Efficient Utilization of the Communal Strength at the Edge of IoT

The Internet of Things (IoT) has been a primary component in building our modern civilization. Apart from monitoring applications, IoT technology plays a major role in various complex applications that involve real-time coordination among dynamic entities, such as multi-robot systems or intelligent transportation. In the traditional approach, an IoT device is mainly used to sense different physical parameters. Due to severe resource constraints, all analysis and decision-making tasks are typically performed either with the aid of cloud or fog devices. However, such external services may not be available all the time or may be occupied with other tasks, which can cause unexpected lag. Efficiently supporting modern advanced applications, therefore, requires an IoT system to be more self-sufficient. There have been efforts to enable IoT edge devices to manage a heavier computational load, which is an essential component for meeting the aforementioned requirements. Some of the approaches prune the computational requirements and ultimately result in a considerable reduction in the precision of the outcome. A set of works divides the computational load into a set of components, which is more promising than pruning. However, the works so far have been tested with a small number of nodes in a laboratory setting, where the components are explicitly allocated to the nodes. In this work, we examine the scenario in a large IoT system and the importance of component distribution. In particular, we study how efficiently the components of the tasks should be distributed so that they can efficiently collaborate with each other and achieve the expected goal. Our analytical study reveals a promising approach to distributing components that efficiently meet practical needs.