Heterogeneity-Aware Approaches to Optimizing Performance of Computing and Communication Tasks
Type of DegreeDissertation
Electrical and Computer Engineering
MetadataShow full item record
As the domain of computing and communication systems grows, heterogeneity among computers and subnetworks employed for a task also increases. It is important to understand how heterogeneity affects performance of computing and communication tasks in order to optimally utilize heterogeneous resources. However, the effects of heterogeneity in heterogeneous computing and communication systems were either not taken into account explicitly or not thoroughly analyzed in the previous research work in this field. In this dissertation, effects of heterogeneity are analyzed, and heterogeneity-aware approaches are proposed for both computing and communication systems. In the computing system context, temporal heterogeneity refers to variation, along the time dimension, of computing power available for a task on a computer, and spatial heterogeneity represents the variation among computers. Effects of heterogeneity on the performance of a target task have been analyzed in terms of the mean and standard deviation of parallel execution time. The results reveal that, in minimizing the average parallel execution time of a target task on a spatially heterogeneous computing system, it is not optimal to distribute the target task linearly proportional to the average computing powers available on computers. Based on the analysis results, an approach to load balancing for minimizing the average parallel execution time of a target task is described. The proposed approach, of which validity has been verified through simulation, considers temporal and spatial heterogeneities in addition to the average computing power of each computer. In the communication system context, the concept of temporal and spatial heterogeneity in the available communication resource is applicable to various levels of a communication network. Effects of heterogeneity on the performance of individual messages in a heterogeneous communication systems are analyzed. A heterogeneity-aware approach to source rate control is proposed, which utilizes the heterogeneity information on the available bandwidth in a UDP-based protocol, to improve throughput, dropping rate, end-to-end delay, and real-time performance. Two main components of the heterogeneity aware source rate control scheme are a receiver side feature extractor and a sender-side adaptive rate controller. The feature extractor captures the dynamic features in the bandwidth heterogeneity, and the source rate controller utilizes the extracted features in the rate control. Performance of the proposed source rate control scheme has been analyzed in detail through an extensive simulation for the single and multiple path media streaming, and multiple HAA and/or TCP flows.