An ant colony approach to regenerator placement with fault tolerance in optical networks

This paper introduces an ant colony optimisation algorithm for the optimal placement of regenerators in optical networks. The algorithm guarantees survivability of the underlying network whilst also minimising the number of regenerators required. These are vital requirements in optical networks as the cost of regeneration equipment is very high. However, minimisation of the number of regenerator sites must be achieved whilst maintaining alternative lightpaths in the event of equipment failure. The latter constraint is one which does not appear to have been adequately addressed by research literature to date. Weaknesses in other published formulations of this problem are analysed and shown not to guarantee fault tolerance in all circumstances or to be too restrictive in establishing the existence of alternative paths. Following this, a new model is introduced in the form of a minimum k-connected k-dominating set problem with shared risk resource groups. A greedy algorithm is then introduced and modified into an ant colony optimisation approach. The two algorithms are tested against one another and the ant colony heuristic is seen to produce superior results with reasonable runtimes.