Description:
After Tropical Storm Irene and the Halloweensnowstorm, everybody understands the need for rapid restoration of electric power. Optimal job scheduling is an NP-complete problem, which means, in ordinary English, that the only known solution is a full enumeration of all possible schedules. As near as we can tell, CL&P uses either a "First Come, First Served" (FCFS) policy or an "Outside In" policy for scheduling their crews. FCFS means that the jobs are scheduled in the order that they're called in, and "Outside In" means that crews are sent to the borders of affected areas and they then work their way in to the center of an affected area. This last method is equivalent to what's called the "Nearest Neighbor" algorithm, which is equivalent to "Shortest Travel Time First" scheduling. The authors wondered whether a scheduling algorithm known as "Longest Remaining Job First" (LRJF) might produce better results. LRJF is a "near-optimal" algorithm, apparently discovered by Prof. Todd in the early 90s, that when used for scheduling jobs for parallel processing, results in faster job completion times.