• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.14 no.6
• /
• pp.295-301
• /
• 2014

In the absence of a sorting algorithm faster than O(n log n), Quicksort remains the best and fastest of its kind in practice. For given n data, Quicksort records running in O(n log n) at best and $O(n^2)$ at its worst. In this paper, I propose an algorithm by which 3-points average P=(L+M+H)/3 is set as a pivot for first array L=a[s], last array H=a[e], and middle array $M=a[{\lfloor}(s+e)/2{\rfloor}]$ in order to find the more fast than Quicksort. Test results prove that the proposed 3-points average pivot Quicksort has the time complexity of O(n log n) at its best, average, and worst cases. And the proposed algorithm can be reduce the $O(n^2)$ time of Quicksort to O(n log n).

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.17 no.5
• /
• pp.157-163
• /
• 2017

The binary and interpolation search algorithms are the most famous among search area algorithms, the former running in $O(log_2n)$ on average, and the latter in $O(log_2log_2n)$ on average and O(n) at worst. Also, the interpolation search use only the probability of key value location without priori information. This paper proposes another search algorithm, which I term a 'hybrid block and interpolation search'. This algorithm employs the block search, a method by which MSB index of a data is determined as a block, and the interpolation search to find the exact location of the key. The proposed algorithm reduces the search range with priori information and search the reduced range with uninformed situation. Experimental results show that the algorithm has a time complexity of $O(log_2log_2n_i)$, $n_i{\simeq}0.1n$ both on average and at worst through utilization of previously acquired information on the block search. The proposed algorithm has proved to be approximately 10 times faster than the interpolation search on average.

• Tribology and Lubricants
• /
• v.36 no.4
• /
• pp.232-243
• /
• 2020

It is essential to predict the amount of wear and surface parameters for a surface where relative motion occurs. In the asperity-based model for wear prediction, only the average contact pressure can be obtained. Hence, the accuracy of wear analysis is poor. In this study, DC-FFT is used to obtain the pressure of each node, and wear analysis is performed by considering the effect of the pressure gradient. The numerical surface generation method is used to create Gaussian, negatively skewed, and positively skewed surfaces for wear analysis. The spatial and height distributions of each surface are analyzed to confirm the effectiveness of the generated surface. Furthermore, wear analysis is performed using DC-FFT and Archard's wear formula. After analysis, it is confirmed that all peaks are removed and only valleys remain on the surface. The RMS roughness and Sk continue to decrease and Ku increases as the cycle progresses. It is observed that the surface parameters are significantly affected by the radius of curvature of the asperity. This analysis method is more accurate than the existing average wear and truncation models because the change in asperity shape during the wear process is reflected in detail.

• Proceedings of the Korean Society for Bioinformatics Conference
• /
• 2004.11a
• /
• pp.244-249
• /
• 2004

There have been many attempts to predict the secondary structure content of a protein from its primary sequence, which serves as the first step in a series of bioinformatics processes to gain knowledge of the structure and function of a protein. Most of them assumed that prediction relying on the information of the amino acid composition of a protein can be successful. Several approaches expanded the amount of information by including the pair amino acid composition of two adjacent residues. Recent methods achieved a remarkable improvement in prediction accuracy by using this expanded composition information. The overall average errors of two successful methods were 6.1% and 3.4%. This work was motivated by the observation that evolutionarily related proteins share the similar structure. After manipulating the values of the frequency matrix obtained by running PSI-BLAST, inputs of an artificial neural network were constructed by taking the ratio of the amino acid composition of the evolutionarily related proteins with a query protein to the background probability. Although we did not utilize the expanded composition information of amino acid pairs, we obtained the comparable accuracy, with the overall average error being 3.6%.

• Smart Structures and Systems
• /
• v.14 no.4
• /
• pp.617-641
• /
• 2014

The energy constraint is still a common issue for the practical application of wireless sensors, since they are usually powered by batteries which limit their lifetime. In this paper, a practical compound energy efficiency strategy is proposed and realized in the implementation of a real time wireless sensor platform. The platform is intended for wireless structural monitoring applications and consists of three parts, wireless sensing unit, base station and data acquisition and configuration software running in a computer within the Matlab environment. The high energy efficiency of the wireless sensor platform is achieved by a proposed adaptive radio transmission power control algorithm, and some straightforward methods, including adopting low power ICs and high efficient power management circuits, low duty cycle radio polling and switching off radio between two adjacent data packets' transmission. The adaptive transmission power control algorithm is based on the statistical average of the path loss estimations using a moving average filter. The algorithm is implemented in the wireless node and relies on the received signal strength feedback piggybacked in the ACK packet from the base station node to estimate the path loss. Therefore, it does not need any control packet overheads. Several experiments are carried out to investigate the link quality of radio channels, validate and evaluate the proposed adaptive transmission power control algorithm, including static and dynamic experiments.

