A Multi-objective Optimization for Laying Optical Fiber Cables

Summary form only given. There have been many research publications on the topic of optical network survivability where the networks is modeled as a graph with a given topology composed of nodes and links. Such modeling does not enable the consideration and optimization of the shape of cables. One glaring example is the 2006 Hengchun earthquake that occurred on December 26 near the south coast of Taiwan and caused catastrophic disruption to Internet services in many Asian countries. This catastrophe was caused as a result of damage to several undersea cables that were located all close to each other in an earthquake prone area. Clearly it makes sense for cables to avoid earthquake prone areas and if it is not possible, they should be kept away from each other in such areas to avoid simultaneous destruction of multitude of cables. Research conducted in 2005 in ETH, Switzerland, has shown that a week of Internet shutdown, in a country like Switzerland, will cost over 1% of its GDP. In addition to consideration of network survivability with the aim to continue to provide services at accepted quality even when disaster strikes, optimizing cable shapes should consider many other factors. Clearly, cost is an important factor, and to save cost, cable length should be minimized. However, laying cables in straight lines between nodes is often neither optimal nor feasible, as many other types of areas should be avoided. These include the following examples: ecologically critical areas, politically sensitive regions including areas where it is difficult to obtain licenses, areas where the topography is characterized by steep slopes and areas with sharp rocks, shipping areas where anchors can damage cables, shark infested areas, and wind turbine farms. The problem of how to minimize cost of a cabling system and meet required survivability measures is the topic of our Collaborative Research Fund Project supported by the Research Grants Council of Hong Kong. This presentation will report the status of this project where we integrate data on topography, ground motion and cable response to earthquake effects to provide methodologies for laying cables considering cable shape, location, and capacity. In our project, we develop a network survivability model involving a multi-objective optimization over manifolds, leading to Pareto front results that trade off cost versus network failure risk. This multi-objective approach is justifiable because of the multiple stakeholders (e.g., cable owners, governments, and insurance companies) which may have different and possibly conflicting interests and objectives. We will present results based on multi-objective cable laying optimization taking account of the above mentioned considerations using theoretical models as well as real world data.