• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.5
• /
• pp.177-188
• /
• 2018

The scale of graph data has been increased rapidly because of the growth of mobile Internet applications and the proliferation of social network services. This brings upon the imminent necessity of efficient distributed and parallel graph processing approach since the size of these large-scale graphs are easily over a capacity of a single machine. Currently, there are two popular parallel graph processing approaches, vertex-centric graph processing and block centric processing. While a vertex-centric graph processing approach can easily be applied to the parallel processing system, a block-centric graph processing approach is proposed to compensate the drawbacks of the vertex-centric approach. In these systems, the initial quality of graph partition affects to the overall performance significantly. However, it is a very difficult problem to divide the graph into optimal states at the initial phase. Thus, several dynamic load balancing techniques have been studied that suggest the progressive partitioning during the graph processing time. In this paper, we present a load balancing algorithms for the block-centric graph processing approach where most of dynamic load balancing techniques are focused on vertex-centric systems. Our proposed algorithm focus on an improvement of the graph partition quality by dynamically reassigning blocks in runtime, and suggests block split strategy for escaping local optimum solution.

• KIPS Transactions on Software and Data Engineering
• /
• v.9 no.5
• /
• pp.169-176
• /
• 2020

Recently, deep neural networks (DNNs) have brought revolutions to many mobile and embedded devices by providing human-level machine intelligence for various applications. However, high inference accuracy of such DNNs comes at high computational costs, and, hence, there have been significant efforts to reduce computational overheads of DNNs either by compressing off-the-shelf models or by designing a new small footprint DNN architecture tailored to resource constrained devices. One notable recent paradigm in designing small footprint DNN models is sharing parameters in several layers. However, in previous approaches, the parameter-sharing techniques have been applied to large deep networks, such as ResNet, that are known to have high redundancy. In this paper, we propose a parameter-sharing method for already parameter-efficient small networks such as ShuffleNetV2. In our approach, small templates are combined with small layer-specific parameters to generate weights. Our experiment results on ImageNet and CIFAR100 datasets show that our approach can reduce the size of parameters by 15%-35% of ShuffleNetV2 while achieving smaller drops in accuracies compared to previous parameter-sharing and pruning approaches. We further show that the proposed approach is efficient in terms of latency and energy consumption on modern embedded devices.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.8
• /
• pp.517-526
• /
• 2013

When developing an android mobile application the androids intent is used as a mechanism to send messages between local equipment of androids application inner part and other applications. But the androids intent does not support sending messages via each android products intent. If there is a way to support each androids equipments to send messages, it will be easier to make non-stopping services. Non-stopping service is used when the user is using the android to do word or searching services and suddenly changes to a different android product but still maintains the progress what was currently being done without waiting the programs to be loaded. It is possible to send messages to each android products by using the socket, but the connection must be maintained stably which is the weak point. In this paper, I am suggesting a BRIF(Broadcasting Remote Intent Framework) framework to send messages to different android products. BRIF is a framework that uses the Googles C2DM service which services asynchronous transmissions to different android products. This is organized with the C2DM server, RemoteContext Api, web server and RISP(Remote Intent Service Provider) which is will be easy to be used for the developers since there are no big changes for coding compared to the intent code.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.24 no.6
• /
• pp.1159-1174
• /
• 2014

As the use of smartphones and tablet PCs has exploded in recent years, there are many occasions where such devices are used for treating sensitive data such as financial transactions. Naturally, many types of attacks have evolved that target these devices. An attacker can capture a password by direct observation without using any skills in cracking. This is referred to as shoulder surfing and is one of the most effective methods. There has been only a crude definition of shoulder surfing. For example, the Common Evaluation Methodology(CEM) attack potential of Common Criteria (CC), an international standard, does not quantitatively express the strength of an authentication method against shoulder surfing. In this paper, we introduce a shoulder surfing risk calculation method supplements CC. Risk is calculated first by checking vulnerability conditions one by one and the method of the CC attack potential is applied for quantitative expression. We present a case study for security-enhanced QWERTY keyboard and numeric keypad input methods, and the commercially used mobile banking applications are analyzed for shoulder surfing risks.

• Journal of KIISE:Information Networking
• /
• v.30 no.1
• /
• pp.75-81
• /
• 2003

User authentication, including confidentiality, integrity over untrusted networks, is an important part of security for systems that allow remote access. Using human-memorable Password for remote user authentication is not easy due to the low entropy of the password, which constrained by the memory of the user. This paper presents a new password authentication and key agreement protocol suitable for authenticating users and exchanging keys over an insecure channel. The new protocol resists the dictionary attack and offers perfect forward secrecy, which means that revealing the password to an attacher does not help him obtain the session keys of past sessions against future compromises. Additionally user passwords are stored in a form that is not plaintext-equivalent to the password itself, so an attacker who captures the password database cannot use it directly to compromise security and gain immediate access to the server. It does not have to resort to a PKI or trusted third party such as a key server or arbitrator So no keys and certificates stored on the users computer. Further desirable properties are to minimize setup time by keeping the number of flows and the computation time. This is very useful in application which secure password authentication is required such as home banking through web, SSL, SET, IPSEC, telnet, ftp, and user mobile situation.

