Byoungchul Ahn

A reverse AODV routing protocol in ad hoc mobile networks

In mobile ad hoc networks, mobile devices wander autonomously for the use of wireless links and dynamically varying network topology. AODV (Ad-hoc on-demand Distance vector routing) is a representative among the most widely studied on-demand ad hoc routing protocols. Previous protocols have shown some shortcomings on performance. AODV and most of the on-demand ad hoc routing protocols use single route reply along reverse path. Rapid change of topology causes that the route reply could not arrive to the source node, i.e. after a source node sends several route request messages, the node obtains a reply message, especially on high speed mobility. This increases both in communication delay and power consumption as well as decrease in packet delivery ratio. To avoid these problems, we propose a reverse AODV which tries multiple route replies. The extended AODV is called reverse AODV (R-AODV), which has a novel aspect compared to other on-demand routing protocols on Ad-hoc Networks: it reduces path fail correction messages and obtains better performance than the AODV and other protocols have. We design the R-AODV protocol and implement simulation models using NS-2. Simulation results show that the reverse AODV provides good experimental results on packet delivery ratio, power consumption and communication delay.

Fault Modeling and Multi-Tone Dither Scheme for Testing 3D TSV Defects

This paper provides a new test technique for detecting defects in Through Silicon Via (TSV) in 3-D ICs and presents a substrate-dependent equivalent electrical model for TSVs. Process-related defects that affect the functional electrical performance of the TSV are identified, and fault models are developed for each individual defect. The fault models are integrated into the equivalent electrical model of the TSV for testing. Our test technique uses an RF carrier signal modulated with a multi-tone signal with added Gaussian white noise to synthesize the test stimulus; the peak-to-average ratio is measured as output response. We find a significant difference in peak-to-average ratio between defect-free and defective TSVs. Our test technique is very sensitive to small defects in these nanostructures, thereby identifying the defects with high accuracy.

A prefetching Web caching method using adaptive search patterns

As the Internet continues to grow exponentially, the Web caching technology becomes one of very important technologies in computer networks. The mechanism of a web caching affects the performance of Internet servers or proxy servers. This paper presents a new approach to predict Web documents by tracking search patterns of users and by managing documents depending upon the number of hits. Several hot spot documents and their linked documents are stored in cache servers and transmitted to clients for fast response. Generally cache servers store not only the first document of a site but also all documents navigated through linked documents. The adaptive prefetching method, using search patterns of users, analyzes documents along with the navigation patterns and marks several popular documents. If one of these marked documents is hit, all marked documents are loaded into cache at a time. Cache servers can save their cache memory space as well as provide fast response to clients. The results show that the response time of the proposed method decreases to 20% compared with other methods.

A Fault-Tolerant Protocol for Mobile Agent

Mobile agent technology has been proposed for a variety of applications. Fault-tolerance is fundamental to the further development of mobile agent applications. Fault-tolerance prevents a partial or complete loss of the agent. Simple approaches such as checkpointing are prone to blocking, replications scheme is expensive since it has to maintain multiple replicas. In this paper, a new approach rooted from checkpointing is proposed. This scheme can guarantee to detect and recover most failure scenarios in mobile agent systems, even if there occurs machine failure.

An Energy-Efficient Data Collection Method for Wireless Multimedia Sensor Networks:

In wireless sensor networks (WSNs), some sensor nodes transmit their acquired data to a far distant base station directly or by linking several relay nodes. WSNs might not be able to operate if a sensor node completely consumes its battery power and does not perform its designated operation. Semiconductor technology drives sensor networks to collect video and audio data as well as environmental data. Therefore, energy is one of the most important resources in wireless multimedia sensor networks (WMSNs), and clustering is proposed to reduce energy consumption. When a mobile sink is used, a lot of energy is saved, and the lifetime of the network can be extended because the sensor nodes do not need to transmit multimedia data to a far distant base station. We propose an energy-efficient data collection method to extend the lifetime of networks that use a mobile sink. By using the neighborhood density clustering method and defining an optimal path for the mobile sink, the proposed method extends the life of the WMSN. Simulation results show that the proposed method increases the lifetime of WMSNs by up to 15 rounds, compared with LEACH. And data collection time is decreased owing to the predefined path for the mobile sink.

