-
公开(公告)号:KR100363889B1
公开(公告)日:2002-12-11
申请号:KR1020000004137
申请日:2000-01-28
Applicant: 한국전자통신연구원
IPC: H04B1/7083
Abstract: CDMA 통신시스템은 정보를 확산코드인 의사난수(PN) 코드를 사용하여 대역확산하여 전송한다. 따라서 CDMA 통신시스템의 전송정보를 복구하기 위하여 반드시 확산코드의 정보를 알아야 하는데, 확산코드인 의사난수 코드에는 여러 종류의 지연이 발생하기 때문에 이를 제거해야 한다. 이를 위하여 종래에는, 확산신호를 제거한 역확산 신호를 비동기 검파 및 동기 검파를 위하여 주어진 처리이득만큼 더한 후 포함된 위상오차를 제거하기 위한 제곱기 및 곱셈기를 이용하여 위상 오차를 제거한다. 그러나, 이 방법은 처리하여야 할 제곱 및 곱셈의 수가 증가하면 이를 처리하기 위한 복잡도가 증가한다.
본 발명에서는, 역확산과 동시에 위상오차 성분을 제거를 하는 바, 즉, 입력데이터를 확산코드를 이용하여 복소수 곱을 하여 확산코드를 제거하고, 이 신호를 레지스터에 위상별로 각각 저장하여 지연시키고, 복수역확산 데이터와 레지스터에 의해 지연된 위상별로 곱셈기를 통해 곱한 다음, 두 결과값을 더한다. 이러한 본 발명은, 여러개의 확산코드를 동시에 검색할 때 하드웨어 사용을 줄일 수 있으며, 여러 코드위치에 대한 동시검색과 여러 가지 종류의 코드에 대한 동시 검색이 가능해지고, 하드웨어를 간단하게 구현할 수 있다.-
公开(公告)号:KR100355452B1
公开(公告)日:2002-10-11
申请号:KR1019990031257
申请日:1999-07-30
Applicant: 한국전자통신연구원
IPC: H03M13/27
Abstract: 본발명에서는터보부호를사용하는터보복호기를 ASIC으로구현하는데하드웨어자원공유를통하여면적을최소화하기위한 MAP(Maximum A Posteriori) 방식을사용한터보복호기에관한것이다. 본발명에서는종래의두 개의 MAP 복호기를하나의 MAP 복호기로사용하여공통하드웨어자원으로만들어하드웨어크기를최소화한것으로서, 그구조는제 1 RAM(20), 제 1 가산기(21), 제 2 멀티플렉서(22), 제 1 가산기와제 2 멀티플렉서에서출력된데이터를받아순방향과역방향메트릭을동시에계산하여입력정보를혼합된순서로 MAP 복호해서출력하는하나의 MAP 복호기(30), 하나의지연소자(23), 제 1 멀티플렉서(24), 제 2 가산기(40), 제 2 RAM(50), 인터리버, 디인터리버, 역인터리버, 역디인터리버(41∼44), 그리고최종터보복호된정보를출력하는경성판정부(60)로구성된다. 이와같이구성되어, 본발명에서는복호시간의지연이없이하드웨어의자원공유를통하여터보복호기의크기를획기적으로줄일수 있다. 이에따른터보복호기는차세대이동통신시스템인 IMT-2000 규격에채택된오류정정분야에있어서경쟁력을확보할수 있게된다.
-
公开(公告)号:KR1020020050395A
公开(公告)日:2002-06-27
申请号:KR1020000079534
申请日:2000-12-21
Applicant: 한국전자통신연구원
IPC: H04N19/53
Abstract: PURPOSE: A system and a method for estimating motions using hierarchy search are provided to reduce the size and circuit area of a memory and decrease consumption power. CONSTITUTION: A system for estimating motions of a moving picture includes a processing element block(260) and a comparator(270). The processing element block receives reference data and current data to obtain a motion vector with which the absolute value of the difference between the reference data and current data becomes a minimum value. The system further includes a down-sampling unit(200,210) and a memory unit(230,240,250). The down-sampling unit down-samples the reference data and current data. The memory stores the down-sampled reference data and current data and provides the stored data to the processing element block.
