Metro networks are today a 10 billion dollar equipment business per annum. The first generation of metro WDM networks used the lightpath technology that involved setting up of optical circuits on wavelengths (lamdas). To make this technology scale beyond the long-haul, especially into the metro and to be able to propose a low-cost capital and operational set up, a new technology was invented by us and is called light-trails. Light-trails are a generalized lightpath such that multiple nodes can time-share the bandwidth offered by a single wavelength. Light-trail communication brings for the first time dynamic bandwidth allocation, optical multicasting and sub-wavelength support at price-points that have been lower than contemporary lightpath and other technologies. Light-trails were prototyped and commercialized in nascent versions. We now investigate new areas of light-trails that are commercially viable and scientifically daunting. Amongst these are:

• Two Stage Auction Protocols for Light-trails: Algorithms that facilitate creation of virtual topology on multi-degree ROADMs that support optical bus features such as dynamic bandwidth provisioning, optical layer multicasting and sub-wavelength spatial grooming.
• Multi-hop Light-trails (MLTs): light-trails are analogous to optical buses. The bus feature guarantees effective dynamic bandwidth allocation, optical multicasting as well as spatial wavelength grooming. However, ROADM loss to support a bus is almost 16dB or even more in the express configuration using advance WSS devices. Hence, we are researching on ways to improve the reach of light-trails. To this end, a new solution is being developed called Multi-Hop Light-Trails or MLT. MLT is a technology whereby one can move from one light-trail to another over an electronic cross over fabric, subject to the constraint that the two light-trails are overlapping (have common nodes). The MLT network design problem is further complicated when we consider multi-degree ROADMs where optimizing card count and minimizing OPEX are key design features.
• Energy efficient Light-trails: Recently the Energy Efficient Ethernet study group has proposed the P802.3az standard draft for energy efficient communications. We are extending this aspect to even optical networks. Since in a light-trail (bus), multiple nodes time-share the bandwidth, the question we ask is can we use the control channel to intimate the end-nodes out-of-band and ahead in time so that they can be switched off when not in use? This is the concept of EE Light-trails that we are currently investigating.

Funding Agencies: Tata Group, Fujitsu, DAE, Government of India, IIT Bombay

A new era of web based services in the Internet have emerged in the last decade. Popularly called Quad play (VOIP, VOD/IP TV, multi-media services and their provision on the last mile backhaul), these services have cast stringent requirements on the backhaul systems of the Internet in terms of consistency, high speed and delay tolerance. Ethernet is a layer-2 technology that has emerged as being adaptable, simple, economical and most importantly has largest deployment base. s. Due to the increasing demand from the newly emerged services, a requirement has been felt to adapt Ethernet IEEE 802.3 standard to speeds of 100Gbps. 

IEEE 802.3 Ethernet encompasses Media Access Control (MAC) Layer, Reconciliation Sub layer (RS), Physical Coding Sub layer (PCS), Physical medium attachment (PMA) and Physical Medium Dependant (PMD) sub layers. Broadly these are called as MAC and PHY layers. Most of these layers need to change to support 100Gbps operation. The output of the PHY layer is in terms of 20 x 5.1562Gbps or 4 x 25Gbps or any other combination that can be supported by an individual/ Enterprise. Considering the evolution of previous versions of Ethernet, it is anticipated that providing a serial 100Gbps interface with a 100Gbps NIC would be the ultimate aim. In view of this, it is pertinent to investigate serial 100Gbps transmission in fiber optics.

Most of the deployments in the core of the internet which use optics for data transport use amplitude modulation. Although, numbers of researchers have investigated multi-level modulation schemes and their viability, amplitude modulation using Optical Time Domain Multiplexing (OTDM) remains attractive due to simplicity of modulation and large deployment base. In this work we investigate the dispersive effects on a pulse in nonlinear medium, methods to achieve OTDM system and number of simulation experiments to establish the viability of the 100Gbps serial transmission.

