Combining circuit and packet switching using a large port-count optical cross-connect for data center networks

The exponential increase in the data center network traffic has posed new challenges for achieving high throughput, low latency and energy consumption. To address these challenges, this paper presents a novel scheme that combines optical circuit and packet switching technologies together in a single optical cross-connect for data center networks. By utilizing fast tunable linecards and parallel array waveguide grating routers, the optical cross-connect can offer low latency, large scalability and high throughput in datacenter networks. A scheduling technique is designed to simultaneously accommodate circuit switching and packet switching in the optical cross-connect. The physical performance of this optical cross-connect in circuit and packet switching modes is investigated by experiments. In addition, the network performance of the optical cross-connect is evaluated through simulations under a mixed circuit/packet-switched traffic pattern. Results show that the network performance of the optical cross-connect does not degrade significantly with an increase in the number of racks, but it is affected by the circuit-switched traffic proportion.     

Effects of the switching time in OPS/OBS networks

In this paper we present an analytical model, which considers the effects of the switching time on the network performance in service differentiated optical packet/burst switching (OPS/OBS) networks. Our results indicate that the switching time must be less than 10 % of the packet/burst duration in order to avoid any significant reductions in the network performance. Furthermore, regarding a network with full wavelength conversion, we show that the benefits of statistical resource sharing are almost non-existent for low priority traffic when the switching time is large.     

光分组交换网络中多组光正交码组合光标签识别处理及实验

报道了光分组交换网络中基于多组光正交码组合光标签的识别与接收实验,光标签采用了码长为19,码重为3的2组光正交码组合光标签方案.给出了基于多组光正交码光标签的光分组交换网络(MOOC-OPS)基本原理,通过分析指出在MOOC-OPS网络中处理多组在时域上非连续、随机、突发光标签是现实MOOC-OPS关键技术之一.设计出了基于展宽网络和放大器级联方式接收多组低功率窄脉冲(MOOC-OPS网络的光标签)的实验方案,完成了实验电路板制作与实验验证.实验结果表明成功地实现了MOOC-OPS网络中周期为2ns基于多组光正交码光标签脉冲的识别接收,验证了该方案的可行性.     

A novel all-optical label processing based on multiple optical orthogonal codes sequences for optical packet switching networks

This paper proposes an all-optical label processing scheme that uses the multiple optical orthogonal codes sequences (MOOCS)-based optical label for optical packet switching (OPS) (MOOCS-OPS) networks. In this scheme, each MOOCS is a permutation or combination of the multiple optical orthogonal codes (MOOC) selected from the multiple-groups optical orthogonal codes (MGOOC). Following a comparison of different optical label processing (OLP) schemes, the principles of MOOCS-OPS network are given and analyzed. Firstly, theoretical analyses are used to prove that MOOCS is able to greatly enlarge the number of available optical labels when compared to the previous single optical orthogonal code (SOOC) for OPS (SOOC-OPS) network. Then, the key units of the MOOCS-based optical label packets, including optical packet generation, optical label erasing, optical label extraction and optical label rewriting etc., are given and studied. These results are used to verify that the proposed MOOCS-OPS scheme is feasible.     

Design of Two-Dimensional Optimized Banyan Networks

A smart two-dimensional optical Banyan network is proposed. Polarization switching logic of light is applied for the design of this network. The main element for the connections of the network is an optical crossbar switch that is composed of a half-wave plate, a programmable polarization switch, and a couple of birefringent plates. By the proposed optical crossbar switches, all permutations of the link patterns required in the network are generated due to the optimized design of the switching logic. As a result, complicated routing control of ordinary Banyan networks is much reduced and a simple two-dimensional optical Banyan network with straight interconnections between successive stages of the networks can be realized.     

Design and Performance Evaluation of Optical Ethernet Switching Architecture with Liquid Crystal on Silicon-Based Beam-Steering Technology

A non-blocking optical Ethernet switching architecture with liquid crystal on a silicon-based beam-steering switch and optical output buffer strategies are proposed. For preserving service packet sequencing and fairness of routing sequence, priority and round-robin algorithms are adopted at the optical output buffer in this research. Four methods were used to implement tunable fiber delay modules for the optical output buffers to handle Ethernet packets with variable bit-rates. The results reported are based on the simulations performed to evaluate the proposed switching architecture with traffic analysis under a traffic model captured from a real-core network.     

