• Electronics and Telecommunications Trends
• /
• v.35 no.5
• /
• pp.57-68
• /
• 2020

This paper reviews the recent trends in optical access technologies and their future directions. As the number of hyper-connected, ultra-low-latency, and hyper-realistic services increases, the wireless path has become shorter and the optical access network has become deeply penetrated into the end user. The optical access network continues to evolve through endless innovation via high-speed, ultra-low-latency, and abstraction/virtualization technologies.

• Journal of Communications and Networks
• /
• v.6 no.2
• /
• pp.147-155
• /
• 2004

In this paper, we propose an efficient sequencing technique, namely minimum Row time scheduling (MFTS), to manage variable-Iength message transmissions for single-hop passive starcoupled WDM optical networks. By considering not only the message length but also the state of the receivers and the tuning latency, the proposed protocol can reduce the average delay of the network greatly. This paper also introduces a new channel assignment technique latency minimizing scheduling (LMS), which aims to reduce the scheduling latency. We evaluate the proposed algorithm, using extensive discrete-event simulations, by comparing its performance with shortest job first (SJF) algorithm. We find that significant improvement in average delay could be achieved by MFTS algorithm. By combining the proposed message sequencing technique with the channel selection technique, the performance of the optical network could be further improved.

• ETRI Journal
• /
• v.45 no.1
• /
• pp.60-74
• /
• 2023

Integration of the machine learning (ML) technique in all-optical networks can enhance the effectiveness of resource utilization, quality of service assurances, and scalability in optical networks. All-optical multistage interconnection networks (MINs) are implicitly designed to withstand the increasing highvolume traffic demands at data centers. However, the contention resolution mechanism in MINs becomes a bottleneck in handling such data traffic. In this paper, a select list of ML algorithms replaces the traditional electronic signal processing methods used to resolve contention in MIN. The suitability of these algorithms in improving the performance of the entire network is assessed in terms of injection rate, average latency, and latency distribution. Our findings showed that the ML module is recommended for improving the performance of the network. The improved performance and traffic grooming capabilities of the module are also validated by using a hardware testbed.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.50 no.7
• /
• pp.131-139
• /
• 2013

To overcome the limitations in performance and power consumption of traditional electrical interconnection based network-on-chips (NoCs), a hybrid optical network-on-chip (HONoC) architecture using optical interconnects is emerging. However, the HONoC architecture should use circuit-switching scheme owing to the overhead by optical devices, which worsens the latency unfairness problem caused by frequent path collisions. This resultingly exert a bad influence in overall performance of the system. In this paper, we propose a new task mapping algorithm for optimizing latency by reducing path collisions. The proposed algorithm allocates a task to a certain processing element (PE) for the purpose of minimizing path collisions and worst case latencies. Compared to the random mapping technique and the bandwidth-constrained mapping technique, simulation results show the reduction in latency by 43% and 61% in average for each $4{\times}4$ and $8{\times}8$ mesh topology, respectively.

• Journal of the Institute of Electronics and Information Engineers
• /
• v.51 no.11
• /
• pp.83-93
• /
• 2014

It is a widespread concern that electrical interconnection based network-on-chip (NoC) will ultimately face the limitation in communication bandwidth, transmission latency and power consumption in the near future. With the development of silicon photonics technology, a hybrid optical network-on-chip (HONoC) which embraces both electrical- and optical interconnect, is emerging as a promising solution to overcome these problems. Today's leading edge systems-on-chips (SoCs) comprise heterogeneous many-cores for higher energy efficiency, therefore, extended study beyond regular topology based NoC is required. This paper proposes an energy and latency optimization topology design technique for HONoC taking into account the traffic characteristics of target applications. The proposed technique is implemented with genetic algorithm and simulation results show the reduction by 13.84% in power loss and 28.14% in average latency, respectively.

