In Communication Networks, Survivability is an important factor to be considered in planning and designing of Survivable Fiber Optic Networks (SFON). Survivabilitydepicts network protection and restoration with respect to span failure, node failure, or even Shared Risk Link Group [SRLG] failure [1-3]. [4-8].
1.1 FIBER NETWORK ARCHITECTURES
Networks survivability resists any interruption or disturbance of a service, particularly by warfare, fire, earthquake, harmful radiation, or other physical or natural catastrophes rather than by electromagnetic interference or crosstalk [9-10].
In the world of fiber technology, bandwidth may not be a constraint. An architecture that uses facility hubbing can best utilize the economical factor of high-capacity fiber systems and reduce the amount of equipment needed for signal transport. As a result, reasonable network architecture and routing strategy is to send all the demands from each office to a central point or hub. Thus demand is aggregated into the largest possible bundle to take advantage of today's fiber technology economically [11-13]. In this architecture each Central Office (CO) is connected to a hub via a fiber-optic system. At the hub, a Digital Cross Connectivity System (DCS) partitions incoming traffic to different destinations and route channels, to the appropriate end office.
1.2 THE SIGNIFICANCE OF CONSIDERING SURVIVABILITY:
The increasing deployment of interoffice optical fiber transmission systems with large cross-sections supported on a few strands of fiber-hubbed network architecture have increased concern about the survivability of fiber communications networks. Service disruption causes both tangible and intangible loss for users as well as for service providers [14-15]. Network failures can be attributed to hardware or software problems or to natural catastrophes.
1.3 SERVICE SURVIVABILITY PLANNING
Service survivability planning involves challenges, opportunities, and regulatory realities. The planning can be categorized into 4 phases Fig.1.1 to ensure service continuity and minimize the level of impact caused by service disruption [16-17]. They are prevention, prompt detection, network self-healing through a robust design and manual restoration.
The first phase focuses on preventing network failures. The second phase focuses on quick detection of network component failures. The third phase focuses on the network self-healing capability during network component failures . The last one focuses on planning and practicing restoration in terms of efficient utilization of available work forces, facilities and the equipment .
1.4 SURVIVABILITY DESIGN CONCEPTS:
To implement survivability, restoration techniques are designed to make active use of available capacity. It also automatically restores when maintainance service fails. These techniques fall into two categories - Traffic Restoration and Facility Restoration. Traffic restoration is applied to switched networks, where as facility restoration is applied to facility transport networks [20-21].
1.4.1 TRAFFIC RESTORATION
It involves routing and individual calls around a failure. A circuit switch performs traffic restoration by routing calls around failed circuits. Other techniques that can perform traffic restoration including Dynamic Non-Hierarchical Routing (DNHR) and state-dependent routing which reroutes not only traffic from failed points, but also efficiently utilize network bandwidth.
1.4.2 FACILITY RESTORATION
It involves in rerouting transmission, in large units around a failure. It requires fewer operations than rerouting each call individually thus it has the potential to restore more services in a shorter time than traffic restoration. For current, high-capacity asynchronous fiber facility networks, an efficient and commonly used transport signal unit is Digital Signal Level (DS3), which carries 45 Mbps of data, rather than Digital Signal Level 0 (DS0), which carries a voice call of 64kbps. This facility restoration is more appropriate than traffic restoration for fiber facility transport systems. Technological advancements play a crucial role in implementing survivable fiber networks .
A network that supports high capacity traffic for telecommunications must be designed in a way that makes it robust to the potential damage from unforeseen events like a cut in a link or a breakdown of some network equipment. A network having survivability is capable of satisfying the demand for the point-to-point services expected of it despite the potential for such disruptive events [23-25]. This is achieved when the network is planned with sufficient extra capacity, is multi-connected, and has the ability to re-route traffic immediately, and if necessary, to avoid any failed network locations. For telecommunication networks a planner has a number of options, e.g., using equipment and architectures based on the Synchronous Optical Network (SONET) and Wavelength Division Multiplexing (WDM) technologies, for configuring a network with sufficient capacity and traffic-switching capability to provide effective and efficient survivability. In these networks when the equipment failure is detected, the (pre-designed) plan for restoring service is automatically and quickly implemented.
