "Call Processor" and "Audiotex" for voice information systems and telephone services with computer-assisted switching. The most important feature is the interactive voice output for central dictation services. This product is representative of the type of speech compression that can be achieved today.

5) Telephone Communications, 11722-D Sorrento Valley Road, San Diego CA 92121, produces an image compression product called "ALICE" that performs 15-to-1 compression without loss of resolution. It is software that has been designed for integration into other products such as teleconferencing, picture data base and surveillance. It is said to be able to send full color images in less than 10 seconds over normal telephone lines or to save 4,000 color images with high resolution on a 50 Mbyte hard drive. This product is representative of the types of compression levels that can be achieved.

6) Dataproducts New England, Barnes Park North, Wallinford CT 06492, produces a product called "DPMUX M-44". This device can accept channel inputs from a telephone switchboard using E&M signaling, or data in baseband format from a digital device to a port. The speech signals are then converted into digital. Speech is converted at the desired bit rate or, in the case of data, they are timed to match the network timing and assigned a time slot on the network total transmission side. The sum side is either a fractional or a complete T1 service. With this device, the operator can make the following configurations: number of channels, bit rate allocation, clock allocation (asynchronous, synchronous, plesiochronous), as well as the class of the voice service. The concept of the parameters that can be configured by the user and downloaded to the remote station via a sum frame can be extended to the automatically configurable generation by means of a control algorithm that is downloaded to the remote station by a sum frame.

7) The network transmission devices manufactured by the various manufacturers have functions which are applicable to the communication processor. These functions include time slot machines and time assigned speech interpolation machines (TASI, voice data as the current bandwidth variable) such as the product IDNX from Network Equipment Technologies, 800 Saginaw Drive, Redwood City, CA 94063. This class of device fits both data and voice, in an adaptable format to use the available bandwidth efficiently. The devices of this class are described as transmission resource managers. The concept of managing transmission resources is applicable to the communications processor in workstation video conferencing.

8) Feature article in the February 1991 issue of Telecommunications entitled "Streamlining Protocols" by William Stalling. The article describes changes to transmission protocols to improve transmission performance. The communication processor uses an adaptive protocol that enables it to communicate with many different electronic digital transmission devices to simplify transmission control during video conferencing between workstations.

9) Feature article in the February 1991 issue of Telecommunications entitled "LAN Interconnections Technology" by Michael Grimshaw. The article discusses the differences between bridges, routers, repeater stations and switching stations that are used in local area networks and the interfaces to the transmission network. The communication processor takes over the network functions bridging, routing, amplification and switching.

10) IX General Assembly of the Internal Telegraph and Telephone Consultative Committee (CCITT) - Document 52, Report R43 of Research Group VII, Recommendation X.200, entitled REFERENCE MODEL OF OPEN SYSTEMS INTERCONNECTION FOR CCITT APPLICATIONS. The model is also referred to as the 7-layer ISO model. The application process referred to in this document is a video conferencing circuit performed by the communications processor.

11) IEEE International Conference on Communication ICC 90, Volume 1, 3/3/90, page 134-139, XP 147391, CH Weis: "Desktop Video Conferencing - An important feature of visual communication in the future" - - deals with desktop Video conferencing and explains how to implement it on a broadband network.

SUMMARY OF THE INVENTION

The invention can be used for telecommunications networks, with different channels and different tariffs, to which the communication processor is connected for and on behalf of the workstation users to provide the desired video conferencing services.

A workstation must exchange audio and video information with other workstations. A picture of the operator or a video with sound would significantly increase personal productivity by significantly improving communication between individuals. The objectives of this description are as follows:

1) Propose a simple video interface from the workstation to the communication processor at low cost. There is more than one solution to this design element, but only one of these solutions needs to be shown here.

2) Propose a simple video evaluation algorithm that gives high quality perception without the need for high bandwidth. There is more than one scoring algorithm that would meet a scoring requirement. Because the rating plan is proportional to the transmission bandwidth, and inversely proportional to the performance of the display, the communications processor selects the appropriate rating plan that performs best under the existing conditions. The use of an algorithm to select an evaluation algorithm is important.

3) Propose an audio management algorithm that would involve the use of multiple rating standards that would allow the communications processor to be compatible for multiple standards, allowing connection to different network circuits. The plan takes into account the type of service requested and the service requested by the other party. The decision is made based on a match between the two. This selection then becomes a criterion for part 2.

4) Propose a network management algorithm for dynamic multiplex transmission on hierarchical T-carriers down to the DSO level, standards for fiber optic media such as SONET and FDDI. The DSO level is the lowest common denominator for an audio / video channel because the development of communication requires compatibility with existing standards and DSO is an interface of the predominant type. A key function of the communication processor is to assemble audio and video signals on at least a small part of a T1 carrier. If the bandwidths allow higher carrier rates, the communication processor uses higher carrier rates.

5) Artificial Intelligence (AI) for the system controls the final processing after the transmission network has been used as an input base for processing decisions. Factors such as connectivity and activity are two of the criteria used to assess connectivity. User requests for services, the capabilities of the workstation and the user availability at the remote station are added to these type factors.

These tasks are generally fulfilled by the solution mentioned in the subsidiary claims.

Further advantageous exemplary embodiments of the present invention are set out in the subclaims.

To accomplish the above objectives, there is provided a system for coupling workstation units in a system that transmits audio and video signals on a carrier and that includes a communications processor that has a digital bus for connecting the elements coupled to the communications processor , The network interface ports of the communication processor include a port for a network carrier signal, a processor connects the various units in the network to one another. A workstation interface, a video processor and an audio processor are connected to each other and exchange digital and analog signals with each other; They also forward digital information via the system digital bus. To control the system, a channel frame processor is provided, which is connected to said digital bus and controls communication via said digital bus. A statistical audio / video multiplexing processor is also provided on the system bus and is connected to the digital bus to dynamically change an allocation bandwidth between audio and video information signals on the digital bus based on changes in an amount of activity of the audio or video signals during transmission of the audio and video information signal.