Abstract translation: 目的:提供一种用于使用层次搜索来估计运动的系统和方法,以减小存储器的尺寸和电路面积并降低功耗。 构成:用于估计运动图像的运动的系统包括处理单元块(260)和比较器(270)。 处理元件块接收参考数据和当前数据,以获得参考数据和当前数据之间的差的绝对值成为最小值的运动矢量。 系统还包括下采样单元(200,210)和存储单元(230,240,250)。 下采样单元对参考数据和当前数据进行缩减。 存储器存储下采样的参考数据和当前数据,并将存储的数据提供给处理元件块。
-
公开(公告)号:KR1020020032157A
公开(公告)日:2002-05-03
申请号:KR1020000063098
申请日:2000-10-26
Applicant: 한국전자통신연구원
IPC: H03H17/06
CPC classification number: H03H17/0607 , H03H17/0226 , H03H17/0621 , H03H17/0657
Abstract: PURPOSE: A finite impulse response filter without using multiplier is provided to process an FIR filter operation a look-up table method suitable in a high speed operation without using a multiplier. CONSTITUTION: An input shift register and selector(100) shifts and stores 4 bit filter input, and selects one among the stored input data in response to a clock signal. An address generator(200) generates an address suited to a look-up table of each counting group in response to the data from the input shift register and selector(100). Look-up table groups 0, 3, 1, and 2(300,400,500,600) receive an address of a corresponding group from the address generator(200) and generate a filter output corresponding to the address by a look-up table. Integrator group 0, 3, 1, and 2(700.800,900,1000) shift and integrate results outputted by each counting groups by an input bit number. 4x1 multiplexor(1100) serially outputs of the integrator group 0, 3, 1, and 2(700.800,900,1000) in response to a control signal.
Abstract translation: 目的:提供一种不使用乘法器的有限脉冲响应滤波器来处理FIR滤波器操作,适用于高速运行的查找表方法,而不使用乘法器。 构成:输入移位寄存器和选择器(100)移位并存储4位滤波器输入,并根据时钟信号选择存储的输入数据之一。 响应于来自输入移位寄存器和选择器(100)的数据,地址发生器(200)产生适合于每个计数组的查找表的地址。 查找表组0,3,1和2(300,400,500,600)从地址生成器(200)接收相应组的地址,并通过查找表生成对应于地址的过滤器输出。 积分器组0,3,1和2(700.800,900,1000)将每个计数组输出的结果移位并整合输入位数。 4×1多路复用器(1100)响应于控制信号串行输出积分器组0,3,1和2(700.800,900,1000)。
-
145.
公开(公告)号:KR1020020010845A
公开(公告)日:2002-02-06
申请号:KR1020000044405
申请日:2000-07-31
Applicant: 한국전자통신연구원
IPC: H04L27/30
CPC classification number: H03H17/0621 , H03H17/0607
Abstract: PURPOSE: A 108 tap 1:4 interpolation finite impulse response filter device for digital mobile communication is provided to be capable of simultaneously performing 4 filter operations without increasing the speed of operation frequency by applying a pipeline technique and a look-up table technique. CONSTITUTION: An input shift register and selector(100) shifts bit filter inputs and outputs 27-bit parallel data, and sequentially the output parallel data one by one. An address generator(200) receives the 27-bit parallel data to generate addresses according to respective counting groups. Look-up table groups(300,400,500,600) generate the filter outputs of count groups by using the addresses generated in the address generator. A pipeline register(700) delays the filter outputs per count group output in parallel from the look-up table groups. A group selector(800) converts the outputs delayed from the pipeline register(700) into DC outputs. A pipeline register(900) delays the outputs of the group selector to match the filter output times of respective channels.
Abstract translation: 目的:提供一种用于数字移动通信的108抽头1:4内插有限脉冲响应滤波器装置,可以通过应用流水线技术和查找表技术,同时执行4个滤波操作而不增加操作频率。 构成:输入移位寄存器和选择器(100)将位滤波器输入和输出移位27位并行数据,并逐个依次输出并行数据。 地址发生器(200)接收27位并行数据,以根据各个计数组产生地址。 查询表组(300,400,500,600)通过使用地址生成器中生成的地址生成计数组的过滤器输出。 流水线寄存器(700)从查找表组并行延迟每个计数组输出的滤波器输出。 组选择器(800)将从流水线寄存器(700)延迟的输出转换成直流输出。 流水线寄存器(900)延迟组选择器的输出以匹配各个通道的滤波器输出时间。
-
公开(公告)号:KR1020010076623A
公开(公告)日:2001-08-16
申请号:KR1020000003874
申请日:2000-01-27
IPC: H03M7/02
CPC classification number: H03M13/2957 , H03M13/3905
Abstract: PURPOSE: A method of normalizing a turbo decoder by using maximum a posterior is provided to reduce a critical path and not to saturate a state metric with a positive value, when the turbo decoder is normalized by using hardware. CONSTITUTION: First, a branch metric is prepared by adding a constant value on a branch metric value of 2 digits to obtain a positive value(S11). The branch metric value is inputted into a metric calculator(S2), and the inputted branch metric value is calculated with a state metric(S12). Then, a value for the normalization is extracted from a previous state metric, and a current state metric is normalized by the extracted value. Accordingly, the performance of the turbo decoder is improved by embodying with the hardware and a critical path does not elongated.
