Discussion of a Master's Thesis

The master's thesis for the student (Thaer Habib Nasser) with a major in (Computer Engineering) was discussed on Sunday 16/10/2022 at the discussion room (Hall No. 9) in the Department of Control and Systems Engineering. The thesis title was

“Performance Evaluation of Multi-Radio Wireless Communication System”.

The discussion committee consisted of:

Prof. Dr. Nadia Adnan Shiltagh / Chairman Asst.

Prof. Dr. Ahmed Raouf Nasser / Member Lect.

Dr. Laila Hatem Abboud / Member Asst.

Prof. Dr. Ikhlas Kadhem Hamza / Member and Supervisor

In heterogeneous computing systems, work scheduling effectiveness is crucial for high-performance software execution. The programs are to be considered as multiple sequences of tasks that are presented as directed acyclic graphs (DAG). The problem of the current work is divided into two parts. The first part shows that increasing broadcast and redundancy lead to an increase in time consumption. The second problem is solving complexity problems when tasks are scheduled in a heterogeneous manner in a computing system, where the processors in the network may not be identical and take different time periods to carry out the same task. The goals of this work are to reduce the total cost of network-wide broadcasting to minimize the search space and to solve the complexity problem when tasks are scheduled in a heterogeneous way in the computing system. The objectives of this research include, firstly, the implementation of the MOCAB algorithm to find the shortest path between nodes (the least expensive) and calculate the alternative paths. Second, the HEFT algorithm is used to calculate the actual finish time (AFT), earliest start time (EST), and earliest finish time (EFT). Third, performance is evaluated by throughput and response time. Finally, the results were improved by 37% using the genetic algorithm. The performance of the MOCAB algorithm was evaluated with that of the HEFT algorithm in terms of the delivery ratio of packets delivered from the source node to the target node in the network. The results showed that the MOCAB algorithm outperformed the HEFT algorithm by 8%. On this occasion, we congratulate the student (Thaer Habib Nasser) for obtaining the master's degree, and we wish him continued success.