• ETRI Journal
• /
• v.43 no.1
• /
• pp.64-73
• /
• 2021

Two-dimensional torus network nodes are typically interconnected using XY routing algorithm for transmitting a packet from a source node to a destination node. In XY routing, if all the paths are used efficiently, the throughput and latency can be improved. In this paper, to utilize all the paths efficiently, we propose a novel binary optical routing algorithm (BORA) to improve the throughput and latency. The throughput is calculated according to the injection rate and number of packets received at the destination. The XY routing algorithm and proposed BORA are implemented using objective modular network testbed in C++ simulation software and the results are analyzed and compared. In this paper, the simulation results show that the network latency reduces to 50% while using the proposed algorithm; moreover, the throughput is also improved.

• Annual Conference of KIPS
• /
• 2021.11a
• /
• pp.878-879
• /
• 2021

Most state-of-the-art CNNs for action recognition are based on a two-stream architecture: RGB frames stream represents the appearance and the optical flow stream interprets the motion of action. However, the cost of optical flow computation is very high and then it increases action recognition latency. We introduce a design strategy for action recognition inspired by a two-stream network and teacher-student architecture. There are two sub-networks in our neural networks, the optical flow sub-network as a teacher and the RGB frames sub-network as a student. In the training stage, we distill the feature from the teacher as a baseline to train student sub-network. In the test stage, we only use the student so that the latency reduces without computing optical flow. Our experiments show that its advantages over two-stream architecture in both speed and performance.

• Journal of the Optical Society of Korea
• /
• v.20 no.6
• /
• pp.678-685
• /
• 2016

The standard high-availability seamless redundancy (HSR) protocol utilizes duplicated frame copies of each sent frame for zero fail over time. This means that even in cases of a node or link failure, the destination node will receive at least one copy of the sent frame, resulting in no network downtime. However, the standard HSR is mostly based on the electrical signal connection inside the node, which leads to the production of considerable latency at each node due to frame processing. Therefore, in a large scale HSR ring network, the accumulated latencies become significant and can often restrict the mission-critical real-time application of HSR. In this paper, we present a novel design for optical HSR (OHSR) that uses beam splitting/combining techniques. The proposed OHSR passes the frames directly to adjacent nodes without frame processing at each node, thereby theoretically generating no latency in any node. Various simulations for network samples, made to validate the OHSR design and its performance, show that the OHSR outperforms the standard HSR.

• Proceedings of the Optical Society of Korea Conference
• /
• 2003.07a
• /
• pp.10-11
• /
• 2003

완전 광 레이블 스위칭 기술은 차세대 광 인터넷에서 요구되는 투명성과 낮은 latency를 주기 위한 핵심기술이다. 패킷 스위칭 시스템은 각 노드에서 레이블을 이용해 패킷을 교환하는 방식으로 각 라우터에서는 이전 라우터에서 전달해 은 레이블을 분석하고 그 레이블 신호에 따라서 목적지로 포워딩 시키게 된다. 레이블을 교환하는 기술은 MLS 또는 완전 광 레이블 스위칭 (Optical label Switching： OLS) 네트웍에서 scalability를 구현하기 위한 핵심 기술이 된다. (중략)

• KIPS Transactions on Computer and Communication Systems
• /
• v.11 no.3
• /
• pp.99-104
• /
• 2022

Convolutional neural network (CNN) generally uses two-stream architecture RGB and optical flow stream for its action recognition function. RGB frames stream display appearance and optical flow stream interprets its action. However, the standard method of using optical flow is costly in its computational time and latency associated with increased action recognition. The purpose of the study was to evaluate a novel way to create a two sub-networks in neural networks. The optical flow sub-network was assigned as a teacher and the RGB frames as a student. In the training stage, the optical flow sub-network extracts features through the teacher sub-network and transmits the information to student sub-network for baseline training. In the test stage, only student sub-network was operational with decreased in latency without computing optical flow. Experimental results shows that our network fed only by RGB stream gets a competitive accuracy of 54.5% on HMDB51, which is 1.5 times better than that on R3D-18.