The audio processor has a speech compression / evaluation subprocessor that is shared by several users coupled to the communications processor system.

In addition, it has a way of performing the compression and evaluation based on an assignable bandwidth provided by a user of the system and based on a decision of the communication processor to assign control of the process with the final bandwidth for the requested bandwidth.

The video processor has a video compression / rating subprocessor shared by several users coupled to the communications processor system. This compresses the video information and is controlled by the communication processor based on the video activity and the assignable bandwidth and rating provided by a system user and based on a decision of the communication processor to assign control of the process with the final bandwidth.

The system is equipped with a statistical audio / video multiplexing processor which is shared by several system users in order to model the communication channel and the usage situation and to calculate the parameters for the channel transmission in the system. The channel frame processor can be coupled so that multiple users share it to assemble a channel frame for insertion into a sum frame.

Furthermore, a common superframe processor for implementing a language of the telecommunication interface is provided for all users, one per port.

The audio / video communications processor system, in common to all users, can process Artificial Intelligence (AI) software that is subordinate to all other system processor elements to synthesize the implementation parameters of digital speech modulation conversion from one channel to a sum frame, to the channel in one reproduce other aggregate frames and to set conditions for facsimile use, so that channel threads can be formed for the connection between system elements in order to connect the individual workstations working with the system mentioned.

These and other features are described in connection with the following drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

Fig. 1 shows the various classes of transmission equipment in the form of a Venn diagram. The circle overlaps represent the transmission equipment, the ellipse represents the communication processor. Since the communication processor has to work with the different network equipment used by the network operator for the provision of tariff services, it must be able to select the corresponding algorithm with all its advantages and constraints. The communications processor must then take advantage and operate within the constraints to assemble the channel packet according to the service required by the user.

The communications processor not only performs the same functions as the transmission equipment, but also performs audio and video processing functions in addition to these functions, making it a new equipment class. In the telecommunications lexicon, DCE is described as a data communication device that belongs to the telecommunications company. The DTE is the data terminal equipment; this belongs to the user of the services provided by the telecommunications company. The distinction between DCE and DTE is applied to the equipment as such. Since it is a new equipment class, its use in the market will ultimately decide whether it is a DTE or a DCE, since elements of both are present. FIG. 1 is referred to as AUDIO / VIDEO COMMUNICATION PROCESSOR, NEW DEVICE CLASS.

2 shows the interface of the communication processor at the level of the workstation. It consists of a front ring that contains the electronics for the forward video and for processing the transmission and reception voice signals, as well as for receiving the feedback and the video of the remote station. It communicates with the communication processor via control data and receives data for the user interface, such as. B. Menu selection data and work session data. The software and interface adapter card with which the workstation is connected to the communication processor is implied, but not shown. The front ring has no direct connection to the communication processor.

FIG. 3 shows how a video evaluation plan can be constructed for presentation in accordance with the display area in FIG. 2.

Fig. 4 shows a channel frame consisting of three types of data. The figure is titled FRAME BEFORE SUM CODING.

5 shows the data output structure at the total port of the communication processor. This figure is titled SCHEMATIC REPRESENTATION OF THE FRAME.

6 shows the high level functions and interfaces of the communication processor. It can be called FUNCTION PLAN OF THE COMMUNICATION PROCESSOR. As you can see, this is the preferred embodiment of the functions. With a view of the workstation, the communications processor would be a DTE. On the network side, network connections are called sum ports. A total frame is transmitted from each port. Each total frame contains the channel frames from the workstations. A channel frame can come in on a port that has no destination at this point, but is forwarded to another port and destination. The information required for this must be returned to the original communication processor in order to be set up and coordinated there correctly.

7A shows the basic functional concept of the communication processor, which is connected to the group of workstations and the telecommunications network. It is titled NETWORK CONCEPT.

Figure 8 shows an image processing system that can be applied to the communications processor

Figure 11 is an extension of Figure 3 with vertical and horizontal line numbers inserted here. The numbers shown were taken from IBM's VGA graphics specification. However, the purpose of the figure is to emphasize that the bandwidth is linked to a video specification that is not driven by it, since each workstation must use a window and a window is part of the overall screen. In addition, two dissimilar graphics specifications can hold a video conference together.

12 shows an idealistic bandpass frequency response of a focal plane image detector. Part of the image processing is the integration of the frequency from the low frequency limit to the high frequency limit in a single value that represents the pel. The picture is titled DETECTOR FREQUENCY INTEGRATION SCALE.

Figure 13 shows the types of signals that would be exchanged between the communications processor and the workstation using the technique shown in Figure 10. The FIGURE is titled AUDIO INTERFACE OF THE COMMUNICATION PROCESSOR. This figure is intended to show that a complex multiplexing process or complex fiber optic arrangements are not required to perform this function. A simple frequency division multiplex procedure on an RG cable with limited bandwidth is sufficient.

14 shows a possible method for processing speech signals. Here, the speech is first treated as a 64 kbps PCM signal, with either u-law or A-law weighting, before any compression or weighting is performed on the signal if the carrier is operating at T1 rates or higher rates and the activity is low. If the bandwidth does not have to be reduced, no further processing for the audio bandwidth is required for the speech signal, although the audio signal can nevertheless be subjected to TASI processing (TASI = Time Assigned Speech Interpolation). TASI works with the video processing section to split the allocated bandwidth between audio and video. The figure is titled AUDIO FORMULA FOR A PACKAGE.

