Self-Organizing Fair Job Scheduling among Mobile Devices

Karin Anna Hummel, Harald Meyer
2008 2008 Second IEEE International Conference on Self-Adaptive and Self-Organizing Systems Workshops  
Ubiquitously available mobile devices can contribute to grids not only for accessing resources but also to provide resources, such as, computational power or memory in mobile scenarios. When utilizing networked mobile resources challenges arise due to, e.g., disconnections, disturbances on the wireless medium, and limited energy sources. We propose an opportunistic job scheduling approach to harness cycles. Mobile nodes decide autonomously which job to take by matching the job's requirements
more » ... inst their capabilities and coordinate among each other based on shared job queues. Reactive and proactive fault tolerance mechanisms assure robustness. The main contribution of this work is the introduction of five different fairness strategies for self-organizing balanced load distribution based on gossiping about the other mobile nodes' states and comparing the own status against this information. In simulation runs, we achieved an improvement of fair load distribution which remained robust even in the presence of faults. Related Work Distributed job scheduling among mobile clusters and grids has recently attracted several research efforts, addressing challenges like energy efficiency [22] and connection failures. Disconnections of wireless links can be masked from applications as proposed for reliable sockets in [21] on This paper is a preprint of the published work and restricted to personal use only.
doi:10.1109/sasow.2008.58 dblp:conf/saso/HummelM08 fatcat:a7fag4jjmnadta5i5s3wdacm6y