Resource allocation and scheduling within the context of fibre network deployment
Abstract
The Resource-constrained Project Scheduling Problem (RCPSP), where a schedule must obey the resource constraints and precedence constraints of various activities over time, is one of the most studied scheduling problems. During the scheduling process, each activity requires a quantity of some resource for each period. The total resource consumption for each of the time periods must be less than or equal to the availability of resources. A precedence graph determines the order in which the activities may be scheduled. Existing project management tools such as MS Project do not take complex objective functions into account and are unable to cater for telecommunication-specific side constraints. A fibre network deployment scheduling model is proposed to assist with the resource allocation and scheduling of project activities. The proposed model will take the time value of money into account and will perform resource allocation and scheduling with the objective of maximising Net Present Value (NPV). In this dissertation, two Mixed-integer Programming (MIP) formulations of the RCPSP are presented, the time-indexed and resource flow formulation. An experimental comparison of instances involving varying activity duration sizes is performed. The impact of these problem characteristics on the performance of the models is evaluated when considering the minimisation of makespan as the scheduling objective. The results from the minimisation of makespan models are used to solve the scheduling of fibre deployment activities with the objective of maximising NPV. Computational results of the formulations presented in this dissertation are compared using datasets from the literature as well as generated datasets. Conclusions of each model are drawn according to the instance characteristics. Based on the results it is found that the resource flow formulation performed better than the time-indexed formulation when the duration of project activities increased. The hybrid approach improves the MIP models by using Constraint Programming (CP) as a primal heuristic to determine initial feasible solutions for the MIP models to increase NPV.
Collections
- Engineering [1403]