Figure 15 shows how speech can be replaced by a fax signal or text in the channel packet. Because fax is an audio tape representation of an image. A complex image can be transmitted from workstation to workstation using the fax invention and the corresponding advantages and features, the following drawings being given as examples.

Figure 16 shows how communication processors are interconnected.

Fig. 17 shows changes in the error bits with respect to the block size.

DETAILED DESCRIPTION OF THE INVENTION

Before we go into detail about the preferred embodiments, an example should perhaps explain how the communications processor would work. A user (# 1) at a workstation decides to request a video conference session with another user (# 2). The user is in the same establishment, but in a different building. Number 1 calls up a menu on the screen. The menu can be controlled interactively with the mouse (mouse: because of the argument), so that the user clicks on the user directory "Who is Who". In the user directory, he selects the name of the person with whom he wants a video conference. The name is linked to a routing directory via a list. The user also selects the window size for the receive video. The names in the directory window are linked to and mapped in the routing directory. Since the communications processor assigns a route to user # 2, it assigns a network port and a telecommunications network interface (ISO levels) for that port. The communications processor then checks the port for available bandwidth, puts the requests in a "jobs to do" list, and then returns to the user at the workstation with suggested window sizes.

A high fidelity symbol is associated with the window size. This symbol represents the voice quality, which must be coordinated with the window size. When the user changes the window size from small to large, the high fidelity icon changes from large to small. The video window consists of two parts: one part is the size, the second part is the speed. The speed at which the video is updated is symbolized by rotating wheels (film reels of a camera). The communications processor controls the speed of the wheels by notifying the workstation of the window area and speed that are permitted at that time. The workstation processes the symbols via the application software and proportions them from a low value to a maximum. The faster the wheels turn, the more natural the video picture is, the slower the wheels turn, the less natural the video picture is. Experience leads the user to an acceptable value.

The communications processor then initiates the session. This process consists of opening a channel to the other communication processor and forwarding a service request. The first communications processor has determined that there is bandwidth. The second communications processor must determine if a second user, # 2, is available.

The improved image processing of the described hardware system, like US-A-4,862,264, allows video signals to be processed in blocks to identify the activity of the block, and then to make a decision whether with full accuracy or with reduced accuracy to be transferred. With the improvement one has the possibility to always transmit at reduced rates to take into account the next potential user. The reduced rates do not necessarily affect the image quality if the transmission selection process does not switch off the information but the redundancy. In contrast to US Pat. No. 4,862,264, the decision-making process in this system can be based on a pixel-numerical dynamic range. The image is always transmitted in the form of binary numbers. The number of binary symbols per picture element is determined by the A / D process; however, the number of bits transmitted and their position in the number field is first determined by the communications processor based on the available bandwidth. The communications processor can choose not to transmit video signals, but to use the bandwidth for audio or data. The user does not necessarily notice a change in image quality because the intent here is to reduce the bandwidth requirement at the source by means of a controlled field of view and feedback-controlled lighting.

It should be noted that the communications processor can be used to improve results in areas such as those described in U.S. Patent No. 3,795,763. The memory is used to store successive video frames and to carry out difference algorithms between them. He then carries out the further processing, unlike in the description of US Pat. No. 3,795,763. For example, one difference is that the pulse synchronization information is not transmitted because the workstation video hardware at the other end may have different video properties than the transmitting workstation. However, it should be appreciated that the communications processor processes workstation video signals, not normal television signals.

After the second communications processor determines that user # 2 is available, the processor continues.

Each workstation can set up the reception status. If a workstation is busy and is currently working in an application, and if the user has set the reception status to "No", the communications processor denies access to the first communications processor with the message "busy". User # 1 will also be informed. If the reception status of user No. 2 was set to "Yes", a query appears on the screen. If the user answers this query positively before a certain time has elapsed, a work session is initiated. The parameters for workstation # 2 correspond to those of workstation # 1, so both have the same fidelity and window size. The sessions can also be customized so that user # 2, who has a more powerful workstation, has a larger video window or a better quality video window. The adjusted parameters are set again by the user via the menus. The response to a receive status also includes information about the window size, which may be defined in an adjustment list.

The case described is a simple example because the second user was in the same establishment, albeit in a different building. In such a situation, one would expect that the telecommunications network would not be used at all, but rather a specially designed fiber or wire line with FDDI bandwidths. If the second user is not in the same company as the first, the telecommunications network is the means for handling the data traffic. In this second situation, there is a larger number of interactions with a larger number of different devices. The communication processors most likely communicate through intermediaries. The communications processor must be at the physical level (see ISO, CCITT X.200; x. etc.) speak the right language up to the network layer, for example 2B1Q or AMI.

The communication processor can act as an intermediary between two other communication processors that want to communicate with one another. This is a necessary function because there is as yet no transmission equipment that can interact with the communication processor to enable video conferencing. In a situation in which a communication processor works as a power supply unit, it can be requested for routing or bridging between two separate transmission circuits. For example, the communications processor may translate from 2B1Q to AMI. The rules for this translation are embedded in the AI ​​software. If there are no rules, the communications processor will not attempt routing or bridging. This area is extremely complex and will not be explained further in this description.

When implementing new technologies, costs play an important role. The cost of establishing a video / audio communication facility on workstations can be significantly reduced by using a common communication processor that is shared by multiple users. The costs are reduced by using the computing power available in the workstations for the display and control. The communication processor should not compete with the Token Ring or Ethernet Local Area Networks (LAN), but should only perform another type of service by providing inexpensive audio and video communications within a company and also for long-distance transmission. The purpose of the communication processor is to provide audio signals and communications with the best possible quality, taking into account the time-dependent constraints of the transmission medium and the current level of utilization or utilization. The communication processor does not rely on specific standards for audio or video, but the bandwidth, resolution and transmission speed are adjustable so that they adapt to the respective constraints at the time a particular service is requested.

