@inproceedings{oai:ipsj.ixsq.nii.ac.jp:00210324,
 author = {Masato, Fukui and Yoichi, Ishiwata and Takeshi, Ohkawa and Midori, Sugaya and Masato, Fukui and Yoichi, Ishiwata and Takeshi, Ohkawa and Midori, Sugaya},
 book = {Proceedings of Asia Pacific Conference on Robot IoT System Development and Platform},
 month = {Mar},
 note = {In recent years, edge computing has been attracting attention for the purpose of offloading advanced processing by IoT and robot services. It needs to efficiently provide services to multiple robotics services while dynamically switching abundant computational resources such as many-core resources. In Many-core research, many studies have been conducted to increase the degree of parallelism of tasks to improve responsiveness and computational efficiency. However, there is still not enough discussion on how to effectively allocate resources in a way that is suitable for integrated services and to achieve both efficiency and responsiveness. In this study, we construct an efficient resource allocation prediction formula as a study of an efficient Many-core allocation method for edge offload. In addition, as a specific service, we decided to evaluate the offload of AI processing of communication robots. In the construction of the prediction formula, it was confirmed that sufficient processing performance and responsiveness can be obtained with the minimum core by automatically calculating the optimum number of cores from the actual application operation., In recent years, edge computing has been attracting attention for the purpose of offloading advanced processing by IoT and robot services. It needs to efficiently provide services to multiple robotics services while dynamically switching abundant computational resources such as many-core resources. In Many-core research, many studies have been conducted to increase the degree of parallelism of tasks to improve responsiveness and computational efficiency. However, there is still not enough discussion on how to effectively allocate resources in a way that is suitable for integrated services and to achieve both efficiency and responsiveness. In this study, we construct an efficient resource allocation prediction formula as a study of an efficient Many-core allocation method for edge offload. In addition, as a specific service, we decided to evaluate the offload of AI processing of communication robots. In the construction of the prediction formula, it was confirmed that sufficient processing performance and responsiveness can be obtained with the minimum core by automatically calculating the optimum number of cores from the actual application operation.},
 pages = {46--50},
 publisher = {情報処理学会},
 title = {Predictive Many-core Allocation Method for Edge Off-Road Services},
 volume = {2020},
 year = {2021}
}