Priority Based Cloud Scheduling Using Analytical Hierarchy Process with Multi-Layer Perceptron

Abstract

Priority of jobs is an important issue in scheduling because some jobs should be serviced earlier than other those jobs can’t stay for a long time in a system. Assigning priorities also helps to preempt task in the middle of execution. Since priorities often vary according to users with respect to various job scheduling parameters in cloud such as bandwidth, completion time, memory, etc., often a model that makes decision to schedule with respect to multi-criteria is relevant. Analytical Hierarchy Process (AHP) is a multi-criteria, multi-attribute decision making model that combines the criteria weights and the options scores, thus determining a global score for each option, and a consequent ranking. Using AHP, the decision making scheduling problem can be decomposed into hierarchy of comprehended sub problems, each of which can be analyzed independently. However, different users can have conflicting preferences. So, there might be inconsistent elements in the reciprocal pairwise comparison matrix created on the process of solving scheduling problem by using AHP. In order to solve matrix inconsistency problem as well as to tackle the situation of missing entries in the comparison matrix which must be filled in accordance to decision maker’s judgments, a MultiLayer

Perceptron neural network model can be used to handle such scenarios after training them. Final result is the priority based schedule of tasks that can be scheduled in the datacenter with fixed resources which is typically suggested by neural network in course of deriving a PVS vector.

Categories and Subject Descriptors:

• Networks~Cloud computing • Computing methodologies~Neural networks

Keywords:

Cloud Scheduling, Analytical Hierarchy Process, Neural Networks, Multi Layer Perceptron