A workstation initiates a service request. This service request is entered by the user through a software controlled on-screen menu. The menu parameters are determined by the communication processor. The selected parameters together become the service request. A service request includes data about the type or type of service and the signal destination. This information is enough for the communications processor to attempt multiple routing threads before an affirmative determination can be made. If an affirmative determination is not possible, the communication processor determines what is possible and suggests these possibilities to the user.

The main concept of this design is the integration of several disciplines according to the principle of systems engineering. Due to the sensible compliance with several communication standards and the application of extended possibilities in high-speed signal processing, in optics and in light wave processing, a cost-effective communication processor is feasible.

The preferred embodiment

We now want to deal with the invention in more detail; Figure 1 shows our preferred embodiment in which the communications processor is specified for processing workstation audio and video data using artificial intelligence techniques for the use of the telecommunications network. This use of the telecommunications network makes it possible for the first time to use a simple and effective means for personal video conferencing. The communication processor is a collection of existing technologies, the collaboration of which has been improved and modified towards a common goal. The most important functions are network functions and are shown in FIG. 1. The development of the network was aimed at increasing the bandwidth. This development was valid for a while, but there was also a development aimed at developing digital bandwidth services that connect the customer to the network at different layers. In the structure typically referred to as a seven-layer ISO or International Standards Organization structure, the lower layers were specified for the basic control of network equipment for access to the physical functions, the data link layer functions, the network functions and the transport network functions. The ISO specification for access to the physical functions, the data link layer and the network are embedded in the communication processor, in particular the AI, for communication with those elements in the network that provide these functions. Starting with the transport layer and continuing upwards, the communication processor only communicates with other communication processors to provide this function. Although it may later be possible to structure the transport layer portion of the communication processor to work with higher layer telecommunications network equipment, the functions of the lower physical layer, data link layer and network layer are currently in the AI ​​section of the communication processor embedded. The bottom three layers, using the definitions of CCITT, are built into the communications processor for communicating with the equipment of the telecommunications network. In addition, the functions assigned to these layers are used by the AI ​​software.

The communication processor has the following basic elements:

1) Speech compression / evaluation subprocessor used by multiple users. The algorithm chosen is based on the assignable bandwidth. This multi-user sharing is important because it allows key bandwidth process control with the communications processor.

2) Video compression / evaluation subprocessor shared by multiple users. The compression algorithm and the evaluation process are controlled by the communication processor based on the speech activity and the assignable bandwidth. The subprocessor works with component 1, the audio subprocessor.

3) Video imaging device (can be embedded in the front ring). Works with the communications processor through the group of interface boards for the workstation and related software.

4) Speech coding processor (can be embedded in the front ring). Provides digital speech capability with either 64,000 bits per second PCM or 32 or 16 Kbps ADPCM or LPC, bidirectional or full duplex.

5) Statistical audio / video multiplexing processor used by multiple users. This function is driven by the AI ​​software in the communication processor. It essentially models the communication channel and the usage situation. It calculates parameters for the channel and sum processor. This is the processor that uses the collected data, such as error-free seconds, to change the error correction code to minimize the overhead.

6) Channel frame processor, which can either be a common processor or a processor that many users share. Part of the communication processor is responsible for assembling the channel frame for insertion in the sum frame. It implements the results of the statistical audio / video multiplexing processor. This function is driven by the AI ​​software in the communication processor.

7) Compact superframe processor for all users, one per port. This function is driven by the AI ​​software in the communication processor. This processor implements the language of the telecommunications interface such as FDDI, 2B1Q or AMI or similar types.

8) Communication processor with AI software for all users, to which all other processors are subordinate. This processor is responsible for synthesizing the implementation parameters of the digital speech modulation conversion from one channel to one sum frame in order to reproduce the channel in another sum frame. He is also responsible for setting conditions for facsimile use.

While the products described under 2) to 5) are independent of one another (not connected to each other), an attempt will be made to establish a specification between them. However, this should not be done in relation to a specific product, but rather in terms of a specific function. These products stand for the designs possible with today's technology. Based on the available options, the communication processor can be developed from these or similar products.

The processors mentioned under 6) and 7) are linked in a unique manner, with the exception that the aim of TASI equipment is to increase the apparent transmission bandwidth, in that more voice channels can be assigned to a digital carrier than with fixed multiplex systems. For example, the T-1 carrier carries 24 fixed channels of 64,000 bps; with TASI equipment, however, it could carry 32 or more channels. The quality of the services is reduced by TASI and the data transmissions are severely impaired. The space or inactivity periods associated with the speech serve for the transmission of voice channels with extra capacity. The sum frame structure contains the information about the remote station in order to reconstruct the voice channels in the correct order. The difference is that a fixed number of 8 bits used to encode the speech signal is dynamically reduced from 8 to 4, depending on the dynamic range and activity. When speech is replaced by data, the TASI algorithm causes enough signal distortion noise to destroy the data. So it is imperative that users know how their data connections are handled by the telecommunications carriers.

Connections purchased under a certain tariff may not necessarily be used for other purposes. The communications processor fits into this category. Since the AI ​​coding is different for each tariff zone, no replacement is possible. A communications processor port designed for the ISDN (2B1Q) B channel cannot be exchanged for a 128,000 bps channel from a programmable T-1 multiplexer, although the nominal tariff rates are the same.

