Joonhyouk Jang

A study on the effect of heat treatment on functional groups of pitch based activated carbon fiber using FTIR

Abstract Pitch based active carbon fibers (ACFs) were analyzed by a FTIR micro-ATR technique by introducing a very thin KBr layer on their surfaces. The ACFs were thermally treated at 600, 1100, and 1200 °C respectively, to investigate the change of their surface functionalities. As the heat treatment temperature increased, the amount of oxygen containing surface functional groups were reduced and the ACFs became more hydrophobic. When the ACFs were thermally treated, the decrease of carboxylic acid groups occurred first and ketone or quinone groups subsequently disappeared at higher heat treatment temperatures. The degree of graphitization of the ACF was increased significantly when it was treated at 1100 and 1200 °C and this was partially attributed to the release of the CO groups in the conjugated ketone or quinone structures.

Protecting Android applications with steganography-based software watermarking

In this paper, we propose a steganography-based software watermarking scheme to protect Android applications from software piracy. Considering the resource limitations in mobile devices, it is difficult to apply previous software watermarking schemes in mobile applications. To decrease the watermarking overhead in a watermarking scheme, such as a slowdown in the execution of applications after embedding a watermark, we proposed a different approach that applies steganography-based techniques in software watermarking. The proposed scheme embeds watermarks by reordering the sequence of instructions in the basic blocks in Dalvik executable files.

An efficient backup-recovery technique to process large data in distributed key-value store

As the demand for the high-speed processing of mass data increases, the use of in-memory systems is increasing. In-memory systems enable data to be accessed in memory, thereby allowing for its high-speed processing. However, as the size and volume of the data increases, its recovery speed from a replicated server gets slower. In this paper, we propose a backup-recovery technique allowing quick recovery by distributing and replicating data onto various slave servers and recovering them data from the simultaneously during disaster recovery. We realized the proposed technique using Redis, an open source distributed memory system. The results showed that the equal key distribution and recovery performance with the proposed technique was improved by 51.4%.

Integrated financial trading system based on distributed in-memory database

The portion of HFT in market is growing rapidly. To process hundreds of thousands of orders per second which are requested by algorithms trading, not by humans, trading systems are required to equip high performance computer, network devices, and efficient software solutions. To achieve performance requirements, financial trading software processes data in memory with delayed writes to databases. However, these systems are getting closer to their physical limits and in most cases fault tolerance is ignored to improve the performance of the system. In this paper, we designed an integrated financial trading system which runs on a distributed manner based on in-memory NoSQL database and InfiniBand technologies.

Concurrency control scheme for key-value stores based on InfiniBand

Using InfiniBand technologies, the performance of key-value stores can be greatly improved because of RDMA features and the ultra-low latency of InfiniBand. However, maximizing the benefits of InfiniBand for key-value stores is still challenging because of the data consistency problem between RDMAs and CPU-aware memory accesses. In this paper, we propose a concurrency control scheme to utilize the RDMA features of InfiniBand in key-value stores. The proposed scheme efficiently handles the race conditions among GET and PUT operations on the key-value store.

Enhancing lookup performance of key-value stores using cuckoo hashing

Recently, many studies have been conducted on key-value store as a platform for distributed storage services. A key-value store manages data as key-value pairs using hash tables. In this paper, we propose a hybrid hash table management technique to improve the performance of single nodes in a key-value store. The proposed scheme uses cuckoo hashing and separate chaining at the same time. Using this hybrid hashing scheme, cuckoo hashing and separate chaining offset the respective disadvantages while preserving the advantages of both. Experiments show that the proposed scheme is much faster than the hash table with separate chaining for temporal locality workloads.

A method for resilient graph-based comparison of executable objects

A software birthmark is a set of characteristics extracted from an executable program. Software birthmark techniques are used to detect program theft by determining the similarity between two different programs. In this paper, we propose a software birthmark based on a graph-based structure comparison scheme. By enhancing the function level comparison and the basic block level comparison algorithms, we show that software birthmarks can have improved resilience compared to those of the original graph-based structure comparison scheme.

An internet protocol stack for high-speed transmission in a non-OS environment

Todays embedded systems are required to have stronger wired/wireless communication capability. Due to strict limitations on how they utilize resources such as power, address space, and processing ability, network protocols for embedded systems are designed to make the best use of constrained resources. They are developed in a way that allows for support from a range of operating systems and thus utilizes platform-independent architectures. This paper presents an Internet protocol stack design for embedded systems that operate without an operating system. Our scheme schedules the transmission and the reception of the data packets and uses cross-layer optimization. We implemented the proposed scheme in a LTE network device. The overhead of our scheme is low and it subsequently meets the constraint of transmission speed in the next generation mobile communication environments.

Analysis of expected processing time for multiple memory backup servers

Data stores in distributed systems maintain backup servers for data recovery. Most existing data stores use non-volatile storage to store backup data. However, high-performance systems may require the use of the main memory for backup storage instead of disks due to the performance gap between the disks and the memory. In this paper, we present an expected processing time analysis for data stores in distributed systems using multiple memory backup servers for recovery. Our analysis shows that the expected processing time decreases as the number of backup servers increases. Thus the number of backup servers can be chosen to meet the performance requirement based on our analysis.

Impact of software watermarking on smart devices

As a binary protection technique, software watermarks can be used to protect the copyrights of applications. However, most watermarking techniques do not consider the resource limitations or the interactivity of smart applications. In this paper, we apply an efficient software watermarking scheme to an actual Android-based smart device and analyze the impact of software watermarks. Experimental results show that watermarked applications perform efficiently in terms of the embedding time and the execution overhead.