Abstract:
PURPOSE: A three dimensional high speed of round-robin scheduling method is provided to allow all clients to occupy a public multi-communication line with an equitable utilization ratio and to utilize all public multi-communication lines with a maximum utilization ratio. CONSTITUTION: In a three dimensional high speed of round-robin scheduling method, Requests signals from all clients are arranged with a sequence of a two dimensional request matrix and thereby a two dimensional matrix which is consisted of rows(N) corresponding to the number of all communication lines and columns(M) corresponding to the number of client groups is formed. The NxM request matrix is divided by an M unit and thereby a K number of MxM request matrix is formed. A K number of MxM request matrixes is copied as the number(K) of the pubic communication lines in which a client group can use simultaneously. A KxKxM number of processors are parallel-arranged and the MxM request matrix is parallel-processed. A two dimensional parallel round-robin scheduling is performed using a round-robin sequence with same start element and a three dimensional parallel round-robin scheduling is performed using a round-robin sequence with a different start element number, thereby obtaining a three dimensional request element. The obtained three-dimensional request elements are traced using a multistep tracing method to allocate to the public communication lines.
Abstract:
1. 청구범위에 기재된 발명이 속한 기술분야 본 발명은 중재 지연 내성의 분산형 입력 버퍼 스위치 시스템 및 그를 이용한 입력 데이터 처리 방법에 관한 것임. 2. 발명이 해결하려고 하는 기술적 과제 본 발명은, 입력 버퍼에 이중의 파이포(FIFO : first-in-first-out) 버퍼를 두고 중앙 중재기에 요청 파이포(FIFO) 버퍼를 두어 전송지연에 무관하게 발생된 요청에 대해 중재를 수행하는 중재 지연 내성의 분산형 입력 버퍼 스위치 시스템 및 그를 이용한 입력 데이터 처리 방법을 제공하고자 함. 3. 발명의 해결방법의 요지 본 발명은, 입력 데이터 처리수단에서 매칭되는 입력포트로부터 입력 데이터를 받아 목적하는 출력포트별로 저장 관리하는 제 1 단계; 상기 입력 데이터 처리수단에서 상기 입력 데이터에 대한 중재 요청 신호를 중재 수단으로 전송하고 중재 요청 신호가 전송된 입력 데이터에 대한 정보를 저장 관리하는 제 2 단계; 중재수단이 전송받은 중재 요청 신호에 대해 입력 데이터 처리수단 및 목적하는 출력포트별로 관리하는 제 3 단계; 입력 데이터 처리수단과 목적 출력포트에 따라 중재 요청을 확인하여 중재를 수행하고 그 결과를 상기 입력 데이터 처리수단과 스위칭 수단으로 전송하는 제 4 단계; 및 상기 입력 데이터 처리수단이 출력 허가 신호를 수신하여 저장된 입력 데이터에 대한 정보를 확인하고 스위칭 수단으로 전송하여 입력 데이터에 대한 처리를 수행하는 제 5 단계를 포함함. 4. 발명의 중요한 용도 본 발명은 스위치 시스템 등에 이용됨.
Abstract:
PURPOSE: A dispersed type input buffer switch system of an arbitration latency tolerance and a method for processing input data using the system are provided to perform an arbitration with respect to a generated request regardless of delay of a transmission of a request signal and a permission signal by providing a dual FIFO buffer in an input buffer and providing a requesting FIFO in a central arbiter. CONSTITUTION: An input buffer(21) includes a virtual output queue(211), a queue controller(212), a cell address FIFO(213), and an idle queue(214). A central arbiter(22) includes a request matrix(221), a request FIFO controller(222) and an arbitration logic(223). An output cell address FIFO buffer(213) is matched the virtual output queues(211) of the input buffer(21), respectively. The central arbiter(22) has a request matrix(221) storing a request signal per an output port according to input buffers. The request matrix(221) has a request signal FIFO buffer per each request matrix element. In addition, a space division switch(23) switches inputted data in accordance with a given command.
