Optical Switching and Networking最新文献

Flexible Ethernet (FlexE)-over-WDM is one of the promising technologies of 5G transport networks that aims to provide a variety of Ethernet MAC rates over large capacity wavelength channels. Ultra-reliable communication is a key scenario of 5G, which poses a big challenge to the transport networks. To achieve ultra-reliability, frame replication and elimination for reliability (FRER) was proposed by IEEE. 802.1CB to provide redundant transmission in layer 2. However, previous studies of FRER considered to replicate MAC frames at source node and eliminate them at the destination, which may cause a waste of network resources, especially for the services that require different reliabilities in a condition of network with different packet loss rates of link and node. To this end, we propose a novel redundant routing scheme, called segment FRER (S-FRER), to balance the reliability and network efficiency. And a FlexE group allocation scheme is coming with S-FRER to assign the Flexible Ethernet channels along the path. Two heuristic algorithms, which are end-to-end FRER (E-FRER) and ordinal FRER (O-FRER), are compared with S-FRER. Simulation results show that S-FRER can satisfy diverse reliability requirements of services, and the resource redundancy and equipment cost are condensed.We further take the resource utilization efficiency as the optimization goal, and propose a FlexE Group allocation strategy based on segmented frame repetition and elimination to realize the resource redistribution of service flows on redundant routes. The experimental results verify the effectiveness of the scheme in terms of resource utilization efficiency.

Light fidelity (LiFi) is a wireless optical communication (WOC) technology that holds the key to solving the challenges of 5th and higher generation mobile networks. LiFi is a two-way wireless communication technology that enables high-speed transmission on both up and down links simultaneously. Today, researchers and manufacturers consider LiFi technology as an essential solution for radio frequency (RF) spectrum limitation due to growing demand of Internet users. In addition, integration of various communication technologies such as wireless fidelity (WiFi) and LiFi can help to overcome the traffic restriction challenge caused by the growing demand of Internet users. Although LiFi provides high-speed data transfer capability, it has some weaknesses such as coverage. So, hybrid WiFi/LiFi network can uses fast data transfer from LiFi and wide coverage from WiFi. Their integration can complement the shortcomings of each of these technologies and increases the network performance. In this work, while describing LiFi theory, the recent studies in this field are presented. Theoretical and practical concepts of LiFi-based indoor networks, issues such as network structures, cell deployment models, modulation techniques, multiple access schemes, criteria for measuring network performance, and scenario-based architectures for implementing a real LiFi-based indoor network based on up-to-date equipment manufactured by some LiFi manufacturers are discussed in detail.

In space division multiplexing-elastic optical networks (SDM-EONs), crosstalk and fragmentation are the significant factors determining overall resource utilization. Crosstalk and fragmentation management spectral resource allocation schemes can enhance the overall resource utilization in SDM-EONs. This paper proposes three crosstalk and fragmentation management bi-directional resource allocation schemes for multi-core SDM-EONs to suppress inter-core crosstalk and network fragmentation. A spatial resource partitioning method is presented to mitigate inter-core crosstalk, and this paper adopts a k-shortest path routing method to suppress fragmentation while spectral resource allocation. To the best of our knowledge, there is no such scheme that handles both crosstalk and fragmentation together while allocating spectral resources in a counter-propagating manner for multi-core fiber (MCF) SDM-EONs. The estimated result shows that the proposed scheme outperforms the traditional co-propagation method and enhances existing resource utilization by reducing blocked requests.

In a long-reach passive optical network (LR-PON), protection for optical network units (ONUs) against their distribution fiber (DF) failures is highly desirable to ensure uninterrupted Internet access to the associated users. In the distribution section of the LR-PON, shared-risk link groups (SRLGs) are formed by DFs due to sharing fiber cables and conduits. Most of the existing SRLG-aware DF protection schemes require residual bandwidth with ONUs. Therefore, they fail to provide protection to all ONUs under high reliability requirements (RRs) and result to be very cost-inefficient as well. We first propose a SRLG-aware reliability framework to compute the connection reliability of every ONU and joint reliability of every ONU-ONU pair and ONU-node pair. Thereafter, we propose two different SRLG-aware nonresidual bandwidth based partial protection schemes and their compatible architectures to provide partial DF protection to ONUs against SRLG failures to satisfy the given RR. Following the proposed schemes, we formulate two different integer linear programming (ILP) based optimization problems to set up backup connections by using backup DFs and other backup optical resources with the minimum protection cost. The proposed ILP-based protection planning models restrict the backup DF lengths to bound the propagation delay to satisfy the strict delay requirement for real-time applications. Since the proposed ILP-based models turn out to be computationally intractable for large network problems, we also propose two heuristic schemes that provide comparable results to that of the ILP-based models. We evaluate the performance of the proposed partial protection schemes with reference to the protection cost and total length of backup DFs (TLBDF). The exhaustive simulation results show that the proposed partial protection schemes not only satisfy high RRs, but also require a much lower protection cost, TLBDF, and optical power margin in comparison to the existing SRLG-aware protection schemes.

