Research Topics

From WikiOna
Jump to: navigation, search

The activities of the Optical Networking Area are distributed in two broad research lines:

  • Optical network control and service management

Optical network control and service management focuses on control and service architectures (distributed and centralized), protocols, traffic engineering algorithms (provisioning, protection and restoration strategies), system design, and hardware and software developments for dynamic end-to-end QoS-enabled services in next generation multi-domain and multi-layer converged optical transport networks. Currently, research work aims at enhancing the intelligence, dynamicity, efficiency, robustness, capacity and cost-effectiveness of Wavelength-Switched Optical Networks (WSON), and Connection-oriented optical IP/Ethernet packet transport networks (e.g., MPLS-TP/PWE3, PBB-TE) through Generalized Multiprotocol Label Switching (GMPLS) and Path Computation Element (PCE).

This research line encompasses the design and performance evaluation of dynamic provisioning, protection and restoration strategies with traffic engineering, automated topology and resource discovery, automated resource reservation, on-line routing and wavelength assignment (RWA) and traffic grooming algorithms, adaptive service provisioning, and self-management and traffic characterization.

  • Optical transmission and subsystems

Optical transmission and subsystems is focused on specific physical layer and transmission aspects of optical communications, aiming at investigating and designing novel adaptive and efficient optical transmission systems for high-speed transparent transport networks. The purpose of this activity is to study advanced modulation formats and flexible technologies, as well as suitable monitoring techniques, for the mitigation of transmission impairments and the enhancement of system performance. Specifically, multicarrier modulation techniques are investigated for optical communication systems based on direct detection and coherent schemes. The objective is to furnish cost-effective solutions for flexible, high-capacity optical networks, optimizing the use of the existing infrastructure and the available resources.

These reaseach lines encompass several specific scientific and technological objectives:

Contents

Multi-layer Multi-Domain network architecture

The deployment of end-to-end services in a globalized context requires the effective cooperation of service providers, both at the content level and at the transport level. In our scope, this maps into the adoption of multi-domain and multi-layer network architectures, which bring new challenges for signaling, routing and path computation control plane components and the respective protocols. In particular, in a multi-domain scenario, routing exchange among domains is restricted to the exchange of summarized topology and TE attributes (e.g., bandwidth), as well as reachability information, which may lead to suboptimal paths if not detailed resource availability is known or even if every path segment is computed independently. Moreover, in the context of transparent optical networks, physical impairments should also be considered in the routing decisions. Therefore, the above routing information should be completed with physical-layer aspects, in order to compute inter-domain routes assuring end-to-end QoS.

On the other hand, in multi-layer networks, the dissemination of resource and TE attributes are not longer limited to tackling just a single (wavelength-) layer when setting up connections. Indeed, routing decisions must be made considering all dynamic layers involved in the transport of the flow, including at what points cross-layer operations should be performed. This gives rise to the so-called grooming operations. With regard to signaling, one of the challenges is to develop efficient signaling extensions for exchanging control information among nodes in order to perform and efficient resource reservation for the dynamic provisioning of wavelength and sub-wavelength paths, taking into account the requirements that must be accomplished at each switching layer, such as the wavelength continuity constraint in wavelength-routed networks. This activity aims at finding solutions for GMPLS RSVP-TE signaling protocol and GMPLS OSPF-TE routing protocol for multi-domain multi-layer networks, and the interaction with the Path Computation Element (PCE).

Network protection and restoration

In WSON networks, a network failure (e.g., fiber cut) would cause a large amount of data loss, degrading the QoS to the worst extent possible. Thus, fast and efficient survivable schemes are needed to minimize these effects as well as recovering disrupted services. However, not all the applications require the same level of reliability, mainly differing in how much customers are willing to pay. Therefore, the aim of the network operator is twofold: first, coping with the client reliability and availability necessities, and second, devising differentiated protection and restoration schemes to encompass efficient network performance in terms of resource usage, restoration time, network cost, etc. Furthermore, within a multi-layer network context, recovery schemes at each layer must be enhanced to provide inter-layer collaboration, and thus avoiding that interrupted traffic is recovered more than once. This activity focuses on studying and designing efficient protection and restoration schemes to provide differentiated recovery within the context of both single (WSON) and dual-layer (e.g., Ethernet over WSON).