1.5 SURVIVABLE NETWORK ARCHITECTURE FOR RESTORATION
In many Operating Telephone Company (OTC) environments, fiber hubbing architecture is an economically attractive, alternative to the current metallic mesh architecture and to expand point-to-point fiber transmissions. It is fiber efficient and robust in a rapidly growing environment. A three-level hubbing architecture is assumed in the Intra-LATA network architecture namely central offices (CO), hubs and gateways. The gateway is also a hub. This three-level hierarchy is assumed because it has worked well for LATA network traffic loads. Each CO is identified as either a special CO or not [26-28]. Such CO's selected by OTC's are given special treatment for failure conditions. A group of CO's served by the same hub is called a cluster, and a group of clusters served by the same gateway is called a sector. Gateways are fully connected to each other by fiber systems.
In order to aggregate demand from a CO the fiber utilization is maximized. All the demand from the CO is multiplexed on a fiber span having terminals in the CO and the hub. Each span may include one or more links in the network topology. At the hub, point-to-point demand supports one or more DS3's. It is cross- connected on a DS3 basis to the multiplex span destined for the proper CO, i.e., not via the hub DCS 3/1. Hub-to-hub DS3 demands are carried by multiplex spans on an economic basis will be explained later. Fig.1.2 shows a diagram of the above multiplex span construction and demand aggregation within the hub [29-32].
1.5.1 PROTECTION SWITCHING
Survivable structures considered in this study include protection switching connectivity. The protection switching approach is commonly used to facilitate maintenance and protect working services, and has the advantage of being totally automatic. The 1:N diverse protection structure is an alternative to the commonly used 1: N protection strategy where working fiber systems share one common protection fiber system. The only difference is the location of the fiber protection system. The 1:N protection structure places the protection fiber in the same route as that of working systems, while the 1:N diverse protection structure places the protection fiber in a diverse route [33-37]. A 1:1 diverse protection arrangement, which provides 100-percent survivability for fiber cable cuts requires a less sophisticated Automatic Protection Switch (APS) than the 1:N diverse protection scheme.
1.5.2 DUAL HOMING
In contrast to the single homing approach commonly used in fiber-hubbed networks aggregating demands from any CO to their destinations through an associated home hub, dual homing is used. It gives a concept of demand balancing that splits demand originating from a special CO between two hubs namely a home hub and a designated foreign hub. Dual homing does not automatically accomplish restoration by itself, but must be used in conjunction with path rearrangement capabilities [37-45]. The dual homing approach guarantees surviving connectivity, but it may take time to restore priority circuits via path rearrangement. On the other hand, dual homing provides protection against hub and DCS disasters. Several options for implementing a multiple span layout for a dual homing architecture have been studied. The most advantageous option found is to build a diverse working span directly connecting a special CO to a designated foreign hub, in addition to a working span connecting it to its home hub.
1.5.3 PATH REARRANGEMENT
The foregoing path rearrangement structures protect priority circuits against fiber cable cuts and other fiber system failures. To protect against DS3 and DS1 level failures in CO's and hubs, each of these structures also need to support path rearrangement. This is partly accomplished by providing additional standby DS3 paths between CO's and hubs and among hubs. Hence by providing one standby DS3 path for each CO-to-hub combination and two standby DS3 paths for each hub-to-hub pair path rearrangement is enhanced [46-52]. In case self-healing architectures are not used, more standby DS3 paths may be desirable to defend against fiber cable cuts. Note that path rearrangement uses DCS 3/1 capability.
