Jaehyun Park

An extended TCP/IP protocol for real-time local area networks

Abstract This paper proposes an extended TCP/IP protocol, LAN-TCP, that can be used for distributed soft real-time systems connected through an Ethernet-based local area network. Since LAN-TCP shows soft real-time performance, while maintaining compatibility and interoperability with the standard TCP/IP, the existing application software can be used without any modification. LAN-TCP also provides a periodic transmission mode (PTM), with which periodic data collection and the updating of the control signals can be efficiently implemented with relatively small traffic overhead. This paper includes a computer simulation and experimental results for the proposed protocol.

Hardware implementation of real-time Petri-net-based controllers

Abstract This paper presents an implementation method for high-speed Petri-net-based controllers for time-critical control with parallelism, based on look-up tables. Unlike microprocessor-based software implementations, the method is based on hardware, and offers enough speed to control fast plants at low cost. The method is easily able to accommodate more complex control problems than hardwired implementations. The data structure and execution module are composed of common memory devices and simple logic gates. A modified matrix-based data structure, named the bi-column matrix-based data structure, is suggested. It has a regular form but retains the efficiency of list-based data structures. A sub-class of Petri nets is defined in order to fit Petri nets to the bi-column matrix-based data structure, and conversion to Petri nets of the sub-class is described. The matrix framework enables pipelined execution, which increases the performance in an efficient way. An event-driven evolution scheme is proposed as well, to enhance response time. This method is demonstrated by an implementation using a Xilinx 4000 series LCA. An example illustrates that the Petri-net-based controller is useful in a high-speed supervisor-control problem.

Architectural design of an RISC processor for programmable logic controllers

In this paper, an architecture of the RISC processor for programmable logic controllers is proposed. Execution characteristics of relay ladder logic (RLL) are analyzed with various application programs in order to determine an optimal architecture for programmable logic controllers (PLCs). A conditional execution mechanism is developed to prevent pipeline hazards caused by the inherent execution behavior of RLL. Instruction sets of three different architectural models are defined. Translators, assemblers, and simulators are developed for three models to evaluate performance and to choose an optimal architecture for PLCs. The proposed processor, which has an accumulator architecture with a four-stage pipeline, exhibits desirable performance much higher than that of recent commercial PLCs.

A content-based music retrieval system using representative melody index from music databases

For content-based music retrieval, since not only the correctness of retrieval results but also the performance of retrievals is important, there are great needs for efficient content-based music retrieval systems that can quickly retrieve the relevant music on demand from large music database with low storage overhead. In this paper, we describe the design and implementation of a content-based music retrieval system in which the representative melody index is systemically constructed and used to support quick and appropriate retrievals to users melody queries. By using the proposed system for digital music libraries, it could save up to 65% of index space than that for the whole motifs index while the appropriateness of retrieval results is maintained.

Implementation of High-Reliable MVB Network for Safety System of Nuclear Power Plant

The computer network plays an important role in modern digital controllers within a safety system of a nuclear power plant. For the reliable and realtime data communication between controllers, this paper proposes a modified high-reliable MVB(multi-function vehicle bus) as a main control network for a safety system of a nuclear power plant. The proposed network supports the state-based communication in order to ensure the deterministic communication latency, and very fast network recovery when the bus master fails compare to the standard MVB. This paper also shows the implementation results using a FPGA-based testbed.

TCP/IP PROTOCOL OVER IEEE-1394 NETWORK FOR REAL-TIME CONTROL APPLICATIONS

Abstract A computer network between the control devices plays an important role to provide the real-time operation in distributed computer control systems. Compare to the field-buses used in past, TCP/IP protocol, widely used in IT(Information Technology) area, provides an economical approach to implement a real-time control network. This paper proposes a modified TCP/IP protocol for IEEE-1394 network that supports asynchronous as well as isochronous data transmission. The experiment results show that real-time data could meet their deadlines even normal non-real-time data are mixed with them. The proposed TCP/IP over IEEE-1394 bus can be adopted for the real-time control applications including the home networking, intelligent robot, and general distributed control systems.

MILESTONE REPORT COORDINATING COMMITTEE ON COMPUTER CONTROL

Faculty of Information Technology and systems, Delft University of Technology, Delft, the Netherlands *Department of Automation Engineering, Inha University, Inchon, Korea **Faculty of Electrical Engineering, Fern Universitat, Hagen, Germany ***Intelligent Control and Systems Group, School of Electrical and Electronic Engineering, Queen’s University, Belfast, UK ****Department of Electrical and Computing Engineering, University of Central Florida, Orlando, USA

A rapid QAR data acquisition and visualization system for aircraft engine maintenances

We developed an aircraft engine maintenance support system that can be used for quick delivery of aircraft QAR data and can support engineers to analyze QAR data to check engine health. Instead of physical logistic, QAR upload subsystem can transfer QAR data electronically from shipside to the remote server via networks. Then, engine-related parameters are extracted from transferred QAR data and stored into the maintenance database. The engine-related parameters in the maintenance database can be shared among several engineers for engine health monitoring at different places and the analysis results will be also used for collaborative analysis.

Lifetime Prediction Routing Protocol for Wireless Sensor Networks

We propose LPDD (Lifetime Prediction Directed Diffusion), a novel energy-aware routing protocol for sensor networks that aims at increasing network survivability without a significant increase in latency. The key concept behind the protocol is the adaptive selection of routes by predicting the battery lifetime of the minimum energy nodes along the routes.

A Content-Based Music Retrieval System Using Multidimensional Index of Time-Sequenced Representative Melodies from Music Database

In content-based music retrieval systems, since both the correctness and the performance of retrievals are important, a few content-based music retrieval systems have the melody index which contains the representative melodies of music to be likely used as users’ queries. In this paper, we describe the development of a content-based music retrieval system in which the multidimensional index of time-sequenced representative melodies extracted appropriately based on musical composition forms is used to support quick and appropriate retrievals to users’ melody queries. From the experimental results, we can see that the developed system can retrieve more relevant results than previous systems with smaller storage overhead than whole melody index.