面向下一代航电系统的数据块填装性能

结合光分组交换(OPS)网络和光纤通道(FC)技术的优势, 提出一种下一代航电系统组网方案——基于光分组交换网络的光纤通道技术(FC over OPS)。建立了数学仿真模型, 研究了一种数据块填装算法与网络实时性之间的关系, 分析不同的参数如发送带宽、最低效率门限、发送定时的选取对网络实时性的影响。进一步完成硬件原型设计和仿真, 比较了软件仿真与硬件仿真的结果, 并分析该数据块填装算法的性能。     

The Impact of Signal Properties with Unbalanced Traffic on SOA Switches in Optical Packet Switched Networks

In this paper, we present the property evaluated modeling of switching matrix based on the XGM in SOA for optical packet-switched networks. Crosstalk-induced bit error rate (BER), packet error rate (PER), the dimension of the switching node and cascaded capacity under unbalanced traffic are examined, which proves to be relevant to the peculiarities of signal pulse such as initial chirp, shape and bit rate.Several rules are drawn to facilitate the design of SOA-based switching matrix under unbalanced traffic.     

An optical multicast packet switching architecture shared with limited-range multi-wavelength converter and full-range wavelength converters

Optical packet switch with multicast capability can inspire a broad range of multipoint to multipoint applications in future optical networks. An optical multicast packet switching architecture, equipped with feedback shared small number of limited-range multi-wavelength converters (LMWCs) and output shared some full-range wavelength converters (FRWCs), is proposed for a wavelength-division multiplexed optical multicast network to improve multicast performance in the paper. The FRWCs are used to overcome the performance degradation in terms of packet loss probability due to only use LMWCs. In the architecture, the two converters are shared by all the multicast packets importing to the optical multicast switch node. A maximum bipartite matching with minimum edges weights strategy was designed to employ fewer wavelength converters and avoid useless degradation of optical signal quality for the architecture to improve the node performance. The simulation results show that the proposed architecture and its wavelength converter scheduling algorithm can reduce the multicast packet loss probability with relatively lesser wavelength conversion cost.     

Performance analysis of an integrated scheme in optical burst switching high-speed networks

A new integrated scheme based on resource-reservation and adaptive network flow routing to alleviate contention in optical burst switching networks is proposed. The objective of the proposed scheme is to reduce the overall burst loss in the network and at the same time to avoid the packet out-of-sequence arrival problem. Simulations are carried out to assess the feasibility of the proposed scheme. Its performance is compared with that of contention resolution schemes based on conventional routing. Through extensive simulations, it is shown that the proposed scheme not only provides significantly better burst loss performance than the basic equal proportion and hop-length based traffic routing algorithms, but also is void of any packet re-orderings.     

An optimal burst assembly approach employing traffic shaping (OBATS) for OBS

Optical burst switching (OBS) aims at combining the strengths of packet and circuit switching and is considered as a promising technology for implementing the next generation optical Internet, required to cope with the rapid growth of Internet traffic and the increased deployment of new services. In this paper, an optimal burst assembly approach employing traffic shaping (OBATS) for OBS networks has been proposed in order to improve network performance in terms of reduced blocking probability, congestion control and better utilization of bandwidth. Particularly, the proposed scheme aims at reducing the average delay experienced by the packets during the burstification process in optical burst switched (OBS) networks, for a given average size of the bursts produced. Reducing the packet burstification delay, for a given average burst size, is essential for real-time applications; correspondingly, increasing the average burst size for a given packet burstification delay is important for reducing the number of bursts injected into the network and the associated overhead imposed on the core nodes. Simulation results show that the proposed burst assembly approach gives better network performance in terms of burst drop, resource contention and delay as compared to conventional burst assembly approaches.     

Analysis of Wavelength-Routed Optical Burst-Switched Network Performance

This paper studies a novel scalable network architecture combining optical burst switching (OBS) with dynamic wavelength allocation to guarantee quality of service (QoS), forming a wavelength-routed optical burst-switched network. All processing and buffering functions are concentrated at the network edge and bursts are assigned to fast tuneable lasers and routed over a bufferless optical transport core using dynamic wavelength assignment. Different burst aggregation mechanisms are evaluated for a range of traffic statistics in terms of delay and packet loss rate. New network performance parameters in an analytical model quantify the advantages of dynamic wavelength allocation. The results define the operational gain achievable with dynamic wavelength assignment compared to quasi-static wavelength routed optical networks.     

Investigation on performance of all optical buffer with large dynamical delay time based on cascaded double loop optical buffers

Optical buffers are critical for optical signal processing in future optical packet-switched networks. In this paper, a theoretical study as well as an experimental demonstration on a new optical buffer with large dynamical delay time is carried out based on cascaded double loop optical buffers (DLOBs). It is found that pulse distortion can be restrained by a negative optical control mode when the optical packet is in the loop. Noise analysis indicates that it is feasible to realise a large variable delay range by cascaded DLOBs. These conclusions are validated by the experiment system with 4-stage cascaded DLOBs. Both the theoretical simulations and the experimental results indicate that a large delay range of 1-9999 times the basic delay unit and a fine granularity of 25 ns can be achieved by the cascaded DLOBs. The performance of the cascaded DLOBs is suitable for the all optical networks.     