1.5.4 SELF-HEALING RING
The self-healing ring, like the 1:1 diverse protection structure, is totally automatic and provides 100-percent restoration capability for fiber cable cuts. It can also provide some survivability for hub DCS failures and major hub failures (.e.g., flooding or fires).
1.6. OPTICAL Ad-Hoc NETWORKS
The explosive growth of the Internet and convergence of optical communication and data networking have jump-started several emerging multihop optical networking technologies. In this environment infrastructures such as optical layout and centralized base stations (control units) are available for networking support. A Large numbers of networking devices are allowed to communicate with one another over the shared medium in an ad-hoc optical network.
Optical ad-hoc network offer convenient, infrastructure-free data communication services to optical users. So researchers have developed numerous resource management algorithms and protocols, e.g., QoS-oriented MAC layer design, packet scheduling and mobility management for effective operation [53-59].
The problem of fair-packet scheduling in a shared-medium, multihop optical network has remained largely unaddressed. Fairness is critical to ensure that well-behaved users are not penalized because of the excessive resource demands of aggressive users. In order to solve this problem fair queuing is adopted. This fair queuing is implemented by using different types of algorithms to resolve the maximum throughput.
The significance of optical ad-hoc networks lies with Optical Channel Capacity, Optical Channel resource Sharing-Spatial locality, Scalability and node mobility Fairness Multi-hop Optical Network, thus maximizing the channel utilization, & QoS.
1.6.1 ANALYSIS OF SURVIVABILITY PERFORMANCE IN OPTICAL
The previous work involves protection and restoration in optical networks with arbitrary mesh topologies. A number of distinct efforts were made in this area. A novel network protection method that can handle both cable cuts and switching equipment failures is developed. The process that is fast, autonomous and distributed. This also restores the network in real time, without relying on a central manager or a centralized database [60-65]. It is also independent of the topology and the connection state of the network at the time of the failure.
While the work focused on optical networks, the methods developed are not network specific and can be applied to many types of networks employing a variety of transmission and switching technologies [66-70].
1. Develop models of optical-broadband access networks and trunk networks based on projected traffic growth.
2. Evaluate the impact of emerging technologies on network architecture design.
3. Develop routing algorithms for optical layered networks.
4. Investigates protection/restoration coordination schemes in the optical layer, i.e. physical layer topology.
5. Investigate the potential for packet switching procedures and burst switching in optical networks, i.e. Logical Layer Topology.
The Performance and Evaluation of Optical networks take into consideration the factors like trade-off between routing traffic at the optical layer, creating dedicated light paths in order to maximize the traffic carried and the availability of spare capacity.
The computational complexity and effectiveness of a concept was dealt in the previous work viz “N-hub Shortest- Path Routing” in optical networks. This allows the routing domain to determine up to N intermediate nodes (“hubs”) through which a packet will traverse before reaching its final destination [71-75].
The dynamic routing algorithm concept is introduced in the previous work deals with the major activity of how to design the virtual topology of light paths through a given physical network topology. A light path is an optical channel that connects two routers in the network. A light path can traverse several physical links and Optical Cross-Connectors (OXC's), thus reducing the amount of routing that is performed on each physical link.
The Previous work has also proposed routing in the logical layer based on the hop-by-hop shortest path paradigm. The source of a packet specifies the address of the destination. Central Office (CO) and each router along the route forwards the packet to a neighbour located “closest” to the destination and also proposed the shortest path for each pair of nodes, and therefore minimizes the bandwidth consumed by every packet in the optical adhoc networks scenario [75-82].
The present work proposes a few techniques, which are unique as they try to combine the best features of both path and line-based approaches by means of an integrated approach. The approach is comparable to shared-mesh path-based restoration in terms of restoration capacity utilization and multiple failure restorability. In addition, it is much faster and uses a significantly lower number of messages than the path-based method during the restoration process.