Funding Agencies: Government of India, Agilent Technologies

In this project, we propose a theory for service-oriented communications based on a concept of space and time auctions. Space and Time Auctions are developed as a new class of resource allocation tools that enable efficient and optimal allocation of service entities. Users are free to align with markets and a set of users aligned with a particular market time-share the resource (typically bandwidth), provided by the market. The markets may be physical or virtual, and users are assumed to be coerced into fair trading of resources across these markets. When a user assumes deceitful strategies for resource utilization, the market reacts by deflating the user’s revenue. Optimal allocation is achieved when users follow truthful elicitation as their dominant strategy. Space auctioning is developed between lateral markets, allowing users to move between markets with a view to maximizing revenue. A relationship between space auctioning and time auctioning is also put forth that leads to a comprehensive theory on resource allocation and logical topology growth. Further, we are in the process of developing the proofs and numerical results of this model.

Funding Agencies: Government of India

Light-trails have recently been proposed as a solution for optical networking to provide support for emerging services such as Triple Play, Video-on-Demand, Pseudo-wires, etc. The features of dynamic provisioning, optical multicasting, sub-wavelength grooming through a low-cost evolutionary setup make light-trails a good choice for deployment in future metro networks. Significant work is available in the literature in the area of optimization of light-trail networks and shown to be analogous to bin-packing. An area that has not yet been considered and which is of service provider importance is the stochastic behavior and growth of the light-trail virtual topology. In this project, we propose a two-stage auctioning algorithm that efficiently and in a fair manner allocates bandwidth to nodes within a light-trail and also grows the virtual topology of light-trails using definition principles that are based on utility theory. The algorithm facilitates growth of the light-trail topology taking into consideration engineering specifications as well as using novel socio-political concepts in accordance with utility theory. The algorithm is stated, analyzed and simulated. The final outcome of the growth algorithm is an autonomic optical network that suffices service provider requirements while lowering operational and capital costs.

Funding Agencies: Tata Group

Bandwidth Problems in India

India has been experiencing flat to modest growth of Internet services in the past decade, despite the fact that India continues to add 8-10 million cellular connections each month. The reasons for this sluggish growth are pricing fallacies of Internet connectivity of which in particular is the usage based pricing model employed by ISPs for retail customers and a weak demand for broadband services in India. We examine the reasons that affect the high cost of bandwidth in India and steps required to reduce the same. We argue that non-existent SLA for last-mile residential users, content coming from overseas and the usage based pricing model, which is widely used for billing retail customers in the country, is hampering the growth of local content and services. Moving the contents to the data centers in India, investment in improving the last mile and a flat rate pricing model will spur demand for Internet services and enable content providers to target the local and emerging market. The project concentrates on how profit margin will be improved if content moves to India.

Funding Agencies: North American Vendor.

We present for the first time an implementation of optical packet transport as a solution for the access area. Two features affect design of optical access networks: (1) Traffic being characterized as – packetized (subwavelength granularity per node), delay sensitive, multi-service oriented. (2) Fiber deployment contributing as high as 90 % of network-layout cost. Light-Mesh architecture has been shown to result in 70~85% capital expenditure savings than contemporary star-PONs. The principle challenge in deploying nodes in mesh configuration with passive and lowcost features is to enable packet oriented quality-sensitive communication. We showcase the light-mesh test-bed that supports packet transport and triple-play traffic. A Light-Mesh test-bed was setup to demonstrate the architecture.

Funding Agencies: European Systems vendor.

WDM channels are now transporting data at speeds well beyond 10Gbps into the 40 Gbps and 100 Gbps per channel range. At speeds beyond 10Gbps per channel, the behavior of the optical layer is significantly different than at lower rates. For efficient transmission and long-reach, the treatment of channels at higher speeds should therefore be different than those for lower speeds. The one-size-fits-all philosophy that has governed the architecture for ROADMS and OXCs (primarily for cost reasons) leads to non-discrimination between WDM channel line-rate, fails on 2 counts in higher-speed channel systems: (1) at higher rate the pass-band especially due to AWG/WSS walls has strong bearing on channel throughput adversely affecting OSNR. (2)  the overall capacity of the system is degraded in terms of both throughput as well as performance. Using the current ITU spacing of 50 GHz and 100 GHz implies a maximum of 100~200 channels in the 1520-1590 nm band not meeting the emerging needs of cloud computing, data centers, video dissemination etc. 