Abstract:
본 발명은 크로스바 방식의 방송스위치에 관한 것으로서, 입력포트에 매칭되며, 유니캐스트 및 멀티캐스트 데이터를 일시 저장하고, 유니캐스트 및 멀티캐스트 데이터를 목적하는 출력포트 별로 전송하기위한 중재요청을 하고, 중재요청이 수행된 패킷 데이터에 대한 정보를 저장 및 관리하는 적어도 하나의 입력버퍼; 적어도 하나의 입력버퍼로부터 중재요청 신호를 받아 그 요청 정보를 저장 및 관리하며, 중재 요청에 따라 중재를 수행하는 중앙 중재기; 입력버퍼로부터 데이터를 전송받아 중앙 중재기의 중재결과정보를 받아 이에 따라 스위칭을 수행하며, 적어도 둘 이상의 크로스바 스위치를 포함하는 크로스바스위치부; 및 크로스바 스위치부의 크로스바 스위치 각각에 매칭되어 출력된 패킷 데이터를 크로스바 스위치 별로 저장하고 관리하는 출력버퍼들로 이루어진 출력버퍼부를 포함함을 특징으로 한다. 따라서 본 발명에 의하면, 방송 패킷 지원 문제를 해결하고 스위치의 고 처리율을 지원함으로써 방송 서비스에 원활히 대처할 수 있으며 병렬 중재기를 이용하고 파이프라인 방식을 이용한 고속 동작을 지원함으로써 고속의 대용량의 스위치를 적은 비용으로 설계할 수 있다.
Abstract:
PURPOSE: A broadcast switch in crossbar type is provided to finish all operations within a processing time of each unit data packet, and to transmit results of each functional block in pipeline method, thereby increasing a switch throughput. CONSTITUTION: At least one input buffer(200,210) matched with input ports temporarily stores unicast/multicast data, transmits an arbitration request to transmit the data to targeted output ports, and stores/manages information on packet data where the arbitration request is performed. A central arbiter(230) receives an arbitration request signal from the at least one input buffer(200,210) to store/manage requested information, and carries out an arbitration process according to the request. A crossbar switch portion(222) receives the data to switch the data according to arbitration result information, and includes at least more than two crossbar switches. An output buffer portion(224) consists of output buffers that store/manage the packet data by crossbar switch.
Abstract:
PURPOSE: A cell/packet switching system configured with planes is provided to configure multiple planes, to compensate an additional time block for bit and word synchronization for output data of an asynchronous cross pointer/cross bar switch for sequential activation and operation. CONSTITUTION: Input/output buffering units(4100,4200,4300) stores cell/packet inputted at a high speed, for waiting. Cross pointer/cross bar switching units(4410,4420,4430) are configured with multiple planes, for interfacing and switching with the input/output buffering units(4100,4200,4300) through a high speed serial bit stream input/output. A cell/packet arbitrating unit(4500) performs arbitration for the output control of the cell/packet stored in the input/output buffering units(4100,4200,4300), outputs shape change information to the cross pointer/cross bar switching units(4410,4420,4430), and sequentially activates the cross pointer/cross bar switching units(4410,4420,4430) to output the shape change information.
Abstract:
PURPOSE: A linear systolic Round-Robin scheduler and its scheduling method are provided to implement a simple circuit structure having an excellent modularity for many shared communication links and perform a 2-dimensional Round-Robin scheduling of a high throughput by scheduling at a high speed many shared communication links with a small circuit size. CONSTITUTION: A Round Rover(310) outputs an output Round Robin seed address(o-seed) which is increased by 1 every Round-Robin scheduling cycle and an input Round-Robin seed address(i-seed) which is increased by 1 every Nth Round-Robin scheduling cycle. A request matrix register block(320) reads a request data of an input port to every output port by one time and propagates it to a rotation block(330). The rotation block(330) rotates the request signals read by line unit every Round-robin scheduling cycle from the request matrix register block(320) by using the output Round-Robin seed address. The rotated request signals are propagated to request propagation blocks(341,342,343). Processing elements(344,345,346) are respectively connected to each request propagation blocks(341,342,343) to form pairs. The request signals propagated from the rotation block(330) are propagated from the first request propagation block(341) to the nth request propagation block(343). In this respect, in each propagation block, the request signals are propagated in a pipeline as the uppermost bit is rotated by single bits to the lowermost bit. An aggregator block(350) is positioned at the end of the overall pipeline stage. The aggregator block(350) determines a use allowance of every input shafted communication link and output shared communication link selected by each pipeline stage and each Round-Robin cycles, and generates a data for configuring a space division switching.