• International journal of advanced smart convergence
• /
• v.10 no.1
• /
• pp.56-61
• /
• 2021

Neighbor path selection is to pre-select alternate routes in case geographically correlated failures occur simultaneously on the communication network. Conventional heuristic-based algorithms no longer improve solutions because they cannot sufficiently utilize historical failure information. We present a novel solution model for neighbor path selection by using machine learning technique. Our proposed machine learning neighbor path selection (ML-NPS) model is composed of five modules- random graph generation, data set creation, machine learning modeling, neighbor path prediction, and path information acquisition. It is implemented by Python with Keras on Tensorflow and executed on the tiny computer, Raspberry PI 4B. Performance evaluations via numerical simulation show that the neighbor path communication success probability of our model is better than that of the conventional heuristic by 26% on the average.

• International journal of advanced smart convergence
• /
• v.12 no.3
• /
• pp.13-18
• /
• 2023

This paper aims to investigate data features for neighbor path selection (NPS) in computer network with regional failures. It is necessary to find an available alternate communication path in advance when regional failures due to earthquakes or forest fires occur simultaneously. We describe previous general heuristics and simulation heuristic to solve the NPS problem in the regional fault network. The data features of general heuristics using proximity and sharing factor and the data features of simulation heuristic using machine learning are explained through examples. Simulation heuristic may be better than general heuristics in terms of communication success. However, additional data features are necessary in order to apply the simulation heuristic to the real environment. We propose novel data features for NPS in computer network with regional failures and Keras modeling for computing the communication success probability of candidate neighbor path.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.28 no.8A
• /
• pp.581-590
• /
• 2003

This paper proposes the routing path selection mechanism for source routing-based PtMP (Point-to-Multipoint) call in ATM switching system. Especially, it suggests PtMP routing path selection method that can share the maximum resource prior to the optimal path selection, guarantee the reduction of path calculation time and cycle prevention. The searching for the nearest branch point from destination node to make the maximum share of resource is the purpose of this algorithm. Therefore among neighbor nodes from destination node by back-tracking, this algorithm fixes the node crossing first the node on existing path having the same Call ID as branch node, constructs the optimal PtMP routing path. The optimal node to be selected by back-tracking is selected by the use of Dijkstra algorithm. That is to say, PtMP routing path selection performs the step of cross node selection among neighboring nodes by back-tracking and the step of optimal node selection(optimal path calculation) among neighboring nodes by back-tracking. This technique reduces the process of search of routing information table for path selection and path calculation, also solves the cycle prevention easily during path establishment.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2013.02a
• /
• pp.357-358
• /
• 2013

The non-linear characteristics of ON states are important for the application to the high density cross-point memory industry because the sneak current in neighbor cells occurred during reading, erasing, and writing process. Kw of above 20 in ON states, which is the writing current @ Vwrite/the current @ 1/2Vwrite, was required in cross-point ReRAM memory industry. The high current density non-linear IV curve of ZnSe selector was shown and the ALD HfO2 switching device has the linear properties of ON states and the compliance current of 100 uA. To evaluate the performance of the selection device, we connected itto HfO2 switching device in series. The bottom electrode of the selection device was connected to the top electrode of the RRAM. All of the bias was applied with respect to the top electrode of the selection device, whereas the bottom electrode of the RRAM was grounded. In the cross-point application, 1/2Vwrite and -1/2Vwrite were applied to the word-line and bit-line, respectively, which were connected to the selected cell, and a zero bias was applied to the unselected word-lines and bit-lines. The current @ 1/2Vwrite of the unselected cells was blocked by the selection device, thus eliminating the sneak path and obtaining a writing voltage margin. Using this method, the writing voltage margin was analyzed on the basis of the memory size.

• Journal of the Korea Society for Simulation
• /
• v.16 no.2
• /
• pp.17-26
• /
• 2007

In this paper, we propose a basic virtual grid-based routing algorithm in order to devise an efficient routing method in ad hoc networks using the location information of nodes, energy level, etc. A packet is forwarded to the X-axis direction at first based on the location information of a destination node, and then it is forwarded to the Y-axis direction as its location becomes close to the destination from the viewpoint of the X-axis. Due to the selection of next hop nodes to deliver a packet from a certain node to a destination node, we can regard the whole network as a virtual grid network. The proposed routing algorithm determines routing paths using the local information such as the location information of a destination and its neighbor nodes. Thus, the routing path setup is achieved locally, by which we can expect reduction in network traffics and routing delays to a destination. To evaluate the performance of the proposed routing algorithm, we used the network simulator ns2 and compared its network throughput with that of an existing routing algorithm.