The current network can be estimated as one with four distinct traffic types: short haul high capacity, medium haul high capacity, long haul high capacity and short haul low capacity. Electronic aggregation (grooming) is used for long haul low capacity and other traffic mixes (not included here) to convert the same into high capacity traffic. Traffic routes diligently follow the 80:20 rule [2], in the sense that there are a large number of short-haul routes with low capacity (sub 2.5Gbps). While one approach to maximizing fiber plant usage is to electronically groom much of this traffic using subsystems called muxponders, this procedure does not alleviate the problem for the other traffic types – notably the long and medium haul high capacity traffic – which is impaired by the ITU spacing as it traverses successive optical nodes loosing signal power and being exposed to unwanted device imparities. Moreover muxponders are expensive to deploy on account of inherent electronic processing as well as are protocol limited, meaning a specific egress protocol at the network side on which client protocols are transported. While the later may not initially seem to be a problem, in fact, for emerging services which require one or more of the properties of dynamic bandwidth provisioning, multicasting and sub-wavelength support, mapping client traffic into a network protocol (Ethernet over SONET or GFP or OTN) implies significant overhead and cost function. Hence, there is some merit in keeping client traffic in lower-speed lanes, from both CAPEX (lower cost transponders) and OPEX (dynamic offering) perspectives.

Our proposal: The ROAMTS proposal is based on the following assumptive principles: (1) the dissemination and commoditization of WSS technology, (2) the commercialization of slow-tunable full C-band lasers. The ROAMTS proposal architecture is one where we facilitate a higher capacity system using the two assumptions in conjunction with a novel node architecture. The central idea is to empower with a wavelength map that is not static, optimized for different traffic requirements yet backward compatible with the current ITU grid. One of the design features of our architecture is to eliminate rigid pass-band characteristics to be exhibited by nodes as the signal passes through. When we combine the advantage obtained through pass-band greasing with an ability to provide a new wavelength map, we obtain an optimized network – for multi-rate distance optimized lightpaths. We propose a node architecture that allows us to support a wavelength map that is optimized for all four traffic types in terms of distance (BER) as well as cost (ability to accommodate a maximum number of channels). This node architecture that we will showcase has the added advantage of being able to move close to the theoretical limit in terms of the bandwidth distance product at the lowest possible cost per bit.

We investigate algorithms to emulate EPLAN, EVPLAN and EVC behavior using TLS Ethernet over SONET/SDH. The key for this is the development of a next generation inverse multiplexing engine that can take advantage of the LCAS and VCAT features of SONET/SDH while implementing a GMPLS like control plane to dynamically manage the network. Key aspects of this new architecture are currently being defined as well as how this can be implemented in a generic framework.

Funding Agencies: Classified funding agency.

Since invented by Bob Metcalfe in 1978, Ethernet has traverse a significantly interesting and application-wise diverse journey. From being CSMA Ethernet in LANs, to switched Ethernet in the WANs, to Ethernet over wireless as in WiFi and WiMax, to its present avatar of Carrier Ethernet – the journey has led to some key innovations and applications. The scope of this project is to build an end-to-end Ethernet system that replicates using advances of Carrier Ethernet features that are contemporarily relegated to higher layers.

Funding Agencies: Ministry of Communications and Information Technology

The client is a world leader in MSPP business and has an intelligent ROADM platform. The current transport tool that the client uses takes a long duration to converge. The tool plots the network and creates a BOM from the network wide features. It also optimizes the network for routing, color assignment (wavelengths), card assignment and other physical layer features. The present version of the tool required severe re-design to improve performance. A patentable solution is proposed to this effect by our group. The solution makes use of a single branch-and-bound method to solve the optimization problem and guarantees quick convergence. The solution also takes into consideration PHY layer characteristics while proposing the end-to-end solution. Our proposed algorithm works in real-time.

