Invitation to SIE PhD Defense by David Almeida, Oct 13 2023
David Almeida, candidate for PhD in Spatial Information Science and Engineering, will be defending his dissertation on Friday October 13 starting at 1 PM in 336 Boardman Hall. All are invited to the public portion.
The title of his dissertation is A Map-Algebra-Inspired Approach for Interacting with Wireless Sensor Networks, Cyber-Physical Systems or Internet of Things
The expanding development and deployment of wireless sensor networks (WSN) and Internet of Things (IoT) as spatially distributed sensor nodes deployed in the environment creates new opportunities for near real time spatial analysis. The approach to consuming data from such deployments has typically been to send data back to central servers for processing and analysis. This thesis develops an alternative strategy for processing, analyzing, and acting on data directly in the environment referred to as Active embedded Map Algebra (AeMA). Active refers to the near real time production of data and analyses by the model, and embedded refers to the architecture of the highly distributed and constrained environment of embedded sensor nodes. The approach presents a number of challenges to be overcome, which include efficiently and flexibly coordinating distributed processes among resource constrained nodes.
Macroprogramming is a style of programming that has been adopted for wireless sensor networks and IoT as a means to address the challenges of coordinating the behavior of multiple connected devices as a systems level behavior with a high-level programming model that abstracts from a series of low-level network details. While several macroprogramming models have been proposed, none to date have adopted a comprehensive spatial model. This thesis takes the unique approach of adapting the well-known Map Algebra model for spatial analysis from Geographic Information Science to extend the functionality of WSN/IoT and the opportunities for user interaction with WSN/IoT. Map Algebra, as an inherently spatial model, renders the Map Algebra-inspired metaphor suitable for the types of computation desired from a network of geographically or spatially-dispersed WSN nodes, where inherently spatial results can be made available “naturally” through the conceptual model.
The AeMA model incorporates a data model aligned with the conceptual model of GIS layers and specific layer operations from Map Algebra. A declarative query and network tasking language, based on Map Algebra operations, provides the basis for operations and interactions in the highly-constrained WSN environment. The model adds functionality to calculate and store time series and specific temporal summary-type composite objects which represents an extension to traditional Map Algebra. The AeMA encodes high-level Map Algebra-inspired operations into an extensible Virtual Machine Runtime system, called MARS (Map Algebra Runtime System), that supports Map Algebra in an efficient and extensible way. Map algebra-like operations are performed in a distributed manner. Data do not leave the network to be analyzed centrally but are analyzed and consumed or operated upon in place. A consequence of processing and analyzing data in place is that information is available in-situ to drive local actions. The conceptual model and developed tasking language are designed to direct nodes as active entities, able to perform some actions on their environment.
Dr. Kate Beard
School of Computing and Information Science
University of Maine