• Proceedings of the KIEE Conference
• /
• 2008.07a
• /
• pp.1697-1698
• /
• 2008

PLS(Pohang Light Source) is 2.5 Gev synchrotron radiation source in Pohang, Korea, which is under operation since 1995. The hardware and software of the old BPM(Beam Position Monitor) data acquisition system for the PLS storage ring was completely upgraded to increase its performance and stability. The new BPM data acquisition system is based on VME-based EPICS (Experimental Physics and Instrument Control System) IOC system. We used 16-bit resolution analog-to-digital conversion board to digitize analog BPM signals. We developed a data average software to average raw BPM data using reflective memory board. We also developed device drivers for VME I/O boards used, IOC database for PV's(Process Variables). The new BPM data acquisition system is currently running for routine operation with good performance and stability. In this paper, we present the hardware and software of the new BPM data acquisition system DTL water skid cooling system and Resonant Control Cooling.

• Industrial Engineering and Management Systems
• /
• v.12 no.3
• /
• pp.172-180
• /
• 2013

We herein consider a stochastic multi-item inventory management problem in which a warehouse sells multiple items with stochastic demand and periodic replenishment from a supplier. Inventory management requires the timing and amounts of orders to be determined. For inventory replenishment, trucks of finite capacity are available. Most inventory management models consider either a single item or assume that multiple items are ordered independently, and whether there is sufficient space in trucks. The order cost is commonly calculated based on the number of carriers and the usage fees of carriers. In this situation, we can reduce future shipments by supplementing items to an order, even if the item is not scheduled to be ordered. On the other hand, we can reduce the average number of items in storage by reducing the order volume and at the risk of running out of stock. The primary variables of interest in the present research are the average number of items in storage, the stock-out volume, and the number of carriers used. We formulate this problem as a multi-objective optimization problem. In a numerical experiment based on actual shipment data, we consider the item shipping characteristics and simulate the warehouse replenishing items coordinately. The results of the simulation indicate that applying a conventional ordering policy individually will not provide effective inventory management.

• Journal of Computing Science and Engineering
• /
• v.7 no.1
• /
• pp.53-66
• /
• 2013

Bounding the worst-case DRAM performance for a real-time application is a challenging problem that is critical for computing worst-case execution time (WCET), especially for multicore processors, where the DRAM memory is usually shared by all of the cores. Typically, DRAM commands from consecutive DRAM accesses can be pipelined on DRAM devices according to the spatial locality of the data fetched by them. By considering the effect of DRAM command pipelining, we propose a basic approach to bounding the worst-case DRAM performance. An enhanced approach is proposed to reduce the overestimation from the invalid DRAM access sequences by checking the timing order of the co-running applications on a dual-core processor. Compared with the conservative approach, which assumes that no DRAM command pipelining exists, our experimental results show that the basic approach can bound the WCET more tightly, by 15.73% on average. The experimental results also indicate that the enhanced approach can further improve the tightness of WCET by 4.23% on average as compared to the basic approach.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2006.05a
• /
• pp.277-282
• /
• 2006

Road traffic noise is not produced by any one factor rather occurs as a composition of various factors. Its occurrence is made by running engine noise, tire frictional, and exhaust noise etc. The quality of the noise depends on the size of the vehicles, rotation and engine speed, vehicle load, package state of the road and incline etc. The occurrence of noise level of heavy trucks appears louder than smaller vehicles and the noise levels produced differs according to speed and load etc between similar size vehicles. Other factors such as traffic density, average speed, mixing rate of heavy vehicles, and the distance between vehicles also generate road traffic noise. In this paper we examine 2, 4, and 6-lane roads in Jeonju. Consequently, this study examined the means used to measure road traffic noise. It was found that when there is a large traffic density and the average velocity is below 70 km/hr, the noise level could receive a relative proper value by the current measuring means. But in the case of night-time, it was found that the current measuring method is inapposite.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.10
• /
• pp.4080-4097
• /
• 2020

In recent years, convolutional neural network (CNN) has become the primary method for face detection. But its shortcomings are obvious, such as expensive calculation, heavy model, etc. This makes CNN difficult to use on the mobile devices which have limited computing and storage capabilities. Therefore, the design of lightweight CNN for face detection is becoming more and more important with the popularity of smartphones and mobile Internet. Based on the CPU real-time face detector FaceBoxes, we propose a multi-task lightweight face detector, which has low computing cost and higher detection precision. First, to improve the detection capability, the squeeze and excitation modules are used to extract attention between channels. Then, the textual and semantic information are extracted by shallow networks and deep networks respectively to get rich features. Finally, the landmark detection module is used to improve the detection performance for small faces and provide landmark data for face alignment. Experiments on AFW, FDDB, PASCAL, and WIDER FACE datasets show that our algorithm has achieved significant improvement in the mean average precision. Especially, on the WIDER FACE hard validation set, our algorithm outperforms the mean average precision of FaceBoxes by 7.2%. For VGA-resolution images, the running speed of our algorithm can reach 23FPS on a CPU device.