1.7 SCOPE AND PLAN OF THE PRESENT THESIS:
The present work represents the Optical Layer in to two subsequent layers. The first one is Physical Layer and the second is Logical Layer. It is reconfigured for N X N node connectivity. It can be noted that inter-working between layers (represented by optical and user's equipment in this case) also makes use of the bottom-up approach since the upper layer(s) need to be enhanced with new functionality concept.
Different network architectures are presented in Physical Layer. Chapter II presents Fiber Span Layout analysis of demand distribution by using Point-to-Point Span architecture. This architecture is very significant from routing optimization point of view. It establishes the related simulations and emphatically account for single node connectivity to multi node connectivity. In this different connectivity pattern parameters like link connectivity, node connectivity and digital cross connectivity are measured. It enables fruitful implementation of physical layer diversity. The fiber network design concepts are implemented by using Central Office (CO), hubs and gateways. Two types of fiber spans, namely Point-to-Point Span and Hubbing Span are generally used in fiber network design. It enhances the dynamic route computation mechanism. In this the total routing path is utilized in order to compute the link weights in demand distribution. Improved span connecitivities under multimode configuration have been evaluated for different node connectivities. An effective Fiber Least Shortest Path (FLSP) algorithm has been proposed to evaluate the bit rate parameters/link/network/demand connectivities.It provides necessary proof and accountability to assure normal operation mode in complex networks with multiple connectivities, with enhancement link estimates and performance characteristics.
Chapter III deals with Fiber Network User Service Survivability (FNUSS) simulations of Survivable Protection Switching System (SPSS). In this point-to-point span architecture is extended further to compute demand distribution in terms of demand routing, multiplexing and restoral schemes. The different restoration schemes like Automatic Protection Switching (APS), 1:1, 1+1 and Diverse Protection (DP) are used for recovery from network failures and maintaining the required existing services from a user perceptive point of view. These schemes are used widely to improve the network connectivities and develop it further in terms of time-scale of operations and resource efficiency etc. The average survivability is estimated by using failure probability of each network component and the average restoration time. The service survivability is also measured by enhancing internetworking with internet protocol (IP) routing and resource management protocols. Thus FNUSS algorithm evaluates different multimode configurations and supports integrated restoration topology bridging both primary and backup paths.
Chapter IV describes Optical Network Demand Bundling Using DS3-Forming and establishes the relationship between facility hubbing and diversity techniques. It is further extended to demand bundling technique. The Optical Network Demand Bundling using DS3-Forming is to implement the Optical Network System (ONS) from a single period demand bundling to multi period demand bundling. It depicts the routing analysis in two different paths, direct path and indirect path. The direct path denotes a digital signal at an intermediate office where as the indirect path consists of two or more digital signals at an intermediate hub location. It provides affordability in terms of network planners with flexible demand requirement. The indirect path is further merged into different parcel lists. It combines point-to-point links into appropriate digital signal demands which are commonly used as input to fiber systems in today's interoffice fiber networks. The end-to-end demand bundling is also achieved by link-by-link bundling process. In this the traffic demand can be rerouted through a physical topology or server topology. Thus bundling optimization is also achieved.
Chapter V, describes Synchronous Optical Network (SONET) which is an integrated approach of the Fiber Span Layout Demand Distribution, Fiber Network User Service Survivability and Optical Network Demand Bundling Using Digital Signal 3 - Forming presented in the previous three chapters. In this the different methodologies like Point-to-Point Architecture with Diverse Protection and Ring Architecture are presented. It is further extended to include multi network demands, by using Multiperiod Synchronous Optical Network Survivability (MSONS) algorithm. It estimates different network connectivities, corresponding signal level transformations and end-to-end multi year demands. It uses different network architectures like Point-to-Point/Hubbing span, APS, Self-Healing Ring (SHR) and reconfigurable DCS mesh network. The planning model objective is to minimize the economical impact of network development over N years, while ensuring that sufficient fibers and equipment are installed in the network to accommodate the growth demand. Thus capacity expansion can be achieved by using SONET with SHR combination. The line rate option over the interval is obtained and the demands are inserted into the Q. The node computations in Q are then sorted in increasing order and hence multi period demand path propagation is achieved.
