| dc.description.abstract | IoT has seen steady growth over recent years with smart home appliances, smart personal gear, personal assistants, industrial assistance and many more. Devices used in the Internet of Things (IoT) are often low-powered with limited computational resources. Whereas, the computation part is done in the backend Cloud server. In this thesis, we compare how the scenario changes when computation is done in edge Cloud, near to the data source and thus reducing the distance of network hop and size of data for IoT scope. We developed a face recognition framework as an IoT application with computational server in two different infrastructures: a local, near to the client as edge Cloud, and also in commercial Cloud platform. Also implementing a part of computation in edge node or gateway can decrease the number of data packets in a huge amount and therefore, reduces network latency. In our thesis, the processing time of our developed system and network latency have been measured and compared. The results demonstrate that using edge Cloud, rather than core Cloud is comparably faster in terms of network latency. Moreover, decreasing the size of the transmitted data by computing in client side, reduces network latency and congestion. | en_US |
