• 한국기계연구소 소보
• /
• s.19
• /
• pp.93-115
• /
• 1989

Cerebellar Model Arithmetic Controller(CMAC) has been introduced as an adaptive control function generator. CMAC computes control functions referring to a distributed memory table storing functional values rather than by solving equations analytically or numerically. CMAC has a unique mapping structure as a coarse coding and supervisory delta-rule learning property. In this paper, learning aspects and a convergence of the CMAC were investigated. The efficient training algorithms were developed to overcome the limitations caused by the conventional maximum error correction training and to eliminate the accumulated learning error caused by a sequential node training. A nonlinear function generator and a motion generator for a two d. o. f. manipulator were simulated. The efficiency of the various learning algorithms was demonstrated through the cpu time used and the convergence of the rms and maximum errors accumulated during a learning process; A generalization property and a learning effect due to the various gains were simulated. A uniform quantizing method was applied to cope with various ranges of input variables efficiently.

• Proceedings of the IEEK Conference
• /
• 2002.07c
• /
• pp.2067-2070
• /
• 2002

In this paper, a VLSI architecture for transform and quantization module, which consists of 2D-DCT, quantization, AC/DC prediction block, scan conversion, inverse quantization and 2D-IDCT, is presented. The architecture of the module is designed to handle a macroblock data within 1064 cycles and suitable for MPEG-4 video codec handling CIF image formats. Only single 1-D DCT/IDCT cores are used for the design instead of 2-D DCT/IDCT, respectively. 1-bit serial distributed arithmetic architecture is adopted for 1-D DCT/IDCT to reduce the hardware area in this architecture. As the result, the maximum utilization of hardware can be achieved, and power consumption can be minimized. The proposed design is operated on 27MHz clock. The experimental results show that the accuracy of DCT and IDCT meet the IEEE specification.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.13-14
• /
• 2006

In this paper, an efficient butterfly structure for Radix-4 FFT algorithm using DA(Distributed Arithmetic) is proposed. It is shown that DA can be efficiently used in twiddle factor calculation of the Radix-4 FFT algorithm. The Verilog-HDL coding results for the proposed DA butterfly structure show 61.02% cell area reduction comparison with those of the conventional multiplier butterfly structure. Furthermore, the 64-point Radix-4 pipeline structure using the proposed butterfly and delay commutators is compared with other conventional structures. Implementation coding results show 46.1% cell area reduction.

• Proceedings of the IEEK Conference
• /
• 2006.06a
• /
• pp.527-528
• /
• 2006

In this paper, we propose an IP design and implementation of System on a chip(SoC) for Discrete Cosine Transform (DCT) and Discrete Wavelet Transform (DWT) processor using adder-based DA(Adder-based Distributed Arithmetic). To reduced hardware cost and to improve operating speed, the combined DCT/ DWT processor used the bit-serial method and DA module. The transform of coefficient equation result in reduction in hardware cost and has a regularity in implementation. We use Verilog-HDL and Xilinx ISE for simulation and implement FPGA on SoCMaster-3.

• Proceedings of the IEEK Conference
• /
• 2002.07c
• /
• pp.1823-1826
• /
• 2002

On modern computer culture, the high quality data is required in multimedia systems. So, the technology of data compression fur data transmission is necessary now. This paper presents the pipeline architecture for the low and column address generator of 2D DCT/IDCT (Discrete Cosine Transform/Inverse Discrete Cosine Transform. In the proposed architecture, the area of hardware is reduced by using the DA (distributed arithmetic) method and applies the concepts of pipeline to the parallel architecture. As a result the designed pipeline of the low and column address generator for 2D DCT/IDCT architecture is implemented with an efficiency and high speed compared with the non-pipeline architecture.

• Proceedings of the KIEE Conference
• /
• 2007.10a
• /
• pp.371-372
• /
• 2007

In this paper we present performance and implementation comparisons of high performance two dimensional forward and inverse Discrete Cosine Transform (2D-DCT/IDCT) algorithm and low power algorithm for $8{\times}8$ 20 DCT and quantization based on partial sum and its corresponding hardware architecture for FPGA in MPEG-4. The architecture used in both low power 20 DCT and 2D IDCT is based on the conventional row-column decomposition method. The use of Fast algorithm and distributed arithmetic(DA) technique to implement the DCT/IDCT reduces the hardware complexity. The design was made using Mentor Graphics Tools for design entry and implementation. Mentor Graphics ModelSim SE6.1f was used for Verilog HDL entry, behavioral Simulation and Synthesis. The 2D DCT/IDCT consumes only 50% of the Operating Power.

• Proceedings of the IEEK Conference
• /
• 2000.07b
• /
• pp.711-714
• /
• 2000

In this paper, we propose a high-speed multiplier-free realization using ROM’s to store the results of coefficient scalings in Combination With higher signal rate and pipelined operations. We show that hardware multipliers are not needed. By varying some parameters, the proposed structure provides various combinations of hardware and clock speed (or through-put). An example is given comparing the proposed realization with the distributed arithmetic (DA) realization. Results show that With Proper Choices of the Parameters the proposed structure achieves a faster processing speed with less hardware, as compared to the DA realization.

• Proceedings of the IEEK Conference
• /
• 2007.07a
• /
• pp.85-86
• /
• 2007

In this paper, a Radix-8 structure for high-speed FFT is proposed. Even throughput of the Radix-8 FFT is twice than that of the Radix-4 FFT, implementation area of the Radix-8 is larger than that of Radix-4 FFT. But, implementation area of the proposed Radix-8 FFT was reduced by using DA(Distributed Arithmetic) for multiplication. The Verilog-HDL coding results for the proposed FFT structure show 49.2% cell area increment comparison with those of the conventional Radix-4 FFT structure. Namely, to speed up twice, 49.2% of area cost is required. In case of same throughput, power consumption of the proposed structure is reduced by 25.4%.

• Proceedings of the IEEK Conference
• /
• 2007.07a
• /
• pp.263-264
• /
• 2007

In this paper, an efficient SAD(Sum of Absolute Differences) processor structure for motion estimation of 0.264 is proposed. SAD processors are commonly used both in full search methods for motion estimation or in fast search methods for motion estimation. Proposed structure consists of SAD calculator block, combinator block, and minimum value calculator block. Especially, proposed structure is simplified by using Distributed Arithmetic for addition operation. The Verilog-HDL(Hard Description Language) coding and FPGA implementation results for the proposed structure show 39% and 32% gate count reduction comparison with those of the conventional structure, respectively. Due to its efficient processing scheme, the proposed SAD processor structure can be widely used in size dominant H.264 chip.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.16 no.12
• /
• pp.1236-1248
• /
• 1991

A new fast algorithm for computing the discrete cosine transform(DCT) Is developed decomposing N-point DCT into an N /2-point DCT and two N /4 point transforms(transpose of an N /4-point DCT. TN/t'and)It has an important characteristic that in this method, the roundoff noise power for a fixed point arithmetic can be reduced significantly with respect to the wellknown fast algorithms of Lee and Chen. since most coefficients for multiplication are distributed at the nodes close to the output and far from the input in the signal flow graph In addition, it also shows three other versions of factorization of DCT matrix with the same number of operations but with the different distributions of multiplication coefficients.