Thesis

Genetic algorithms for large resource allocation problems

Downloadable Content

Download PDF
Creator
Rights statement
Awarding institution
  • University of Strathclyde
Date of award
  • 1995
Thesis identifier
  • T8316
Qualification Level
Qualification Name
Department, School or Faculty
Abstract
  • Telecommunications management systems must undergo some radical transformations in order to support the next generation of giant gigabit networks. In particular, Artificial Intelligence (AI) will be necessary to automate some aspects of network management. This thesis concentrates on one particular aspect of management, namely resource management. A large number of telecommunications applications can be viewed as resource allocation problems and the telecommunications operators require efficient techniques to solve these problems. However, due to the complex nature of these scheduling problems, large instances cannot be tackled easily by conventional techniques such as the deterministic and analytical procedures provided by Operational Research. Alternative approaches such as the ones proposed by AI have to be considered. This thesis examines genetic algorithms (GAs) - one of these AI techniques - applied to vehicle routing problems with time-windows and technological constraints - a special class of resource allocation problem. In particular, this thesis proposes a new class of GAs. By working directly on the schedules rather than using an indirect representation, these Gas overcome some important limitations shown by traditional GA-based systems. These direct GAs are examined over a wide range of problems. In a first set of experiments, the direct GAs totally dominate the traditional GA-based systems. In a second study, which compares the performances of these new GAs with the performances of random search, hill-climbing, simulated annealing and tabu search, it is shown that there is no universally best algorithm. However, under certain conditions some algorithms should be preferred to others. In particular, when time is available, the direct GAs should be chosen for the solution of under resourced problems.
Advisor / supervisor
  • Smith, D. G. (D. Geoffrey)
  • Prosser, Patrick
  • Magill, E. (Evan)
Resource Type
DOI
EThOS ID
  • 881587

Relations

Items

Thumbnail Title Date Uploaded Visibility Actions
file details PDF of thesis T8316A 2026-03-31 Public Download
file details PDF of thesis T8316B 2026-03-31 Public Download