Multiaccess in all-optical networks with wavelength and code concurrency

A major obstacle in realizing fast packet switching in all-optical networks is the large tuning delays of tunable optical devices. This article proposes a multiaccess scheme for all-optical local area networks that employs both wavelength and code concurrency. In this scheme, several users share a wavelength channel through code multiplexing. The delay performance of hybrid wavelength/code division multiaccess is obtained under a simple, suboptimal access protocol based on cyclic search. Due to the reduction in the number of wavelength channels without an associated reduction in transmission concurrency, hybrid multiaccess is robust against tuning delays. At a given network throughput, the hybrid scheme achieves considerably lower delays than that of Wavelength Division Multiple Access even with a small amount of code concurrency. Conversely, the hybrid network can support a higher load when there is a maximum allowable value for the average packet delay.     

全光地址识别机理的研究

在面向分组业务的全光交换系统中,采用全光地址处理技术可以减少处理时延,大大提高信息处理速度和网络吞吐量,对全光网络的研究具有重要意义。分析了光正交码的自相关性与互相关性,并将其特性应用于全光传送网中,以针对未来基于光分组业务的全光环网为例,提出了一种光分插复用器的新型结构,给出了总体应用设计规则,重点研究和分析了全光地址识别机理和相关的应用设计方法,给出了针对全光环网组网规模的仿真分析结构和基本实验结果。结果表明采用所提出的设计原则和设计方法来实现全光地址识别是可行的。     

Nature-Inspired Routing and Wavelength Assignment Algorithms for Optical Circuit-Switched Polymorphic Networks

Polymorphic optical networks simultaneously support several optical switching paradigms over a single physical network. In this way, they provide service differentiation at the optical layer by employing the most appropriate paradigm for each service. One type of such architecture is the optical circuit-switched polymorphic network (OCSPN), which combines optical circuit switching paradigms with different grades of dynamism. The performance of this network relies on the utilization of efficient routing and wavelength assignment (RWA) algorithms. In this article, we review the fundamentals of OCSPNs and present a set of efficient RWA algorithms based on ant colony optimization and genetic algorithms.     

Analysis of Wavelength-Routed Optical Burst-Switched Network Performance

This paper studies a novel scalable network architecture combining optical burst switching (OBS) with dynamic wavelength allocation to guarantee quality of service (QoS), forming a wavelength-routed optical burst-switched network. All processing and buffering functions are concentrated at the network edge and bursts are assigned to fast tuneable lasers and routed over a bufferless optical transport core using dynamic wavelength assignment. Different burst aggregation mechanisms are evaluated for a range of traffic statistics in terms of delay and packet loss rate. New network performance parameters in an analytical model quantify the advantages of dynamic wavelength allocation. The results define the operational gain achievable with dynamic wavelength assignment compared to quasi-static wavelength routed optical networks.     

Research on application of MOOC in optical packet switching

An experiment on receiving and identifying multiple optical orthogonal codes （MOOC） sequence-based optical labels in an optical packet switching （OPS） network is reported. Two groups of MOOC with a code length of 23 and a code weight of 3 are used to identify the optical labels. The scheme and principle of OPS networks based on MOOC sequence-based optical labels are presented. Because optical labels contain OPS and routing information, the importance of optical label processing with multiple inconsecutive, random, and burst optical labels in an OPS network is pointed out. Considering the MOOC-based optical labels, we design a circuit consisting of a broadened network and a cascaded amplifier to receive multiple groups of low-power narrow pulses （optical labels in the MOOC-OPS network） with a period of 2 ns. The successful experimental results demonstrate that the designed scheme is feasible.     

Nature-Inspired Routing and Wavelength Assignment Algorithms for Optical Circuit-Switched Polymorphic Networks

Polymorphic optical networks simultaneously support several optical switching paradigms over a single physical network. In this way, they provide service differentiation at the optical layer by employing the most appropriate paradigm for each service. One type of such architecture is the optical circuit-switched polymorphic network (OCSPN), which combines optical circuit switching paradigms with different grades of dynamism. The performance of this network relies on the utilization of efficient routing and wavelength assignment (RWA) algorithms. In this article, we review the fundamentals of OCSPNs and present a set of efficient RWA algorithms based on ant colony optimization and genetic algorithms.     

Wavelength conversion in optical transport networks

The introduction of optical technology in the path layer of a transport network is analyzed; in particular, the impact of all-optical wavelength conversion is examined. Two basic optical cross-connect schemes based on space switching and on wave-length switching, respectively, and two types of wavelength converters are considered. The wavelength converters are, respectively, based on four-wave miring (FWM) in semiconductor optical amplifiers (SOAs) and on cross-phase modulation (XPM) obtained by two SOAs put in an interferometric configuration. An analysis of transmission performances of optical transport networks employing both types of wavelength conversion and both the node architectures is reported. The performances are evaluated by means of a system modeling that facilitates considering physical phenomena occurring during transmission and all the important transmission impairments. The analysis reveals that both the types of converters can be employed with different features and limitations, according to the considered network and switching architecture.