• Journal of Korean Institute of Industrial Engineers
• /
• v.24 no.4
• /
• pp.661-671
• /
• 1998

It is important to minimize the number of disk accesses which is necessary to transfer data in disk into main memory when processing transactions in physical database design. A vertical file partitioning method is used to reduce the number of disk accesses by partitioning relations vertically and accessing only necessay fragments. In this paper, SOFM(Self-Organizing Feature Maps) network is used to solve vertical partitioning problems. This paper shows that SOFM network is efficient in solving vertical partitioning problem by comparing approximate solution of SOFM network with optimal solution of N-ary branch and bound method. And this paper presents a heuristic algorithm for allocating duplicate attributes to vertically partitioned fragments. As branch and bound method requires particularly much computing time to solve large-sized problems, it is shown that SOFM network is able to overcome this limitation of branch and bound method and solve large-sized problems efficiently in a short time.

• Journal of the Korean Institute of Telematics and Electronics
• /
• v.27 no.7
• /
• pp.1025-1032
• /
• 1990

We proposed a neural network energy function for the optimal graph partitioning and its optimization method using Boltzmann Machine. We composed a Boltzmann Machine with the proposed neural network energy function, and the simulation results show that we can obtain an optimal solution with the energy function parameters of A=50, B=5, c=14 and D=10, at the Boltzmann Machine parameters of To=80 and \ulcorner0.07 for a 6-node 3-partition problem. As a result, the proposed energy function and optimization parameters are proved to be feasible for the optimal graph partitioning.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.1
• /
• pp.9-15
• /
• 2021

Recently, it is increasingly necessary to run high-end neural network applications with huge computation overhead on top of resource-constrained embedded systems, such as wearable devices. While the huge computational overhead can be alleviated by distributed neural networks running on multiple separate devices, existing distributed neural network techniques suffer from a large traffic between the devices; thus are very vulnerable to communication failures. These drawbacks make the distributed neural network techniques inapplicable to wearable devices, which are connected with each other through unstable and low data rate communication medium like human body communication. Therefore, in this paper, we propose a distributed neural network partitioning technique that is resilient to communication failures. Furthermore, we show that the proposed technique also improves the inference accuracy even in case of no communication failure, thanks to the improved network partitioning. We verify through comparative experiments with a real-life neural network application that the proposed technique outperforms the existing state-of-the-art distributed neural network technique in terms of accuracy and resiliency to communication failures.

• Proceedings of the KIEE Conference
• /
• 1999.07c
• /
• pp.1067-1069
• /
• 1999

This paper presents a network reduction using weak coupling method. Weak coupling method of identifying coherent generator groups are proposed. The partitioning technique used in this paper is based on a property of sparse matrix factorization. When a matrix has been factorized, a system is divided into study area, boundary buses and external area. A reduction process for external system starts with the load bus elimination and coherent generator aggregation. An identification of coherent generator group, network partitioning and network reduction are presented.

• Proceedings of the Society of Korea Industrial and System Engineering Conference
• /
• 2002.05a
• /
• pp.105-111
• /
• 2002

Presented is a Genetic Algorithm(GA) for dynamic partitioning an ATM LANE(LAN Emulation) network. LANE proves to be one of the best solutions to provide guaranteed Quality of Service(QoS) for mid-size campus or enterprise networks with a little modification of legacy LAN facilities. However, there are few researches on the efficient LANE network operations to deal with scalability issues arising from broadcast traffic delivery. To cope with this scalability issue, proposed is a decision model named LANE Partitioning Problem(LPP) which aims at partitioning the entire LANE network into multiple Emulated LANs(ELANs), each of which works as an independent virtual LAN.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.34C no.7
• /
• pp.70-81
• /
• 1997

In this paper, we propose HMCNN(hybrid multiple component neural networks) that enhance performance of MCNN by adapting new pattern partitioning algorithm which can cluster many input patterns efficiently. Added neural network performs similar learning procedure that of kohonen network. But it dynamically determine it's number of output neurons using algorithms that decide self-organized number of clusters and patterns in a cluster. The proposed network can effectively be applied to problems of large data as well as huge networks size. As a sresutl, proposed pattern partitioning network can enhance performance results and solve weakness of MCNN like generalization capability. In addition, we can get more fast speed by performing parallel learning than that of other supervised learning networks.