Funding Agency: North American IP+Optical Transport vendor

Our group has strong interaction with providers in India, Europe and North America. In particular we are part of the design process of several brownfield and greenfield networks. Our focus has been on the WDM, metro, access and Carrier Ethernet segments.

The client is an Indian network systems integrator. The client would like to deploy a home-grown solution for the Indian market. We are in the process of developing a next generation MSTP that combines guaranteed transport and service delivery over a high-speed switching and routing core using optical interfaces. The technology makes use of some of the significant advances in service oriented communication – relying on telco and utility class service guaranteed platforms. The box is a standalone source of power and bandwidth – ideal for the utility, gas, telco and defense sectors.

Funding Agency: Commtel Networks

GEYSER's vision is to qualify optical infrastructure providers and network operators with a new architecture, to enhance their traditional business operations. Optical network infrastructure providers will compose logical infrastructures and rent them out to network operators; network operators will run cost-efficient, dynamic and purpose-specific networks by means of integrated control and management techniques. In the GEYSER concept, high-end IT resources at users' premises are fully integrated with the network services procedures, both at the infrastructure planning and connection provisioning phases.

Following this vision, GEYSER will specify and implement a novel optical network architecture able to support 'Optical Network + Any-IT' resource provisioning seamlessly and efficiently. Energy consumption metrics for the end-to-end service routing are part of this efficiency.

GEYSER is proposing to:

• Specify and develop mechanisms that allow infrastructure providers to partition their resources (optical network and/or IT), compose specific logical infrastructures and offer them as a service to network operators. This will be done overcoming the current limitations of networks/domain segmentation, and will support dynamic and on-demand changes in the logical infrastructures.
• Specify and develop a Network Control Plane for the optical infrastructure, by extending standard solutions (ASON/GMPLS and PCE), able to couple optical network connectivity and IT services automatically and efficiently, and provide them in 1 step, dynamically and on-demand, including infrastructure replanning mechanisms.
  

These achievements will enable infrastructure providers, network operators and application providers to participate in new business scenarios where complex services with complex attributes and strict bandwidth requirements can be offered economically and efficiently to users and applications. The GEYSER outcomes will be validated in an EU-wide optical network test-bed.

Funding Agency: European Commission Brussels

As the name suggests a new communication platform for high-speed communication systems in hostile environments is being developed. The platform uses some of the home-grown technologies that are based on highly reliable and available fabrics.

Funding Agency: Defense sector.

In this project the client requires us to perform a comparison between the PBB-TE and MPLS-TP flavors of Carrier Ethernet in core networks. The comparison was done through both analytical models as well as detailed simulations. An application level comparison was also done for data-center, video distribution, mobile-backhaul type applications in high-speed networks. The comparison gave a parameterized model that allowed a provider to deploy the appropriate technology.

Multi Protocol Label Switching (MPLS) is a 2.5 layer technology. MPLS provides protocol agnostic operation and makes routing simple. The benefits of 100Gbps serial transmission can be extended to MPLS by a very few changes in the existing architecture. Here, we investigate binary routing and extend the concept of binary routing to MPLS and do a complete optical switching along the entire Label Switch Path (LSP) such that within multiple LSPs that are established on the path from source to destination, the electrical conversion of the light takes place only where the paths are disjoint. This will extend the benefits of optical 100Gbps serial transmission in the domain of binary label switching. MPLS was further improved by incorporating changes that were felt during its fielding. This came to be known as T-MPLS. During this time a new architecture, which supported purely Ethernet for switching from source to destination using bridges (PBT) was standardized. This led to lot of debate as to which technology T-MPLS or PBT will play a crucial role from the future perspective. Further to our work on optical binary MPLS, we analyze both the technologies (T-MPLS, PBT) and investigate the differences between them in order to decide which technology is a better choice.

Funding Agency: IP Routing vendor

The client requires us to build a next generation layer-1-3 box in a single card low-channel count scenario. The target application is towards the rural areas in India. The single card box uses CWDM physical layer design that is supported by both a Carrier Ethernet based switching fabric and a deeply selective IP routing fabric.

Funding Agency: Tata Teleservices