Routing and wavelength assignment

In WSON networks, Routing and Wavelength Assignment (RWA) algorithms are executed to serve each optical connection request. To this end, two routing or path computation approaches are considered, namely, distributed (source-based) and centralized (PCE). In each approach, the aim of the employed RWA algorithm (Figure 3) is to find out the optical resources (i.e., wavelength channels) that accommodate the connection request through satisfying a set of specific constraints and TE requirements. In particular, in transparent WSON, the RWA strategies must be enhanced to efficiently address -the adequate optical signal quality. In other words, due to the physical impairments (e.g., noise, fiber loss, etc,) in the transmission medium, the optical signal is degraded whilst travelling from the source to the destination node. Consequently, the optical connection may not be feasible due the unacceptable optical signal quality at the receiver. The idea is thus to enhance the RWA algorithms to not only consider TE aspects but also to deal with requirements concerning physical impairments. Such algorithms are referred to as Impairment-aware RWA mechanisms.

Traffic grooming algorithms

In a unified multi-layer control plane (e.g., Ethernet over WSON), the path computation entity operates with detailed information about the network resources in each layer. Consequently, an optimal performance in terms of efficient resource usage may be attained at all the involved layers by means of the grooming strategies. The purpose of the grooming is to merge/group low-speed and flexible (Ethernet) connections with small bandwidth necessities into high-speed static (optical) tunnels with coarse bandwidth. By doing so, the grooming states a trade-off between the low cost of switching in the lower layer and the optimal resource utilization of switching in the higher layers. Despite the benefits provided by the grooming decisions in terms of the network resource optimization, the required amount of control information represents the main shortcoming. Specifically, handling detailed resource information per layer may dramatically increase the complexity and processing at the control plane. This activity focuses on the study, design and assessment of new path comptuations algorithms to improve the network performance within multi-layer (Ethernet over WSON) networks exploiting the grooming functions

Self-management and traffic characterization

This research topic focuses on the service plane as a means to forecast the behavior of Internet traffic and automate the generation of provisioning requests by IP routers to the GMPLS control plane in Wavelength-Routed Optical Networks (WSON). The application of self-management to all-optical networks was raised in 2007 related to the physical layer (optical performance monitoring being the enabler for such self-management) and to the provisioning of lambda connections (flow detection and management of lambda-connection provisioning requests being the enablers for such self-management). This activity will set the basis for a service plane optimized for WSON and at the same time easily upgradable to Ethernet over WSON. Generically, the functions of a service plane can be split in two logical elements: one containing information about the services (Service Data Element, SDE) and another managing the translation of IP traffic and services/applications into provisioning requests. This last element, called Service Control Element (SCE), is the one performing connectivity adaptation and service virtualization using information from the SDE, as well as QoS policies from the Service Level Agreements. Taking this into account, the basic functionalities of an SCE, especially those related to the characterization of IP traffic into flows that can be translated into lambda services will be designed and validated. These IP flows include heavy-hitter flows and other "big" IP flows from specialized applications that require huge amounts of bandwidth.

  • Research timeline: 2Q 2009 -
  • CTTC researchers involved: Carolina Pinart, PhD (2Q 2009 - 1Q 2011)
  • Related research topics: Adaptive service provisioning
  • Related research collaborations: --
  • Related publications:
    • C. Pinart, Anticipation of traffic demands to guarantee QoS in IP/optical networks, Journal of Future Internet, special issue "QoS in wired and wireless networks", vol. 2 no. 3, pp. 417-430, September 2010.

[ Useful biblio ]

Adaptive service provisioning

This research topic focuses on the service plane as a means to automate the connectivity adaptation and service virtualization necessary to handle complex services, e.g., a mesh of Virtual Private Networks (VPN), prior to requesting the provisioning of connectivity that can be "understood" by the GMPLS control plane, e.g., lightpaths or Ethernet bandwidth. This adaptation and virtualization will be performed in single- and multi-layer optical networks. A Service Data Element (SDE) will be designed, and its interactions with the Service Control Element (SCE) will be anayzed and validated. This activity will also encompass the translation strategies from composite services, e.g., VPN, into provisioning requests to the GMPLS control plane in charge of setting up and tearing down connections in the Ethernet over WSON network. These translation strategies will contribute to the common goal of achieving adaptive service provisioning through the accomplishment of service customization. Because of the diversity of service logics, we will focus on decomposing complex services into lower-order services, i.e., basic services that will translate directly into provisioning requests to the GMPLS control plane and willl focus on a set of complex services.