MapReduce functions to analyze sentiment information from social big data

Opinion mining, which extracts meaningful opinion information from large amounts of social multimedia data, has recently arisen as a research area. In particular, opinion mining has been used to understand the true meaning and intent of social networking site users. It requires efficient techniques to collect a large amount of social multimedia data and extract meaningful information from them. Therefore, in this paper, we propose a method to extract sentiment information from various types of unstructured social media text data from social networks by using a parallel Hadoop Distributed File System (HDFS) to save social multimedia data and using MapReduce functions for sentiment analysis. The proposed method has stably performed data gathering and data loading and maintained stable load balancing of memory and CPU resources during data processing by the HDFS system. The proposed MapReduce functions have effectively performed sentiment analysis in the experiments. Finally, the sentiment analysis results of the proposed system are very close to those of manual processes.

An Efficient Power-Aware Scheduling Algorithm in Real Time System

Personal portable devices such as DMB (digital multimedia broadcasting)s, PMP (portable multimedia player)s and PDA (portable digital assistant)s require long battery life without reducing their performance. Power consumption is one of important factors in ubiquitous environments. To reduce the dynamic power consumption, many researches have been conducted on the dynamic voltage and frequency scaling (DVFS). DVFS can reduce the power consumption proportionally to the square of the supply voltage. But processor operation under the critical speed increases the leakage current. Larger leakage current increases the static power consumption. This paper proposes a new power-aware scheduling algorithm that reduces the static and dynamic power consumption to use frequency scaling between two discrete voltage levels and under the critical speed while meeting the deadline of given applications. The simulation and experimental implementation show that the proposed scheduling algorithm results in up to 25% energy gains over traditional DVFS algorithms with two task scheduling policies.

An Adaptive Prefetching Method for Web Caches

This paper presents a new approach to predict web documents by tracking search patterns of users and by managing documents depending upon the number of hits. Several hot spot documents and their linked documents are stored in cache servers and transmitted them to clients for fast response. The adaptive prefetching method, using search patterns of users, analyzes documents along with the navigation patterns and marks several popular documents. If one of these marked documents is hit, all marked documents are loaded into the cache but only the requested document is transmitted to clients. Cache servers can save their cache memory space as well as provide fast response to clients. The results show that the average response time of the proposed method decreases to 20% compared with other methods and the cache hit rate is increased to 18% of other methods.

A Design and Implementation of Low Power Management for Personal Multimedia Systems

To reduce the power consumption of portable devices, many researches have been conducted on the dynamic voltage scaling(DVS) and dynamic power management(DPM). Embedded systems such as DMBs, PMPs and PDAs require long battery life without reducing their performance. Also, power consumption is the most important factor in ubiquitous environment. Most researches are focused on power saving methods by hardware circuits. One of them is DVS, which has proven as a highly effective technique to reduce dynamic power consumption. But, software design and control should be considered to minimize the total energy consumption. This paper presents a power management design and implementation for embedded multimedia systems by guaranteeing QoS. The implemented results show that the system saves the total power consumption up to 5% compared with the DVS on the experiment board.

Robust algorithm for scene analysis on compressed video

This paper presents a robust method to detect scene changes from compressed video streams. Scene changes are detected with DC components of DCT coefficients in MPEG encoded video sequences. Instead of decoding full frames, partial macroblocks of each frame, horizontal and vertical macroblocks, are decoded to perform scene changes. This method detects abrupt and gradual scene changes by decoding minimal blocks and saves a lot of computations. The experiment results show at least 95% detection performance for dramas, news and animations.