Abstract translation: 目的:提供一种通过使用最大后验来对turbo解码器进行归一化的方法,以便在通过使用硬件对turbo解码器进行归一化时,减少关键路径,而不使饱和具有正值的状态度量。 规定:首先,通过在2位数的分支度量值上加上常数值来获得正值来准备分支度量(S11)。 分支度量值被输入到度量计算器(S2),并且用状态度量计算输入的分支度量值(S12)。 然后,从先前状态度量提取用于归一化的值,并且通过提取的值对当前状态度量进行归一化。 因此,通过使用硬件来实现turbo解码器的性能,并且关键路径不会延长。
-
公开(公告)号:KR1020010076563A
公开(公告)日:2001-08-16
申请号:KR1020000003768
申请日:2000-01-26
Applicant: 한국전자통신연구원
IPC: H03M7/04
CPC classification number: H03M13/2957 , H03M13/3905 , H03M13/6502
Abstract: PURPOSE: A turbo decoder using base 2 log MAP(maximum a posteriori probability)is provided to reduce size of the hardware without degradation of performance by converting complex E function to simple 2 function. CONSTITUTION: The turbo decoder extracts respective excess information from the first and second information sets using the maximum a posteriori probability(MAP) algorism. The turbo decoder has a 2 function device. The 2 function device transforms the MAP algorism using base 2 logarithm. The turbo decoder adds 2-exponential values of the respective first and second information sets each other, defines 2 function taking base 2 logarithm to the added value and calculates the 2 function in the transformed MAP algorism. Therefore it reduce size of the hardware without degradation of performance by converting complex E function to simple 2 function.
Abstract translation: 目的:提供使用基础2 log MAP(最大后验概率)的turbo解码器,通过将复杂E函数转换为简单的2函数来减少硬件的大小,而不会降低性能。 构成:turbo解码器使用最大后验概率(MAP)算法从第一和第二信息集提取相应的多余信息。 turbo解码器具有2功能设备。 2功能设备使用基数2对数来转换MAP算法。 turbo解码器将相应的第一和第二信息集合的2-指数值相加,定义了将基数2对数作为附加值的2个函数,并计算变换MAP算法中的2函数。 因此,通过将复杂的E功能转换为简单的2功能,可以减少硬件的尺寸,而不会降低性能。
-
公开(公告)号:KR1020010011737A
公开(公告)日:2001-02-15
申请号:KR1019990031258
申请日:1999-07-30
Applicant: 한국전자통신연구원
CPC classification number: H04B1/708 , H04B1/70735 , H04B7/2628 , H04J13/0022
Abstract: PURPOSE: A high-speed parallel code searcher for code division multiple access(CDMA) communication systems is provided to implement an initial synchronization searcher in parallel code search by using a small amount of hardware, and to perform parallel processing by using the implemented code searcher. CONSTITUTION: A signal converter(210) receives in-phase data and quadrature-phase data for decimation. A local diffusion code generator(700) consecutively generates local diffusion codes for reverse diffusion. A parallel correlator(200) separates phase error components from the decimated signals, and multiplies the separated components by the diffusion codes generated in the local diffusion code generator(700) for reverse diffusion for integration in parallel by a microprocessor(800). An energy value calculator(300) squares and adds an integrated result for non-coherent calculation, and simultaneously calculates energy values in code positions in parallel. A local maximum energy value detector(400) sequentially selects the energy values calculated in the energy value calculator(300), to detect a local maximum energy value by the code search block without setting up a reference value. If the detected local maximum energy value is larger than a presently stored energy value, a maximum energy value storage(500) stores the detected local maximum energy value as a new local maximum energy value. The microprocessor(800) sets initial values of the number of integration blocks, coherent integration and code search blocks, and controls the diffusion codes to be consecutively outputted if the input data is inputted. And the microprocessor(800) controls the integration of the parallel correlator(200), and reads the stored local maximum energy value every end of search corresponding to the number of the code search blocks, to decide whether to detect the local maximum energy value. And a control value storage(600) reads or writes a control value of the microprocessor(800).
