Engineering Dynamic Scheduler for Work Manager

Abstract

Provision of service to business and residential customers, network maintenance and fault repair are core activities of large telecommunications companies which involve thousands of technicians every day. Managing a large-sized workforce efficiently in a highly dynamic environment is a challenge that BT has addressed by automating its work management processes within a single application known as Work Manager. For the most part, the success of Work Manager rests upon the ability of its work allocation component in determining the best assignments at any time and in any environment.This paper presents a scheduler for Work Manager which is both predictive and reactive and whose computational model reconciles genericity and efficiency. Coupling Constraint Satisfaction and Local Search, this scheduler can generate long-term schedules that satisfy the conflicting objectives of feasibility, productivity, quality of service and cost minimisation. It also provides the capability to adjust schedules before despatch through on-line heuristic repair/improvement. Equipped with a powerful interface, this system has been successfully trialled within the field and is now deployed nationally.