Background:The incidence of infectious diarrheal disease in Korea has decreased over the past decade, but traveler's diarrhea (TD) is increasing in frequency. We therefore investigated the distribution of the causative agents of TD. Methods: A total of 132 rectal swab specimens were acquired from TD patients who entered the country via Gimhae International Airport. The specimens were screened for 12 bacterial pathogens by real-time PCR, and target pathogens were isolated from the PCR positive specimens using conventional microbiological isolation methods. Results: A total of 93 specimens (70.5%) showed positive PCR screening results, and of these specimens, nine species and 50 isolates (37.9%), including Vibrio parahaemolyticus (18 isolates) and ETEC (17 isolates), were isolated. No specimens were PCR

One of the key challenges to develop applications for multicorebased embedded systems is enduring software complexity and performance requirements raised by parallel programming. Adaptive, self-aware or deterministic computing has been proposed as one method to help application developer cope with these problems. To provide these advanced features, the developer should monitor their application to know the current state of the application for optimizing or adapting to meet their goals. One solution is to make thread progress index of a program for tracking their current state. To guarantee accuracy of the tracking method, we find that the process of a thread should be repeatable and accurate. If the progress does not have the properties, it is impossible to guarantee the accuracy and determinism of the method for tracking progress of thread. Additionally, we minimize the overhead to measure and manage the progress even though providing the cycle-accurate progress tracking with inevitable runtime overhead. In this paper, we propose an efficient method to monitoring thread progress with a cycle-accurate deterministic progress counter and an efficient management technique for adaptive scheduling algorithm. As a case study, we implement 24 video decoders on a quad core system using the proposed runtime system and show the experimental result which yields a geometric mean overhead of about 2.1% to monitor each thread and these applications yield a mean performance improvement of 6% relative to normal execution when running on 4 cores.

With the development of wireless communication and sensor technology, sensor network applications in various fields have been applied. To minimize the power consumption of sensors in sensor network is one of the important factors in oder to extend the system life. The power consumption of each sensor within sensor network can be different depending on the communication method between head(sink) node and its node. In this paper, we propose a new hierarchical communication method to minimize the power consumption of each sensor. The proposed method divides the area of sensor network into four areas using divide-and-conquer method and selects the nearest node to head node in each area as a child node of the node. Next the hierarchical tree in the same way is constructed recursively until each area is no longer divided. Each sensor can communicate to head node using this hierarchical tree. The proposed results were compared with the previous methods through simulation, and showed excellent results in the energy efficiency of sensor network.

No abstract

One of the key challenges in developing applications for multicore systems is dealing with the nondeterministic behavior of applications. This makes parallel programming hard and programmers should spend substantial amounts of time to debug their application. To remove the nondeterministic behavior of applications, we could make the behavior of applications deterministic through the control of thread's execution order based on their progress. To guarantee the deterministic behavior of a thread, the measured progress of a thread should be repeatable and cycle-accurate. If the progress does not meet these requirements, it is impossible to guarantee the deterministic behavior of threads. In this paper, we propose a novel method to measure a cycle-accurate deterministic progress of a thread and a technique to guarantee the deterministic behavior of applications with the proposed method. As a case study, we implement 24 video decoders on a quad core system using the proposed technique and evaluate our technique on SPLASH-2 benchmark. The experimental result yields a geometric mean of overhead of about 7.2% on SPLASH-2 benchmark.