Augmenting The Performance Of MobileDevices Through The Use of DynamicPartition Offloading With HeterogeneousMobile Clouds
Year: 2020
Type: dissertation
Abstract: Smart mobile devices (SMD) are a part of our daily lives and sometimes we cannot do without them. With this, they are expected to have high functionalities and resources which is sometimes not the case i.e they are resource constrained. The Mobile Cloud Computing (MCC) paradigm was introduced to solve this issue and previous works have been implemented but are not without their flaws. Most works do not consider the context of the SMD when making offloading decisions and utilize only one offloading resource option which is not particularly ideal and limits the usage of the paradigm. In this paper, we present a context aware heterogeneous approach to the MCC paradigm. We apply heterogeneity by using serverless computing functions and remote mobile clients as our offloading options. We make offloading decisions based on the SMDs context (battery, network & memory) during runtime by applying dynamic partitioning. The core of our system is the decision making engine and remote configurations. Remote configurations have been implemented to change parameters in our approach without having to modify source code. This provides flexibility. Results gotten from experiments carried out showed our system was able to make offloading decisions based on changing context and also up to 66% reduced execution time (Zhou et al. produced around 65%) for offloaded tasks as compared to local executions thus reducing power utilization. Our solution also proved to be cost efficient when utilizing serverless computing as opposed to using cloud VMs as an offload resource.
Source: NORMA
Author: Emmanuel Okechukwu Weje
Institution: National College of Ireland
Language: English
Topic: IoT and Edge/Fog Computing
Subfield: Computer Networks and Communications
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W3161657415