Benchmarking is commonly used in research fields, such as computer architecture design and machine learning, as a powerful paradigm for rigorously assessing, comparing, and developing novel technologies. However, the data centre network (DCN) community lacks a standard open-access and reproducible traffic generation framework for benchmark workload generation. Driving factors behind this include the proprietary nature of traffic traces, the limited detail and quantity of open-access network-level data sets, the high cost of real world experimentation, and the poor reproducibility and fidelity of synthetically generated traffic. This is curtailing the community's understanding of existing systems and hindering the ability with which novel technologies, such as optical DCNs, can be developed, compared, and tested.

We present TrafPy; an open-access framework for generating both realistic and custom DCN traffic traces. TrafPy is compatible with any simulation, emulation, or experimentation environment, and can be used for standardised benchmarking and for investigating the properties and limitations of network systems such as schedulers, switches, routers, and resource managers. We give an overview of the TrafPy traffic generation framework, and provide a brief demonstration of its efficacy through an investigation into the sensitivity of some canonical scheduling algorithms to varying traffic trace characteristics in the context of optical DCNs. TrafPy is open-sourced via GitHub and all data associated with this manuscript via RDR.

Underwater optical wireless communication (UOWC) is an emerging technology for underwater wireless sensor networks (UWSNs). Compared with acoustic waves and radio frequency waves, optical waves can provide higher data rates and lower latency. However, given the limited transmission range and energy supply, as well as the existence of communication void regions in UOWC which may further deteriorate its transmission performances (e.g. packet deliver rate), effective multi-hop routing schemes are highly desirable for UOWC to expand its transmission range with satisfactory performance. In this paper, we propose a dual-hop topology-aware (DHTA) routing algorithm for UOWC. By adopting dual-hop routing strategy, the proposed algorithm reduces the packet entry possibility into void regions, and thus improves the packet deliver rate. And by selecting the next relay node considering both the available distance of transmission and the deviation of transmission direction, the proposed algorithm can effectively reduce the average energy consumption of the network. Besides, a threshold is employed by the proposed algorithm for energy balance to prevent excessive energy consumption of some nodes. The simulation results verify that the proposed DHTA routing algorithm effectively alleviates the effluence of void regions on forwarding packet and increases the packet delivery rate (PDR), with reduced average energy consumption and energy standard deviation.

Time division multiplexing passive optical networks (TDM-PON) have emerged as a promising technology for mobile fronthaul for small cell CRAN. The performance of such TDM-based PON mobile fronthaul for small cell CRAN greatly depends on efficient dynamic bandwidth allocation (DBA) mechanisms. In this paper, a novel DBA mechanism based on the concept of dynamic service interval for ITU-based TDM-PON (XG-PON) for mobile fronthaul has been proposed. The proposed scheme is implemented using network simulator −3 (NS-3) simulation tool for evaluation of network performance parameters. The performance evaluation of the proposed DBA against the existing schemes such as optimized round robin (ORR) and Group-GIANT DBA shows a significant improvement in upstream mean delay, packet loss ratio and grant-request ratio. In addition to this, the proposed scheme shows that larger number of ONU(RRH) can be supported by the network for front haul uplink traffic transmitted via XG-PON in small cell CRAN.

Over the last decades, the emerging paradigms, e.g., centralized mobile cloud computing (MCC), multi-access edge computing (MEC), and collaborative computing offloading (CCO), have attracted extensive attention in adaptive 5G low latency communication for Tactile Internet. Besides, a novel integrated wireless optical transport network emerges in access networks and supports joint fronthaul and backhaul services. Therefore, considering that the task offloading response time, the needs of proximity, and the ultra-dense 5G small cell deployment, a novelty hybrid cloudlet deployment scheme over fiber and wireless backhaul-aware infrastructure is created with efficient cost optimization in mind. In this work, the envisioned Fiber Wireless networks (FiWi) consists of optical backhaul and wireless fronthaul (i.e., integrated access network and backhaul link), whereby base stations (BSs) with fiber and wireless backhauls are referred as wired-BS (WBS) and unwired-BS (UBS), respectively. Therefore, to meet the quality of service (QoS) delay constraints, cloudlets can be deployed in either UBSs, WBS, or remote node (RN). However, we apply mixed-integer line programming (MILP) to resolve the convex optimization problem in terms of minimization deployment cost. Besides, we describe the QoS-aware hybrid cloudlet placement cost algorithm for WBS and UBS coverage areas against different network conditions. It was shown through experimental measurements that the proposed architecture can achieve the scalability in different deployment scenarios. Also, the dependency of minimization cloudlet deployment cost on variable network parameters in terms of user density, network framework, and network QoS can be validated.

Elastic Optical Networks (EONs) using a flexible wavelength allocation technique with smaller granularity of spectrum can improve the efficiency and flexibility for diverse traffic transmission. However, due to the small granularity in EONs, the resource allocation problems become more complicated than that of the existing network architecture. By classifying the integer linear programming (ILP) models for solving the resource allocation problem on EONs into four categories, i.e., path/slot, path/channel, node/slot and node/channel, the characteristics of each are reviewed and tighter lower bounds estimation schemes for these ILP models are designed. Also, for channel models (i.e., path/channel and node/channel models), we propose two improved channel generation methods in pre-processing to reduce the number of variables. In the comparative experiments, we show the superiority of our methods, and by using them, analyze the characteristics of the four models.