First distinctive element: network architecture. Several workstations are connected to the communication processor to set up a video conference. The video conference may be local or remote to the origin and destination. Distant targets need to be similar to the originating unit in relation to the communication processor and the workstation, but not identical. Workstations on the same communications processor can also hold a video conference. The workstation works with software that can be used to perform several tasks in connection with the communication processor. The five tasks are as follows:

1) Prompt the user to enter the desired service using the menus. Two different types of information are required for this:

a) General class of the desired video service, for example the window size, and the refresh rates from low to high.

b) General class of the desired audio service, for example the clarity of the sound reproduction from low to high and the delay time from zero to one second.

A high fidelity symbol is assigned to the window size. This symbol is an icon of a gramophone. The high fidelity symbol stands for the voice quality, which must be coordinated with the desired window size. If the user changes the window size from small to large, the high fidelity symbol changes from large to small. The video window consists of two parts: one part is the size, the other part is the speed. The speed at which the video is updated is represented by rotating wheels (film reels of a camera). The video symbol is a film camera with coils on top and a lens at one end of the camera. The side of the camera is a box of variable size in which the receiving video will be located. The communication processor controls the speed of the wheels by notifying the workstation of the window and speed range that is possible at that time. The workstation processes the symbols via the application software and proportions them from a low value to a maximum. The faster the wheels turn, the more natural the video is, the slower they turn, the less natural the video is. Experience leads the user to an acceptable value. If enough bandwidth is available, there may be little or no interaction between the high fidelity symbol, the coils, and the window size. If the bandwidth is small, a small movement can quickly lead to a very large interaction. The communication processor controls the size of the symbols at all times and the user cannot request or cause impossible service. An impossible service would be just video and not speech or vice versa, or a situation that is not programmed into the AI ​​software.

The audio service will also be briefly discussed. As part of the audio quality tuning, fidelity is reconciled with video quality. This includes the synchronization of the language with the video. The communication processor will always try to synchronize the language with the picture. However, the speed of the film reels is linked to the synchronization of the speech with the video. The communication processor allows a one second difference between voice and video synchronization

2) Prompt the user to enter a phone number associated with the remote workstation. Video conference routing uses the phone numbers in conjunction with specific names and passwords and call acceptance criteria on a time basis. The communication processors exchange this information among themselves in the superframe in order to set up connection command tables. The phone numbers are important because there is a large investment in the phone number database. The communication processors are programmed with the identity of each user service. This information is actively shared by the individual communication processors. Part of this information is the user reception status. The communications processor will not attempt a call if the receive status is set to "No". Since the communications processor network is an important part of the user directory, the "Who is Who" directory begins with a geographical map that shows the locations with the user concentrations. A software-controlled dial-up to one of these locations calls up the names and telephone numbers of the users at this location. There is also a receive status indicator that prevents users from wasting time and effort. This information is exchanged among the communications processors using free superframe total capacity.

Second distinguishing element: workstation video specifications.

One solution, with the possibility of reducing costs, is the division of the optical and electronic camera functions with the aim of accommodating the camera electronics on the communication port. The basic equipment of a camera for converting images into signals includes focusing optics and image conversion electronics. Such a camera can be implemented using focusing optics in conjunction with a fiber optic focal plane image capture device on the communication processor. The purpose of this method is to use only high-speed electronics to serve multiple users. If the imaging device is fast, it can serve multiple users at the same time. The evaluation algorithm and the multiplexing algorithm can be used simultaneously on the communication processor. Fig. 2, entitled "Video input on the workstation", shows the interface between the image capture circuitry located on the workstation. The imaging lens and the image acquisition system have approximately the size of a video endoscope or a corresponding modification of such a device. To keep the depth small, the primary lens can be designed to be connected to a right angle mirror. The mirror can achieve the desired focal length along the width of the front ring. In this way, the front ring depth can be small. The imaging device or the electronic focal plane are then at right angles to the image. There is also the option of not accommodating the electronic focal level in the front ring, but in the communication processor. The imaging lens and the detection system then transmit the image in optical form in an optical fiber to the communication processor. Both techniques work, but in terms of implementation costs, the electronic focal level in the front ring would be the preferred method; however, the second option is further developed in the section on the alternative solution.

Let us return to the preferred embodiment; Let us assume that we proceed according to the conventional method, using a miniature camera housed in a front ring on the workstation screen, as described under 3). In this case, only a single wire cable is required for the connection between the workstation and the communication processor. The bandwidth of an average RF cable is sufficient to support frequency division multiplexing for transmission to receive video, audio and data. Figure 10 shows that using this method, the interface to the communications processor is simple compared to the other types of signals transmitted on the cable.

One of the keys to the invention is the video processor rating algorithm required to maintain image quality and conserve transmission bandwidth. Compression is also required for evaluation. Compression and evaluation are not mutually exclusive, but must be designed together for compatibility. Let us consider the following case; an image is divided into scan zones and a distinct central zone is defined as the area in which the eye remains more than 50% of the time during focusing; a clear concentric peripheral zone is defined as the area in which the pupil remains 35% of the time during focusing. The remaining concentric peripheral zone is the remaining 15%. FIG. 11 is a graphical representation of an applicable scoring scheme.