The Physical Layer techniques are further expanded to Logic Layer in order to achieve the maximum channel utilization and global fairness model by using optical ad-hoc network methodologies.
In Chapter VI Two-Tier Algorithm is introduced and it guarantees the packet scheduling in optical ad-hoc network design issues. It provides a single physical channel C for multipath propagation of packets by means of transmission flow viz. slot queue and packet queue. Thus fair queuing is achieved interms of an efficient, scalable and localized manner and broadband connectivity in ad-hoc network architecture analysis. The parameters like Weighted Graph (WG), Weighted List (WL), local fairness model, packatized fair queuing and flow information propagation are used in order to avoid consequent collisions to obtain location-dependent contention. Thus maximum speed of ad-hoc network connectivity results in different phases depending on their location and packet delivery procedures. Also different applications like Quality of Service (QoS), rate-sensitivity, delay-sensitivity are presented. It provides a minimum fair allocation of the channel bandwidth for each packet flow and maximizes spatial reuse of bandwidth by using centralized packet scheduling algorithm. It supports effective communication intensive applications like web browsing, video conferencing, remote transfer and etc…
Chapter VII describes BFMLM-FQ Algorithm to determine the concept of node mobility and scalability. It describes the multihop flow propagation which is divided into a number of single hops and thus global topology independent fairness model is achieved. In this fair queuing flow achievement and flow information propagation has been used. Also the stastical short term throughput and fair distribution of bandwidth is achieved. It retains the distributed fair queuing in multi hop network connectivity. The fair share of each packet flow is defined with respect to the corresponding flow contending graph. Emerging applications for the ad-hoc networking technology proves the effective packet scheduling in optical ad-hoc networks. Thus maximum throughput rate is achieved.
Finally in Chapter VIII Hybrid Algorithm is presented which is an integrated approach of Two-Tier Algorithm and BFMLM-FQ Algorithms thus achieving the throughput of global fairness model. In this local fairness and fluid fairness are achieved. Fluid fairness model ensured local fairness in the time domain and global fairness in frequency domain. This model also achieves fair bandwidth sharing with effective throughput. The extensive simulations confirmed the effectiveness of self-coordinating localized design in providing global fair channel access. Thus higher aggregate throughput and higher spatial reuse in optical ad-hoc networks is achieved.
Numerical results have been evaluated for the Physical Layer and Logical Layer methodologies with simplification procedures mentioned above and are implemented in C Language and by using ns2 simulator. In Fiber span Layout Demand Distribution, the DCS factor for different network connectivities like 1 X 5, 3 X 3, 5 X5 and 9 X 9 are measured and obtained which is about 80% as against 20% in the work reported earlier. In Fiber Network User Service Survivability, the demand connectivity factor for different network connectivities like 1 X 5, 3 X 3,
5 X5 and 8X8 are simulated and it is about 86%, where as the same is 20% in the work reported earlier. In Optical Network Demand Bundling Algorithm using DS3 - Forming computed the multi period connectivity by using demand distribution routing mechanisms such as direct and indirect path as compared to single period connectivity reported earlier. In the Synchronous Optical Network (SONET) module the growth of demands for multi period survivability planning periods upto Nth year are considered as compared to the earlier work for one single year. In Two-Tier Algorithm maximum co-ordination of fair queuing in ad-hoc networks achieved is 95% for multi-hop network connectivity, as compared to 25% of the previous work. The BFMLM-FQ determines the fairness model by using node mobility and scalability and throughput is 92% for multihop connectivity as compared to 15% of the previous work. The Hybrid Algorithm results in an integrated approach of local and global fairness models with maximize channel reutilization of 83.9% and 98.2%. Even though the results are given in a consolidated form here they are included in the respective chapters in the text.
Finally conclusions and scope of future work is discussed.