Abstract translation: 目的:提供用于码分多址(CDMA)通信系统的高速并行码搜索器,通过使用少量的硬件来实现并行码搜索中的初始同步搜索器,并通过使用实现的码搜索器来执行并行处理 。 构成:信号转换器(210)接收用于抽取的同相数据和正交相位数据。 局部扩散码发生器(700)连续地产生用于反向扩散的局部扩散码。 并行相关器(200)将相位误差分量与抽取的信号分开,并将分离的分量乘以在局部扩散码发生器(700)中产生的用于反向扩散的扩散码,以便由微处理器(800)并联。 能量值计算器(300)对非相干计算进行平方并加上积分结果,同时并行计算代码位置的能量值。 局部最大能量值检测器(400)顺序地选择在能量值计算器(300)中计算的能量值,以便通过代码搜索块检测局部最大能量值而不设置参考值。 如果检测到的局部最大能量值大于当前存储的能量值,则最大能量值存储(500)将检测到的局部最大能量值存储为新的局部最大能量值。 微处理器(800)设置积分块数量,相干积分和码搜索块的初始值,并且如果输入数据被输入则控制要连续输出的扩散码。 并且微处理器(800)控制并行相关器(200)的积分,并且对应于代码搜索块的数量的每个搜索结束读取存储的局部最大能量值,以决定是否检测局部最大能量值。 并且控制值存储(600)读取或写入微处理器(800)的控制值。
-
公开(公告)号:KR1020010009726A
公开(公告)日:2001-02-05
申请号:KR1019990028261
申请日:1999-07-13
IPC: H03M13/00
CPC classification number: H03M13/2732 , H03M13/6502
Abstract: PURPOSE: A channel encoder for the digital communication is provided to carry out a convolutional-encoding and an interleaving operation at once by using two LAMs for buffering a frame input data having the low memory usage alternately. CONSTITUTION: The device carries out a convolutional-encoding and interleaving operation of a frame input data at once by using two encoder RAMs(417). A microcontroller controls to store a frame input data input to a frame input data register(411) into the first encoder RAM(416) while parallel-inputs into a parallel CRC generator(412) generated by the number of a given CRC input bit and a production polynomial. A device carries out an XOR logic-operation of the input parallel CRC input value and the previous CRC state value to parallel calculate a desired CRC output value. A device stores the CRC output value with the frame data stored in the first encoder RAM and reads the stored CRC output value to carry out the convolutional encoding and the interleaving. The convolutional encoding and interleaving for the frame input data, and storing the next frame input data into the second encoder RAM are carried out at once.
Abstract translation: 目的:提供用于数字通信的信道编码器,通过使用两个LAM来交替地缓冲具有低存储器使用的帧输入数据来一次执行卷积编码和交织操作。 构成:该装置通过使用两个编码器RAM(417)一次执行帧输入数据的卷积编码和交织操作。 微控制器控制将输入到帧输入数据寄存器(411)的帧输入数据存储到第一编码器RAM(416)中,同时并入输入到由给定CRC输入位的数量产生的并行CRC生成器(412) 一个生产多项式。 设备执行输入并行CRC输入值和先前CRC状态值的异或逻辑运算,以并行计算所需的CRC输出值。 设备将CRC输出值与存储在第一编码器RAM中的帧数据进行存储,并读出所存储的CRC输出值,以执行卷积编码和交织。 帧输入数据的卷积编码和交织,并将下一帧输入数据存储到第二编码器RAM中一次进行。
-
公开(公告)号:KR100204597B1
公开(公告)日:1999-06-15
申请号:KR1019960064135
申请日:1996-12-11
Applicant: 한국전자통신연구원
Inventor: 조한진
IPC: H03J1/00
Abstract: 본 발명은 주파수 혼합기의 출력 단에 LC 병렬 공진 회로를 연결하여 저역 통과 필터의 필요성을 제거하고, 동시에 전력 이득을 극대화 하고자 하는 저역 통과 필터를 내장한 고성능 주파수 혼합기에 관한 것이다.
LC 병렬 공진회로는 공진 주파수에서 무한대의 임피던스를 갖고 그 이외의 주파수 대역에서는 적은 임피던스를 갖는 특성을 가지고 있다. 따라서, 본 발명의 주파수 혼합기에 연결된 LC 병렬 공진회로가 원하는 주파수 대역에서는 큰 임피던스를 갖고 나머지 대역에서는 적은 임피던스를 갖게 설계하면 혼합된 주파수 대역에서는 큰 전력 이득을 가지며 나머지 주파수 대역의 신호들은 크게 감소하게 된다. 그러므로 기존에 주파수 혼합기 출력에 필요하던 저역 통과 필터는 불필요하게 된다.
-
-
-
-
-
-
-
-
-