The picture frame, which consists of a number of N x J vertical and horizontal scan lines, is digitized in M ​​resolution bits. As an example, M stands for the numerical value 12 and N = J = 525 picture elements. Then each picture element is represented by 12 resolution bits. The total number of bits is (N * J * M) or 3145728 bits. A serial channel that transmits at a speed of 1,544 Mbps would only take a little over 2 seconds to transmit. If the compression allows a reduction of this number by 15 times, the number is 209715. The first video frame is only transmitted in compressed form, since there is no previous frame for the comparison. The next frame is evaluated in comparison with the first by subtracting it from the first frame, for which purpose the individual picture elements are compared piece by piece. If the resulting value is less than a predetermined delta amount, the resultant for that pixel is zero. The position of the first picture element, which is not equal to zero, is noted with (x, y) of (N, J). The position of the next picture element at position (x, y + 1) should also be non-zero. Sigma is the deviation calculated from the next r scan lines from non-zero picture element comparisons, where r times s is about 50% of the central picture part and 50% of the total number of picture elements maximum or less, if calculated , Therefore, (x, y) to (x + s, y + r) is the central part of the picture. To calculate a starting point for the transmission of 12 bit numbers, however, x- is selected for the value of the initial row and y for the column value. The last value in this sequence is x + s and y + r. The total number of bits is (256 * 256 * 12) or 786432 bits maximum or less if the comparisons result in zero values ​​before 50% of the central image area occurs first. If it is subsequently compressed, the maximum number of bits is reduced by an order of 15 (average compression possible) to 52,428. Since we chose DSO as the base carrier rate, 11,570 bps are available for voice and overhead. 11 shows a possible evaluation algorithm.

After digitization, a video evaluation is carried out. This algorithm assumes that the video is 640 pel by 480 pel; however, it could also be a different number of pel. Together this results in 307,200 pels. Of this number, as any definition, 50% will get 12 resolution bits, 35% 8 resolution bits, and 15% 4 resolution bits. Values ​​12, 8 and 4 were chosen to clarify the description of the rating; however, several such values ​​could be used by the communication processor. The image is transferred in shades of gray, with the central part showing 4096 shades, the remaining parts 256 or 16 shades. Each area is compressed using a compression algorithm with limit dependency.

Compression factors from 50 to 100 to one are possible in real time, but are not required for the communication processor in real time. The image can be presented as a series of images at a rate that is updated in proportion to the request, bandwidth cost, and system usage conditions for the service. The signal / noise ratio of the image transmission is important for user acceptance. To improve the signal-to-noise ratio at the source, a technique is shown in which a weak laser is used to illuminate the object in the short to medium IR short wavelength range.

Illuminating with Fresnel scattering lens is a good way to get a high photometric value while maintaining a low radiometric level. It is recognized that eye safety can be assured if the radiometric level can be kept low with the Fresnel scattering lens. At the same time, the contrast transmission of the receiver should be as far as possible, from the visible to the medium short wavelength in the IR range. FIG. 12 shows the integration scale of the receiver contrast behavior and the effect of integrating all the wavelengths shown. In practice, the imaging systems do not control the lighting, except perhaps in the sense that they have enough light to take a good picture. The spectral content and short-term behavior play an important role in image processing. The goal of image lighting is to exclude stray lighting, such as the AC characteristics of the power line, and the blue / green spectral content in fluorescent lighting, which would mean more computing for the communications processor. Office lighting depends on AC power and fluorescent light sources for the primary source.

The integration of the amplitude over the frequency bandwidth results in a signal with a high signal / noise value. This signal is reproduced at the remote station as an image with black and white grayscale. Strictly speaking, white is not a monochromatic color, but can be considered as such in this case since it is only a single value function of the amplitude (the area under the curve of Fig. 12). This solution was simplified for reasons of space and time. It is recognized that the final signal is a convolution of the function of the source illumination, the fiber optic bandpass behavior and the bandwidth of the detector.

Third distinguishing element: audio management specification.

The communication processor can be connected to the workstation via quality coaxial cables. The cable is divided into several frequency management units using frequency division multiplexing. As a result, voice data can be transmitted in duplex mode via the cable, it can work with feedback for the reception of image data in simplex mode, data can be transmitted in duplex mode and network control data in duplex mode. FIG. 14 shows how language is processed at the PSO level.

Fig. 7A shows the general configuration of an overall network of the system unit for the audio / video communication processor shown in Fig. 7B. Generally, it will be seen that the system provides one or more communication processors that are responsible for a group of workstations for processing in the transmission of audio and video information in a video conference circuit.

The communications processor uses AI software to read the connection. The conversion rules are in tables so that the system can react to the communication environment. The system is coupled for the processing of optical signals for cost-effective communication and video conferencing with audio and video communications in the short and long-distance transmission range. The communication processor takes care of the audio and video communications taking into account the current constraints of the transmission medium and the current level of utilization or utilization. The bandwidth, the resolution and the transmission speed can be adjusted to match the constraints at the time a service request is received. A workstation initiates a service request. A service request contains data about the type or type of service and the signal destination. This information is sufficient for the communications processor to perform multiple thread attempts before an affirmative determination can be made. If an affirmative determination is not possible, the communications processor determines what is possible and issues a message to the user so that the user may be able to change his request. Referring to Fig. 7B, it can be seen that the audio-video communications processor has network ports, a local port, and the system communications processor unit, which is equipped with the additional units connected to the system bus. Connected to the bus is a common superframe processor, an audio / video statistical multiplexing processor, a channel frame processor and the workstation subprocessors, which include the video processor with evaluation and compression, the audio processor with compression and evaluation, and the workstation interfaces - circuits. The workstation subprocessors have both analog and digital interconnections. The system has telephone interface ports for connecting the workstations to the overall system.

The telephone interface is an outstanding concept because it enables the independent units, e.g. B. facsimile, a connection to the workstations. The workstation can work with fax emulation software that converts documents in electronic form into a fax signal and sends them to the remote fax machine. The telephone interface serves to support the audio packets in the network. The telephone interface can accept telephone signaling information and generate signaling information.

The voice signals from the workstation are accepted by the communication processor. One of N voice input lines is accepted for processing by multiplexing. The sampling is carried out in accordance with current practice (1st order DSO carrier) at 64 Kbps. If the description in the routing table indicates that a conversion is necessary, as in a transatlantic video conference, the communication processor automatically carries out a recoding between u and A law. After the evaluation, the signal is processed further in order to reduce the transmission rate and to assemble the resulting data in a numerical sequence suitable for a packet network. The transmission speed of the packets depends on the network load. The communications processor network should not compete with the telephone network, but rather voice and video should be added, with the option to select only language.

