Discrete Network Design (DND) is a classical combinatorial problem in transportation planning. According to its exponential nature, the problem has a costly exact solution. In recent decades, however, the advent of parallel computing facilities motivated researchers to apply exact solutions on many combinatorial problems. As a result, parallel Branch-and-Bound (B&B) algorithms emerged as robust methods to address combinatorial problems. This dissertation will investigate the parallel behavior of B&B algorithms in DND through some case studies. Parallelization results suggest that parallel B&B algorithms can achieve linear speed-ups in DND for a large number of processors
Discrete Network Design (DND) is a classical combinatorial problem in transportation planning. According to its exponential nature, the problem has a costly exact solution. In recent decades, however, the advent of parallel computing facilities motivated researchers to apply exact solutions on many combinatorial problems. As a result, parallel Branch-and-Bound (B&B) algorithms emerged as robust methods to address combinatorial problems. This dissertation will investigate the parallel behavior of B&B algorithms in DND through some case studies. Parallelization results suggest that parallel B&B algorithms can achieve linear speed-ups in DND for a large number of processors