Efficient calculation of reliability and performability of large computer networks

Abstract

This thesis introduces two algorithms for the calculation of reliability and performability metrics of computer networks. Both have several key advantages over existing approaches. Firstly, they have greatly decreased memory usage, including constant space complexity and linear time complexity for families of networks with identical inter-connectivity of devices. Further, they are the first algorithms to address the Expected Hop Count problem, and can be efficiently applied to a wide range of network models.