[ Useful biblio ]

Advanced modulation formats and novel multiplexing techniques

The purpose of this activity is studying advanced modulation formats and flexible technologies to design high spectrally efficient optical transmission systems. Specifically, OFDM-based modulation/multiplexing techniques are addressed within this objective. In fact, new advanced modulation formats enhance the supported bit rate per channel and offer increased spectral efficiency, but are sensitive to the accumulation of impairments. Optical OFDM technique is less sensitive to distortions and requires less complex systems. It offers alternative electronic dispersion compensation and also provides an increasing of the system spectral efficiency, especially if combined with high order modulation formats, orthogonal band multiplexing and/or polarization multiplexing, resulting a promising candidate for 100 Gb/s transmission systems.

Subsystems for direct detection and coherent transmission

DD and coherent schemes trade simplicity against increased sensitivity. DD systems are simpler and can be implemented by using commercial components; no laser is required as local oscillator (LO) at the receiver and a simple photodetector is used. The simplicity of this cost-effective solution is at expenses of the spectral efficiency and its effectiveness depends on the system linearity. Coherent detection is more costly and sensitive to phase noise and frequency offset; narrow linewidth lasers are required as Los, which must be phase-shifted by means of an optical 90º hybrid. However, combined with electronic processing, coherent systems are suitable for high-speed long-haul transmission, giving superior performance in terms of spectral efficiency, receiver sensitivity and robustness against dispersions. The S&T objective consists of first designing simple, cost-effective optical OFDM systems using DD; in particular, discrete multitone modulation (DMT) for the transmission of real-valued OFDM signals will be studied. Intensity modulation systems and power efficient techniques will be investigated and designed for cost-sensitive applications at speed beyond 10 Gb/s. In order to support higher bit rate (up to 100 Gb/s), enable the aggregation of multiple OFDM signals, and enhance the spectral efficiency and the attainable distance of the transmission system, the investigation of coherent schemes will be further addressed within this activity

Optical signal processing

A suitable signal processing plays a fundamental role in designing optical systems. In fact, the implementation of transmission schemes can be realized by using electronic and/or optical processing techniques. The former takes advantage of available digital signal processing (DSP) capabilities; the latter provides bit rate and signal format independent flexible schemes, preserving end-to-end optical transparency. Therefore, a proper identification of design requirements and technology limitations is a key issue. Specifically, transform-based signal processing for optical transmission technologies will be addressed within this activity. The suitability of using other transforms in place of FFT for the OFDM modulation/demodulation (e.g. real trigonometric transforms or wavelet transforms) will be assessed, in order to enhance the performance of optical OFDM processing in terms of flexibility and cost, and provide novel modulation schemes able to support higher spectral efficiency. Furthermore, potential all-optical architectures based on planar lightwave circuit (PLC) will be investigated, to enhance the performance of optical transmission systems, avoiding the electronic bottleneck.

Optical Performance Monitoring techniques

Numerous OPM techniques (e.g., optical spectrum analysis, polarization nulling, RF spectrum analysis, histogram sampling methods, and electrical equalizer coefficient comparison), have been proposed for monitoring various optical parameters such as the channel power, optical frequency, OSNR, Q-parameter, and dispersion (CD and PMD). An S&T objective is to identify the suitable technique and monitoring parameters for the investigated and designed optical transmission systems. Particular attention will be direct towards RF spectrum analysis with pilot tones, due to their potentially fast response time and their straightforward applicability to O-OFDM transmission technique. In fact, channel estimation is required for the equalization and correct recovery of OFDM signals. This leads to an acquisition of different parameters that can be used to track the impact of the sources of distortions and to monitor multiple simultaneous impairments (e.g. OSNR, CD, PMD, etc). The overhead amount (e.g. pilot tones, training symbols, cyclic prefix) depends on many factors and also on the system design and scenario (e.g. DD or coherent schemes), but in any case additional monitoring devices are not required.