• The KIPS Transactions:PartD
• /
• v.18D no.3
• /
• pp.157-168
• /
• 2011

Recently flash memory has been being utilized as a main storage device in mobile devices, and flashSSDs are getting popularity as a major storage device in laptop and desktop computers, and even in enterprise-level server machines. Unlike HDDs, on flash memory, the overwrite operation is not able to be performed unless it is preceded by the erase operation to the same block. To address this, FTL(Flash memory Translation Layer) is employed on flash memory. Even though the modified data block is overwritten to the same logical address, FTL writes the updated data block to the different physical address from the previous one, mapping the logical address to the new physical address. This enables flash memory to avoid the high block-erase cost. A flashSSD has an array of NAND flash memory packages so it can access one or more flash memory packages in parallel at once. To take advantage of the internal parallelism of flashSSDs, it is beneficial for DBMSs to request I/O operations on sequential logical addresses. However, the B-tree structure, which is a representative index scheme of current relational DBMSs, produces excessive I/O operations in random order when its node structures are updated. Therefore, the original b-tree is not favorable to SSD. In this paper, we propose AS(Always Sequential) B-tree that writes the updated node contiguously to the previously written node in the logical address for every update operation. In the experiments, AS B-tree enhanced 21% of B-tree's insertion performance.

• KIPS Transactions on Computer and Communication Systems
• /
• v.9 no.3
• /
• pp.45-58
• /
• 2020

In the current trend of increasing network traffic due to the popularization of cloud service and mobile devices, WAN bandwidth is very low compared to LAN bandwidth. In a WAN environment, a WAN optimizer is needed to overcome performance problems caused by transmission protocol, packet loss, and network bandwidth limitations. In this paper, we analyze the data deduplication algorithm of WANProxy, an open source WAN optimizer, and evaluate its performance in terms of network latency and WAN bandwidth. Also, we evaluate the performance of the two-stage compression method of WANProxy and Zstandard. We propose a new method to improve the performance of WANProxy by revising its data deduplication algorithm and evaluate its performance improvement. We perform experiments using 12 data files of Silesia with a data segment size of 2048 bytes. Experimental results show that the average compression rate by WANProxy is 150.6, and the average network latency reduction rates by WANProxy are 95.2% for a 10 Mbps WAN environment and 60.7% for a 100 Mbps WAN environment, respectively. Compared with WANProxy, the two-stage compression of WANProxy and Zstandard increases the average compression rate by 33%. However, it increases the average network latency by 2.1% for a 10 Mbps WAN environment and 5.27% for a 100 Mbps WAN environment, respectively. Compared with WANProxy, our proposed method increases the average compression rate by 34.8% and reduces the average network latency by 13.8% for a 10 Mbps WAN and 12.9% for a 100 Mbps WAN, respectively. Performance analysis results of WANProxy show that its performance improvement in terms of network latency and WAN bandwidth is excellent in a 10Mbps or less WAN environment while superior in a 100 Mbps WAN environment.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.10
• /
• pp.3210-3218
• /
• 2000

In a IS-95 cellular systm, blocking will occur when the reverse link user interference power reaches a predepermmed level which is set to maintam acceptable signal quality. In this paper, it is assumed that a mobile rdio channel is a shadowing channel and Erlang capacity is calculated for the reverse limk of an imperfect power controlled IS-95 cellular system. the blocking probability is derived using lognornal pproximation and the results according to guassian and lognormal approximation method are compared and analyzed respcctively. Assuming that blocking probability is 1% at the data rate of $R_b$=9.6kbps and $R_b$=14.4kbps, it is shown that Erlang capacity using Iognormal approximation is 13.68 Erlang and 7.08 Erlang and then the approximation erroris occurred about 24.4% and 40.4% inthe garssian approximation, respectively. It is also observed that if the power control becomes periect, the Erlang capacity is increased more 6.99 and 4.21 Erlang than that of the imperfect power control that the power contrl error is 2.5dB, and if voice activity is considered as 10%, the Erlang capacity is increased more 8.21 and 1.25 Erlang than that using non voice activity, respectively.

• The KIPS Transactions:PartD
• /
• v.11D no.6
• /
• pp.1221-1230
• /
• 2004

In mobile environments, the locations of moving objects such as vehicles, airplanes and users of wireless devices continuously change over time. For efficiently processing moving object information, the database system should be able to deal with large volume of data, and manage indexing efficiently. However, previous research on indexing method mainly focused on query performance, and did not pay attention to update operation for moving objects. In this paper, we propose a novel moving object indexing method, named ACAR-Tree. For processing efficiently frequently updating of moving object location information as well as query performance, the proposed method is based on fixed grid structure with auxiliary R-Tree. This hybrid structure is able to overcome the poor update performance of R-Tree which is caused by reorganizing of R-Tree. Also, the proposed method is able to efficiently deal with skewed-. or gaussian distribution of data using auxiliary R-Tree. The experimental results using various data size and distribution of data show that the proposed method has reduced the size of index and improve the update and query performance compared with R-Tree indexing method.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.7
• /
• pp.1804-1820
• /
• 1997

In this paper, an efficient call control procedure is presented for next generation wireless ATM networks and its performance is mathematically analyzed using the open queueing network. This procedure is based on a new scheme called as the cell clustering. When we use the cell clustering scheme, at the time that a mobile connection is admitted to the network, a virtual cell is constructed by choosing a group of neighboring base stations to which the call may probabilistically hand over and by assigning to the call a collection of virtual paths between the base stations. Within a microcell/picocell environment, it is seen that the cell clustering can support effectively a very high rate of handovers, provides very high system capacity, and guarantees a high degree of frequency reuse over the same geographical region without requiring the intervention of the network call control processor each time a handover occurs. But since mobiles, once admitted, are free to roam within the virtual cell, overload condition occurs in which the number of calls to be handled by one base station to exceed that cell site's capacity of radio channel. When an overload condition happens, the quality of service is abruptly degraded. We refer to this as the overload state and in order to quantify the degree of degradation we define two metrics, the probability of overload and the normalized average time spent in the overload state. By using the open network queueing model, we derive closed form expressions for the maximum number of calls that can be admitted into the virtual cell such that the two defined metrics are used as the acceptance criteria for call admission control.