Advanced scheduling in data transmission and mobile media cloud computing

• Main Author: Yang, Ming
• Other Authors: Cai Jianfei
• Format: Theses and Dissertations
• Language: English
• Published: 2015
• Subjects: DRNTU::Engineering::Computer science and engineering::Information systems::Information systems applications
• Online Access: https://hdl.handle.net/10356/62259
• Institution: Nanyang Technological University

In general, scheduling can be considered as the way to efficiently arrange the usage of some shared resource (e.g. communications bandwidth and computing time) to complete multiple tasks or sub-tasks. In this thesis, three types of scheduling are investigated, including the data transmission scheduling in wireless MAC protocol, the CPU scheduling for media computing on mobile devices and the dynamic job scheduling in cloud system. In the data transmission scheduling, we study how to efficiently schedule a large amount of data transmission jobs in the resource constrained scenarios, particularly in the long-RTT (round-trip time) and low-bandwidth networks scenario. The MAC protocol of underwater networks is chosen as a study case to investigate how to overcome the limited resource. A new underwater MAC protocol is proposed where the design tackles the costly protocol handshakes by reducing the protocol handshake overhead and increasing the data transmission opportunities. An efficient dynamic polling-based handshake operation is designed which offers time-bounded collision-free channel assignment. Through extensive simulations, the results show the proposed HCFMA protocol outperforms its counterparts, in terms of throughput and delay, such as ALOHA with carrier sensing (ALOHA-CS) and slotted floor acquisition multiple access (SFAMA). Then we study the energy efficient scheduling for media computing on mobile devices. Specifically we take video encoding, a CPU intensive application, as the study case. The CPU running frequency on video encoding process is scheduled to minimize the total expected energy consumption, while meeting the requirements of quality-of-service (QoS). A probabilistic QoS model is adopted, in which the encoding process should complete with a target probability within a specified delay deadline for each GOP (Group of Pictures). The optimization problem is solved analytically and closed-form solutions are obtained for both the optimal clock frequency scheduling and the minimum energy consumption. The numerical results suggest that significant amount of energy can be saved by using our optimal solution. Finally, we focus on the job scheduling for media computing in cloud system. Taking video transcoding as the study case, a dynamical video configuration scheduler is proposed to minimize QoS degradation while satisfying the criteria of job completion delay. The trade-off between job completion delay and QoS of video transcoding is explored in cloud system. The Lyapunov optimization framework is applied as the problem solver for the scheduler to adaptively set up the video transcoding configurations. Compared to the static configuration strategy, the proposed framework obtains smooth transcoding QoS degradation when system load becomes heavier and also achieves better delay performance. In conclusion, in this thesis we study the advanced scheduling for wireless data transmission, particularly in efficient MAC protocol design of underwater acoustic network. We also study the scheduling for cloud media computing, particularly in energy efficient video coding on mobile device and optimized video transcoding job scheduling on cloud system.