Test-bed Modelling and experimental performance evaluation

Many leading research, industrial, academic and governmental organizations are deploying optical networking platforms and testbeds, in order to validate and assess advanced concepts aiming at decreasing operations costs and increasing revenues. Community researchers agree on the fact that there is a potential for change in the way that operators manage their high-cost transport networks, although still a long way from deployment in real production systems. This research topic drives and encompasses the experimental performance evaluation and research activities of the area. This is justified since the study and analysis in terms of performance of complex systems exclusively by means of theoretical analytical studies and/or simulations cannot capture the whole set of interdependencies between system components. Realistic models of data transport networks as replacement of actual research objects can only address specific and concrete aspects of the systems, for analytical models that can be considered realistic and generic become intractable.

In this framework, the experimental research activities not only validate assumptions or predictions conceived in more theoretical studies, but also to provide: a) inputs and feedback from actual system implementation and deployment justifying eventual (re-)design, while highlighting aspects of the observed system or phenomenon that are deemed worthy of further research and b) to provide meaningful conclusions, results and implementation guidelines on their own, which may be hard to obtain and comprehend from more analytical and fundamental research. Such activities consider the high level design of optical and multi-layer testbed architectures for demonstration and experimental performance evaluation of concepts, algorithms and protocols and assert the importance of real systems implementation.

The activity has a wide scope, including different aspects such as: i) measurement methodologies and tools, with the definition and evaluation of data collection systems and concepts, statistics generation methodologies and the synthesis of results into meaningful key performance indicators (see, for example Figure 4, which shows the traffic generation application); ii) automatic model-based testbed configuration, allowing the configuration of as many different network topologies as possible on top of the same physical infrastructure. This involves the modeling the desired experimentation scenarios and processing models with automatic tools and iii) Software and hardware development and modeling: this activity is conceived as a necessary step towards the concretization of research results and concepts into deployable systems, and acquires significant importance both for experimental research and technology transfer area objectives.

Due research topics

Non-intrusive monitoring for performance and fault management

This research topic studied minimum-intrusion, in-service monitoring alternatives combining L1/2/3 parameters to obtain enough parameters for the verification of Service Level Agreements (SLA) and for fault management, as well as reasonably good estimations of important SLA parameters (e.g., the BER) with minimum extra cost. The rationale behind this is that most of the performance monitoring parameters and techniques applied in deployed optical networks are digital and focused on particular quality parameters. With the introduction of transparency in optical networks, as well as new services with strict quality demands, such as latency (video conferencing) or availability (Grid), the needs and elements for monitoring, especially in-service, are changing. The first designs of equipment for WDM and photonic switches boosted research on optical signal monitoring, especially non-intrusive. At the time of writing, the desired, simple and reliable monitoring method did not exist (yet) for optical signals. In the same way, the identification and location of faults and failures in all-optical networks is complex due to fault propagation (i.e., redundancy and/or false alarms), the lack of digital information (unless some nodes in the network have multilayer capabilities or intrusive monitoring is used) and the processing time, which results from the bounded delay for triggering protection/restoration actions imposed by SLAs. In the literature, fault localization in all-optical networks has been approached in centralized and distributed ways. Taking into account the advantages and drawbacks of these two approaches, and the fact that the problem of locating multiple faults is NP-complete, this activity introduced hierarchical fault localization, and studied and assessed the possibility of using multi-layer performance information to achieve faster localization.

Main achievements of this activity (see related publications):

  1. Design and performance evaluation of a non-intrusive fault and performance management system in single-layer, all-optical, circuit-switched networks
  2. Design of minimum-intrusion alternatives for in-service BER estimation in all-optical, circuit-switched networks, including a closed-form expression to relate the OSNR and PLR in all-optical IP–Ethernet–WDM networks
  3. Proposal of a multi-layer fault localization framework for IP over all-optical multi-layer networks based on the above designs
Retrieved from "http://wikiona.cttc.es/ona/index.php/Research_Topics"
Personal tools