| Summary: | Each winter the Climate-Ecological Observatory for Arctic Tundra (COAT) project deploys a range of small devices to measure and monitor the climate changes that occur in the Arctic regions in an attempt to gain better understanding of how the changes are affecting the Arctic tundra ecosystems. The deployed devices are often limited in terms of energy and connectivity range. Due to this, researchers face the issue of not being able to efficiently extract data from the devices placed on the Arctic tundra - this is often a manual and tedious task as researchers have to themselves collect data from the devices. This dissertation describes and implements a simulation of detached, interconnected sub-networks consisting of energy efficient Observation Units (OUs) placed on the Arctic tundra. A mobile data gathering device, a Mobile Ubiquitous LAN Extension (MULE), moving between the sub-networks creates a dynamically, temporary on-demand network which the detached networks may utilize to store and forward data reliably back to persistent storage. Dynamic Mobile Network Infrastructure (DMNI) presents a three layered architecture which forms the basis of the thesis - the application layer consisting of backend services, the network layer consisting of MULEs and the data layer with the isolated partitioned ad hoc networks of interconnected OUs. By utilizing data MULEs, we show through simulation and experiments that we can mitigate the limitation that systems placed in remote areas may face - permanent partitioning and complete disconnection from backend systems. By using a mesh-like structure in the sub-networks, we show that a MULE only require a single connection to an OU part of the network to accumulate all data - actively reducing the time, power and complexity to collect data. Simulation and experiments show that we can reduce the package-loss ratio to below 5%, even as low as 3.01%, by using a MULE to OU ratio of 30%. It also shows that the system has a low CPU and memory footprint on a real device, only using 2.2% total device CPU and 1.3% total device RAM. DMNI provides a solid first step towards a more refined MULE based system for data accumulation from remote, partitioned ad hoc networks of interconnected OUs in the Arctic.
|
|---|