۱- Introduction

۲- Related work

۳- Problem Formulation and proposed algorithm

۴- Simulation Results and Discussion

۵- Conclusions

References

Abstract

To meet the increasing user bandwidth demands, the ICT networks are constantly expanding. The optical fiber technology has completely revolutionized the bandwidth capacity of both the core networks and the access networks. In core networks, the optical links provide very high bandwidth connectivity over long distance. Thus, any link failure due to disastrous events like earthquake, flood, landslide etc. can lead to massive service outages and huge fiscal losses. Normally, optical fibers are laid in 1 + ۱ configuration to route the traffic to the alternate path in such scenario. However, a natural disaster event may lead to simultaneous failure of multiple lightpaths. Therefore, routing algorithm, running on network nodes (routers or switches), are required in such case to establish new routes. Nevertheless, generally the routing schemes follow the least hop count and shortest distance approach to route the traffic to another backup path. Nonetheless, this approach may result in congestion on some links while other links may have unutilized capacity. This also makes it progressively tougher to fit more connection requests from the access network. Hence, implementation of more advanced path computation capabilities is required at the network nodes of the core network to ensure efficient routing of network traffic in disastrous scenario. This problem is referred to as capacity-bounded lightpath (CBL) problem. We proposed an exact algorithm which addresses this problem by considering the channel capacity of each link in addition to distances. The performance of the proposed algorithm is evaluated through simulation for three parameters: link capacity, connection requests and un-used links. It is revealed that existing shortest path algorithm improves the performance in terms of blocking probability of links and lightpaths at the cost of underutilization of the network capacity. Whereas, proposed algorithm regulates the capacity utilization by prioritizing link capacity over link length to establish more optimal shortest lightpath against connection requests.

Introduction

The services and resources delivered by information and communication technology (ICT) networks have revolutionized the world into a global village. For most of the world population, the Internet access has become a necessity rather than a comfort. In ICT network, the role of optical fiber has significantly increased and it has been estimated that about 99% of the global internet traffic is carried by undersea fiber optic cables [1,2]. The dramatic increase in the demand for ICT services has up surged the bandwidth requirements. It is investigated that bandwidth requirement of the users is rising faster than the network capacity to deliver it [3]. According to Cisco forecast 2017, internet traffic will grow threefold over the next five years, whereas more than 63% of internet traffic will be caused by wireless and mobile devices data in 2021 [4]. According to ITU 2016 report, 95% of the world population lives in the area that is covered by cellular networks whereas mobile broadband networks (3G or above) covered area encompasses 85% of the global population [5]. This ever-increasing bandwidth requirement ultimately contributes towards overloading of network resources or network congestion. It also means available network capacity cannot fulfill the total connection requests in the network. This may happen for several reasons like low bandwidth, multicasting, bad configuration, too many hosts in the broadcast domain, broadcast storm (can be a busy day for e-commerce or Black Friday sales) or the re-routing of disrupted or blocked traffic due to disasterbased failures. Such situations lead to network congestion which might also cause service outages, data losses, service downtimes as well as financial losses to network operators. It has been estimated that the losses due to service downtime may range from 25 to 150 thousand USD per hour [6].