A note on the telephone voice bandwidth. The 64,000 bps speed is the result of a Nyquist sampling rate of 8,000 samples per second. Each sample has 8 bits of binary data. Two to eight gives the decimal value 256. Since the electrical voice signal should be symmetrical around the zero voltage axis, 256 is divided into +128, -127 and 0. The sum of these position holding numbers is 256. This means that a full 4,000 Hertz signal bandwidth is available for speech. The type 500 handset normally used in a telephone set has a nominal bandwidth of 3,000 hertz. The first limiting filter is then the handset. A handset or headset that is part of the communication process equipment is not so limited. It has a nominal high-fidelity voice bandwidth of 4,000 hertz. FIG. 14 shows how an audio transmission package would be put together for a video conference. The process, called Linear Predictive Code Compression (LPC), can reduce the bandwidth of the speech data to a serial rate of up to 1,200 bps. This language is poor in terms of elements such as speech recognition and lost words and causes the speech to be repeated if the transmission medium performs poorly. Since the medium is expected to be of high quality, the only deterioration of the 1200 bps language is in the algorithm. Speech is processed with 2,400 bps LPC and 16,000 or 32,000 bps adaptive differential pulse code modulation (ADPCM) and 64,000 bps PCM. The choice depends on the available bandwidth, the requirements of the user and the compatibility of the remote station. These factors are tracked by the AI ​​software running on the communication processor. FIG. 4 shows how variable partitioning is used. Variable partitioning reduces redundant processing and saves bandwidth.

If a fax transmission is desired, the audio signal can be replaced by a fax signal. The front ring or workstation adapter has an input for fax signals. The communications processor does not generate fax signals, but a side of the workstation that is not linked to the communications processor interface adapter can generate fax signals.

The dynamic allocation of bandwidth is based on audio as the first priority if the signal is a fax signal, taking into account the network load. Periods without audio activity are used to transmit video signals. Audio signals are compressed dynamically from 64,000 bps using a good predictive algorithm. The expectations for voice data transmission are between 2,400 and 1,200 bps. The audio compression can also be zero, so the rate would be DSO or 64,000 bps, as would be the case with fax signals. The packet control section of the workstation is also dynamic. If the information is redundant, retry semaphores from the previous packet are used to convey this information.

The communications processor must collect all fax packets before sending them on to the workstation. The fax packets must not be discarded, as is the case with the audio or video packets. A buffer may be required for this function. A buffer is also required for speech processing. The video signal does not require a buffer. A video package can be discarded if desired. The workstation does not assemble a new frame until the frame is complete and intact. It is recognized that video image hardware under a variety of teleconferencing stations must be upgraded over telephone lines, a point which was not addressed by US-A-4,682,225.

The communication processor with the video processor discards picture elements or pels. They are discarded because the receiving station requested a window size for the video. Pels that do not have an address in the window as determined by an addressing algorithm are discarded. This is the first step in reducing unnecessary bandwidth. All pels that have an address in the remote party's window are processed by compression or evaluation. No single representative pel is used to represent a group of pels.

Fourth distinctive element: network management specifications.

The key to the concept of a communication processor is to embed the compatibility of the network standards between the different transmission systems (standard bridge). The T-carrier system is common and tariffs are available for this service. T carriers can be used in ISDN networks. The Synchronous Optical Network (SONET) uses a different multiplexing scheme, which could also carry ISDN signals. The lowest level is OC-1 at 51.84 Mbps. This level can be used for the local loop (LATA) or as feed into long-distance networks. It has a high capacity for high quality communications and is the preferred medium. The communications processor must be compatible with signaling techniques such as DS2 / DS3. The communication processor has no connection to a telephone switch part at any level, but the long distance network transmission equipment can take care of the signaling pattern and could immediately replace the network routing information. Knowing the tariffs, the AI ​​software knows when a carrier is able to do this and therefore does not allow critical control information to be assigned to those bit positions that are most receptive. This property also applies to the ISDN carrier channel.

An FDDI carrier is the preferred choice for operations in local access lines, i.e. those that are located in a company. One would not have to rely on a telecommunications carrier for this choice. 16 shows how the local access line and the telecommunications network differ from one another. It is also possible for a regional carrier to offer a 64,000 bps carrier from the telephone system but no telephone connection. Such a connection would likely be inserted and dropped by a DACS machine before the T-1 carrier could connect to the telephone switch parts.

Let us consider the connection possibilities of Fig. 16. The current constraints of CP2 to CP4 differ enormously from those of CP3 to CP5. The services available from CP2 to CP4 are reduced accordingly. CP2 may be able to use a fixed DSO connection. Another option is an ISDN interface. Each has its own characteristics, throughputs and protocols.

The communications processor can connect to everyone on the network sum side. The key to this is the AI ​​software, which keeps track of which tariff offer is connected to which network port with all its capabilities and restrictions. The communication processor must also maintain the connection information, an archive file network connection with attributes such as bit error rate and error-free seconds, in order to select an error correction code which leads to the lowest possible transmission overhead. The selected code is sent in the packet control part of the frame. Expert rules are drawn up to take account of error statistics. If the bit error rate is at least 10-5 or better and / or if the error-free seconds are 10 or less, ECC cannot be used. Note that the error distribution has more effect in choosing a suitable EEC than the size of BER. Referring to Fig. 17. The communication processor is opportunistic, an analogy, similar to the Meteor-Bundle communication, the communication processor can take advantage of a good runtime up to the transmission rate by reducing the overhead. Figure 17 is an example of fault data statistics that could be compiled by the communications processor. Figure 17 shows how the block size can affect the quality of the transmission. The automatic repeat request is a measurable parameter. In addition, a given protocol can keep track of how often it must intervene to replace an error bit. The communication processor can track these statistics and make changes to the protocol selection so that a maximum data bandwidth and a correspondingly good quality can be achieved.

