|
|
 |
| |
|
|
|
Finding the Optimal Shortest Path in Stochastic Networks Using the Markov Decision Process (MDP) |
|
|
|
PP: 2811- 2825 |
|
|
|
doi:10.18576/isl/120830
|
|
|
|
Author(s) |
|
|
|
D. Al-Tayar,
Z. Alisa,
|
|
|
|
Abstract |
|
|
| This work introduces new routing algorithms for stochastic networks. The problem addressed here considers multiple stochastic, time-dependent disruption levels on different links. This type of network makes routing decisions a challenging problem due to its stochastic nature. A novel Markov Decision Process (MDP) has been proposed and developed to handle the issue of multiple disruption levels (four levels were tested in this article). In addition, the developed approach has a hydride policy for the optimal path between two nodes depending on online and offline data and when to switch between them. A modelling framework that implements available online and offline network data and a novel cost structure that estimates the probable change in travel time resulting from expected disruption levels is delivered to get an optimal, reliable path. The proposed offline and online algorithms based on the suggested approaches are proven to efficiently handle the problem of stochastic network routing with multiple disruption levels. Results showed the findings demonstrated the efficacy of the suggested approach, contingent upon the incorporation of the Expected transition Cost (ETC) function into the primary computational equations. These equations were evolved from the calculation methodology employed for the protocols.
|
|
|
|
|
|
|
|
