General Datacomm, Inc.

Generaldata Corp

Robert Alan Zakrzewski

James Patrick Kurdzo

A voice server module for an ATM switch is provided and generally includes an interface block which receives and multiplexes high speed preprocessed data, a voice processing block which compresses voice data and which is coupled to the interface, an ATM adaptation layer (AAL) processor which receives data from the voice processing block and which formats the data into ATM AAL2 format, and a management processor which configures the interface block, the voice processing block, and the AAL processor. The voice processing block includes an array of digital signal processors (DSPs), including at least two and preferably eight rows and at least two and preferably eight columns. Each row is also provided with a field programmable gate array (FPGA). Data transmission between (to and from) the FPGAs and DSPs is according to a frame. The FPGAs are utilized to direct data received from the interface block and from the AAL processor to appropriate DSPs for processing. The DSPs programmed to conduct various desired voice processing algorithms, including silence removal or insertion, ADPCM and CELP encoding and decoding, echo cancellation, fax/modem tone detection, etc.

ATM switch voice server module having DSP array

A voice server module for an ATM switch is provided and generally includes an interface block which receives and multiplexes high speed preprocessed data, a voice processing block which compresses voice data and which is coupled to the interface, an ATM adaptation layer (AAL) processor which receives uncompressed data and compressed data and which formats the uncompressed data into ATM AAL1 format and the compressed data into ATM AAL2 format, and a management processor which configures the interface block, the voice processing block, and the AAL processor. The interface means receives data which originated from up to four E1 or T1 lines, and multiplexes the data into a high speed serial data stream. The data stream is then sent both to the voice processing block and to the AAL processor, with the voice processing block ignoring all but the voice data portion of the data stream, and the AAL processor ignoring the voice data portion of the data stream. In this manner, AAL1 data is passed directly to the AAL processor without burdening the voice processing block, while data destined to become AAL2 type data is passed to the voice processing block for compression and then forwarded in a compressed form to the AAL processor.

Growing community of inventors

Meriden, CT, United States of America

Robert Alan Zakrzewski