In the case of expert communication processing (AI), error statistics must be taken into account. A key factor for throughput performance is the quality of the connection. The bit error rate alone is insufficient for the selection of an appropriate error correction code (ECC).

Error-free seconds (EFS) and an error distribution across the packet size are required in addition to the BER (bit error rate). It is important that the communication processor measures these values ​​and sets up an archive for each network port. The choice of block size would also be decided in connection with ECC.

Let's take the following example. A key derivative factor to be calculated is the throughput per block size, which must include a factorization of the dynamic block growth by adding in ECC. For example, in Fig. 17, the block size grows geometrically to be added in an ECC candidate (other ECCs can grow arithmetically or linearly). 17 shows the archive of a network port. The statistics show that at an average of 50% a block of 116k size contains a sequence of 7 consecutive wrong bits. This means that one of two blocks is corrupted. If an ECC can correct 6 wrong bits in a sequence, including a continuous sequence, then that ECC would be chosen, along with a smaller 48k block size, as appropriate for the present conditions. ECC is only used in comparison with tax information. Audio or video data does not receive an ECC. However, the error statistics for the control information apply equally to voice and video data when calculating the block size. If the situation is such that ECC is no longer effective, then a dynamic evaluation of the connection is required so that the call can be ended due to unsatisfactory performance before the normal service is ended. A call can be tried several times in an hour with the intention of collecting statistical information if bandwidth space is available. These call attempts are not necessarily initiated by the user, but can be based on previous usage patterns. The statistics of the trials are saved and are used to calculate the success of the ECC types as the probability of success in later trials.

A summary of the network management plan shows the key elements of the plan.

1) Description table for each interface on the network port side, all parameters and characteristics of the tariff service.

2) Jobs-in-service table, which contains all parameters that cause a service.

3) Service request table (proposed channels in service) queued according to arrival time, bandwidth request and connection complexity.

4) Quality performance table (contains the measured performance statistics and the calculated statistics) for each network port.

5) Calculation of the requested services and customized requirements for each user in a performance table, which concern each request in a table (for adding new channels).

6) Calculation of limits of the performance table of the total frame per network interface port. (System stops).

7) Model of each current total port as well as collected data for the projection of the bandwidth requirements and for the calculation of the bandwidth allocations. Develops all the main parameters required by the sub-processors.

Other characteristics

It is a feature of the invention that the communications processor shares the video subprocessor in the communications processor with multiple users. How this division works is not important here, for example analog switching to an A / D or discrete A / D per workstation connection. The processing of video with evaluation or compression of several users is carried out by a communication processor with the subordinate help of a video processor for all users.

The communication processor uses a video subprocessor to select the picture elements of each user according to the specifications set by the communication processor. The video subprocessor always endeavors to reduce the bandwidth as much as possible, so that transmission capacity is available for the next potential user. Depending on the conditions, the video is either compressed or rated. Evaluated means that the assigned dynamic range of the picture element is reduced to an active dynamic range, which in practice looks such that fewer bits are distributed over the entire assigned dynamic range. For example, if 12 bit positions suppress the full range and the Pel has range fluctuations of 4 bit positions, then only the 4 bits will be transmitted. A special protocol tells the receiving end which 4 bits of the assigned 12 bits are transmitted. Each pel also has a specific address at the image storage level of the workstation. No information has to be transmitted for vertical or horizontal synchronization. The communications processor hardly needs to worry about compression or the techniques of such compression. The compression can be achieved via the hardware or the software, the type of compression is of no importance for the invention.

It should also be noted that, unlike US-A-4,733,299, the system has traffic types that are the same. The communication processor is not concerned with successive or interlocked scanned video images, but only with the video data. Some workstations may use interlaced video, others may use successive video. All users of the communications processor conduct video conferencing regardless of the type of scan. The basic video conferencing structure of the combined page frame is the same for all users, except that some users are allocated more bandwidth than others. There are several reasons for this, but when a connection is made, bandwidth is allocated for the duration of the work session. There may be instantaneous fluctuations in bandwidth, either in the audio or video area, but these are not a factor for the user. Each user has a channel in a combined page frame that can only be decoded by another communication processor. The communications processor ideally uses high speed digital carriers such as T1 and other types on the combined side, but these cannot be processed by the transmission equipment as DS1, DS2 or DS3 carriers, such as a digital access cross connect, to to switch on the voice or data switching points. Nor can they be processed as a variable fractional T1 service. If the service on the combined side is a fractional T1 service, then it must be a fixed fractional service.

Suppose a group of optical lenses is used to focus the image on the end of a fiber optic cable that is connected to the communications processor. The quality of the optical fibers differs from the quality of those fibers that carry high speed binary data. The fiber must maintain the rectilinear spatial image of the focal plane, disregarding the time scatter along the Z axis. There are fiber optic lenses that have a large focal plane and in which the lens diameter gradually narrows into the cable diameter. The cable ends at the communication processor in a lens arrangement and an electronic aperture. The aperture is positioned so that the image is projected onto a facet of a multi-level entrance and single level exit lens (complex prism). The output of the complex prism is an arrangement of a focusing lens in the focal plane detector. 11 shows the evaluation.

The video controller selects the electronic iris in a multiplex sequence, captures the image from the focal plane detector (CCD) in analog form and converts the signal into digital format. High-speed CCD performance provides a cost advantage for asynchronous multi-user operation, but is not essential for the performance of the communications processor.