The Internet of Things (IoT) has revolutionized the world, finding widespread utilization across diverse fields, industries, healthcare, and city management. Despite its success, the exponential growth of the IoT market has presented challenges for traditional IoT systems that demand attention. In urban scenarios, the significant influx of data from media-rich sensors poses a strain on local networks with limited resources, impacting their efficiency and effectiveness. Additionally, the impracticality and high cost of deploying sensors and network infrastructures in remote areas creates obstacles to IoT deployment.This thesis aims to explore the utilization of mobile entities as a solution to address these challenges. Specifically, we emphasize the usage of planned entities with predictable or controllable movements to enhance the sensing and networking capabilities of IoT systems, particularly in time-sensitive community IoT applications. Time-sensitive scenarios demand data collection and analysis within specific time frames to preserve data value. To comprehensively investigate the usage of mobility, this thesis explores strategies for integrating mobile entities in diverse scenarios across urban and remote areas, each presenting unique time-sensitivity requirements and design challenges.
In the first scenario, we investigate the utilization of public transit fleets in network-constrained smart city applications. Our proposal involves using these fleets and network infrastructures along their routes to establish a cost-effective backbone network for long-range sensor data transmission, effectively addressing the limitations of local network resources.To achieve this, we develop approaches for optimal deployment of network infrastructure along with planning data collection from public transit fleets, considering the heterogeneity of delay tolerance and priority of sensor data, as well as the trade-off between data delivery delay/loss and network infrastructure installation cost. This thesis evaluates the proposed approaches using real-world bus networks in Orange County, CA and compares them with several other methods.
As a second use case, we investigate the use of drones to enhance sensing coverage in mission-critical IoT applications, specifically high-rise fire monitoring, where in-situ sensors are unavailable due to extreme conditions. We design and implement a drone-based IoT platform for real-time data collection in fire settings. The platform provides a dashboard for firefighters to visualize monitoring areas, user interfaces for commanding tasks to drones, and automatic flight planning for multiple drones to fulfill specified monitoring tasks. We propose multiple-drone flight planning approaches, optimizing data collection processes while considering the heterogeneity of monitoring tasks in terms of periods and priorities, as well as the trade-off between sensing coverage and data quality. The proposed algorithms are evaluated in a simulated high-rise fire scenario with real building structures at UCI. Additionally, we assess the applicability of the proposed system by implementing it in a lab-based testbed with mockup high-rise fires.
In the third scenario, we focus on utilizing drones to assist mobile sensing and sensor data transmission of IoT-based monitoring systems in remote areas with limited in-situ sensors and poor network conditions, particularly for wildland fire monitoring. We automate the drone-based monitoring system by enabling real-time perception of the physical world based on sensor data, automatic task generation for mobile entities, and dynamic planning and control of their movements to continuously monitor dynamic environments. We propose a rule-based task generation procedure for spatial-temporal monitoring requirements based on fire status and prediction. Additionally, we investigate approaches for multiple-drone flight planning, considering data collection timeliness, the trade-off between sensing coverage and data quality, and network disconnection during flights. The proposed flight planning algorithms are evaluated using simulated wildland fire burns at the Blodgett Forest Research Station, and the system's applicability is assessed through lab-based testbed implementation.
Overall, this thesis offers valuable insights into using mobile entities to address challenges in traditional IoT systems and enhance time-sensitive IoT applications. The exploration of different scenarios, from leveraging public transit fleets as a cost-effective backbone network to employing drones for high-rise fire monitoring and remote area sensing, demonstrates the versatility and practicality of mobile solutions in advancing IoT technologies.