• Journal of Communications and Networks
• /
• v.16 no.6
• /
• pp.667-676
• /
• 2014

Connectivity is a crucial quality of service measure in wireless sensor networks. However, the network is always at risk of being split into several disconnected components owing to the sensor failures caused by various factors. To handle the connectivity problem, this paper introduces an in-advance mechanism to prevent network partitioning in the initial deployment phase. The approach is implemented in a distributed manner, and every node only needs to know local information of its 1-hop neighbors, which makes the approach scalable to large networks. The goal of the proposed mechanism is twofold. First, critical nodes are locally detected by the critical node detection (CND) algorithm based on the concept of maximal simplicial complex, and backups are arranged to tolerate their failures. Second, under a greedy rule, topological holes within the maximal simplicial complex as another potential risk to the network connectivity are patched step by step. Finally, we demonstrate the effectiveness of the proposed algorithm through simulation experiments.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2020.06a
• /
• pp.115-115
• /
• 2020

Water network partitioning (WNP) is an initiative technique to divide the original water distribution network (WDN) into several sub-networks with only sparse connections between them called, District Metered Areas (DMAs). Operating and managing (O&M) WDN through DMAs is bringing many advantages, such as quantification and detection of water leakage, uniform pressure management, isolation from chemical contamination. The research of WNP recently has been highlighted by applying different methods for dividing a network into a specified number of DMAs. However, it is an open question on how to determine the optimal number of DMAs for a given network. In this study, we present a method to divide an original WDN into DMAs (called Clustering) based on community structure algorithm for auto-creation of suitable DMAs. To that aim, many hydraulic properties are taken into consideration to form the appropriate DMAs, in which each DMA is controlled as uniform as possible in terms of pressure, elevation, and water demand. In a second phase, called Sectorization, the flow meters and control valves are optimally placed to divide the DMAs, while minimizing the pressure reduction. To comprehensively evaluate the WNP performance and determine optimal number of DMAs for given WDN, we apply the framework of multiple-criteria decision analysis. The proposed method is demonstrated using a real-life benchmark network and obtained permissible results. The approach is a decision-support scheme for water utilities to make optimal decisions when designing the DMAs of their WDNs.

• ETRI Journal
• /
• v.24 no.6
• /
• pp.473-476
• /
• 2002

This letter proposes an efficient kernel-based partitioning algorithm for reducing area and power dissipation in combinational circuit designs using don't-care sets. The proposed algorithm decreases power dissipation by partitioning a given circuit using a kernel extracted from the logic. The proposed algorithm also reduces the area overhead by minimizing duplicated gates in the partitioned sub-circuits. The partitioned subcircuits are further optimized utilizing observability don't-care sets. Experimental results for the MCNC benchmarks show that the proposed algorithm synthesizes circuits that on the average consume 22.5% less power and have 12.7% less area than circuits generated by previous algorithms based on a precomputation scheme.

• Journal of Communications and Networks
• /
• v.8 no.1
• /
• pp.13-20
• /
• 2006

In cellular mobile communications, how to achieve optimum system capacity with limited frequency spectrum is one of the main research issues. Many dynamic channel assignment (DCA) schemes have been proposed and studied to allocate the channels more efficiently, thus, the capacity of cellular systems is improved. Reuse partitioning (RP) is another technique to achieve higher capacity by reducing the overall reuse distance. In this paper, we present a network-based DCA scheme with the implementation of RP technique, namely dynamic reuse partitioning with interference information (DRP-WI). The scheme aims to minimize the effect of assigned channels on the availability of channels for use in the interfering cells and to reduce their overall reuse distances. The performance of DRP-WI is measured in terms of blocking probability and system capacity. Simulation results have confirmed the effectiveness of DRP-WI scheme. Under both uniform and non-uniform traffic distributions, DRP-WI exhibits outstanding performance in improving the system capacity. It can provide about 100% capacity improvement as compared to conventional fixed channel assignment scheme with 70 system channels.