• ETRI Journal
• /
• v.28 no.4
• /
• pp.453-460
• /
• 2006

In this article, we introduce a new class of product codes based on convolutional codes, called convolutional product codes. The structure of product codes enables parallel decoding, which can significantly increase decoder speed in practice. The use of convolutional codes in a product code setting makes it possible to use the vast knowledge base for convolutional codes as well as their flexibility in fast parallel decoders. Just as in turbo codes, interleaving turns out to be critical for the performance of convolutional product codes. The practical decoding advantages over serially-concatenated convolutional codes are emphasized.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.29 no.5C
• /
• pp.611-617
• /
• 2004

The optimum convolutional error correction codes for recently standardized Feher-patented quadrature phase-shift keying (FQPSK-B) modulation are proposed. We utilize the continuous phase modulation characteristics of FQPSK-B signals for calculating the minimum Euclidean distance of convolutional coded FQPSK-B signal. It is shown that the Euclidean distance between two FQPSK-B signals is proportional to the Hamming distance between two binary data sequence. Utilizing this characteristic, we show that the convolutional codes with optimum free Hamming distance is the optimum convolutional codes for FQPSK-B signals.

• Journal of Information Processing Systems
• /
• v.10 no.1
• /
• pp.69-80
• /
• 2014

The free distance of (n, k, l) convolutional codes has some connection with the memory length, which depends on not only l but also on k. To efficiently obtain a large memory length, we have constructed a new class of (2k, k, l) convolutional codes by (2k, k) block codes and (2, 1, l) convolutional codes, and its encoder and generation function are also given in this paper. With the help of some matrix modules, we designed a single structure Viterbi decoder with a parallel capability, obtained a unified and efficient decoding model for (2k, k, l) convolutional codes, and then give a description of the decoding process in detail. By observing the survivor path memory in a matrix viewer, and testing the role of the max module, we implemented a simulation with (2k, k, l) convolutional codes. The results show that many of them are better than conventional (2, 1, l) convolutional codes.

• Journal of IKEEE
• /
• v.20 no.3
• /
• pp.303-306
• /
• 2016

In this paper, we propose a method to accelerate convolutional neural network by utilizing a GPGPU. Convolutional neural network is a sort of the neural network learning features of images. Convolutional neural network is suitable for the image processing required to learn a lot of data such as images. The convolutional layer of the conventional CNN required a large number of multiplications and it is difficult to operate in the real-time on the embedded environment. In this paper, we reduce the number of multiplications through Winograd convolution operation and perform parallel processing of the convolution by utilizing SIMT-based GPGPU. The experiment was conducted using ModelSim and TestDrive, and the experimental results showed that the processing time was improved by about 17%, compared to the conventional convolution.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2017.05a
• /
• pp.335-338
• /
• 2017

Currently Artificial Inteligence and Deep Learning are social issues, and These technologies are applied to various fields. A good method among the various algorithms in Artificial Inteligence is Convolutional Neural Network. Convolutional Neural Network is a form that adds convolution layers that extracts features by convolution operation on a general neural network method. If you use Convolutional Neural Network as small amount of data, or if the structure of layers is not complicated, you don't have to pay attention to speed. But the learning time is long as the size of the learning data is large and the structure of layers is complicated. So, GPU-based parallel processing is a lot. In this paper, we developed Convolutional Neural Network using CUDA and Learning speed is faster and more efficient than the method using the CPU.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.26 no.8A
• /
• pp.1437-1440
• /
• 2001

We present a modification method for runlength limited codes based on convolutional codes. This method is based on cosets of convolutional codes and can be applied to any convolutional code without degradation of error control performance of the codes. The upper bound of maximum zero and/or one runlength are provided. Some convolutional codes which have the shortest maximum runlength for given coding parameters are tabulated.

• Journal of Communications and Networks
• /
• v.17 no.1
• /
• pp.12-20
• /
• 2015

Convolutional codes which are terminated by direct truncation (DT) and zero tail termination provide unequal error protection. When DT terminated convolutional codes are used to encode short messages, they have interesting error protection properties. Such codes match the significance of the output bits of common quantizers and therefore lead to a low mean square error (MSE) when they are used to encode quantizer outputs which are transmitted via a noisy digital communication system. A code construction method that allows adapting the code to the channel is introduced, which is based on time-varying convolutional codes. We can show by simulations that DT terminated convolutional codes lead to a lower MSE than standard block codes for all channel conditions. Furthermore, we develop an MSE approximation which is based on an upper bound on the error probability per information bit. By means of this MSE approximation, we compare the convolutional codes to linear unequal error protection code construction methods from the literature for code dimensions which are relevant in analog to digital conversion systems. In numerous situations, the DT terminated convolutional codes have the lowest MSE among all codes.

• Journal of Communications and Networks
• /
• v.16 no.3
• /
• pp.270-279
• /
• 2014

The Viterbi decoding algorithm, which provides maximum - likelihood decoding, is currently considered the most widely used technique for the decoding of codes having a state description, including the class of linear error-correcting convolutional codes. Two classes of nonbinary convolutional codes are presented. Distance preserving mapping convolutional codes and M-ary convolutional codes are designed, respectively, from the distance-preserving mappings technique and the implementation of the conventional convolutional codes in Galois fields of order higher than two. We also investigated the performance of these codes when combined with a multiple frequency-shift keying (M-FSK) modulation scheme to correct narrowband interference (NBI) in power-line communications channel. Themodification of certain detectors of the M-FSK demodulator to refine the selection and the detection at the decoder is also presented. M-FSK detectors used in our simulations are discussed, and their chosen values are justified. Interesting and promising obtained results have shown a very strong link between the designed codes and the selected detector for M-FSK modulation. An important improvement in gain for certain values of the modified detectors was also observed. The paper also shows that the newly designed codes outperform the conventional convolutional codes in a NBI environment.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.21 no.6
• /
• pp.1143-1148
• /
• 2017

Currently, Artificial Intelligence and Deep Learning are rising as hot social issues, and these technologies are applied to various fields. A good method among the various algorithms in Artificial Intelligence is Convolutional Neural Networks. Convolutional Neural Network is a form that adds Convolution Layers to Multi Layer Neural Network. If you use Convolutional Neural Networks for small amount of data, or if the structure of layers is not complicated, you don't have to pay attention to speed. But the learning should take long time when the size of the learning data is large and the structure of layers is complicated. In these cases, GPU-based parallel processing is frequently needed. In this paper, we developed Convolutional Neural Networks using CUDA, and show that its learning is faster and more efficient than learning using some other frameworks or programs.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.26 no.1
• /
• pp.197-208
• /
• 2016

In this paper, we propose a method based on a convolutional neural network which is one of the deep neural network. So, we convert a malware code to malware image and train the convolutional neural network. In experiment with classify 9-families, the proposed method records a 96.2%, 98.7% of top-1, 2 error rate. And our model can classify 27 families with 82.9%, 89% of top-1,2 error rate.