PLAnetary Scale events impact all of us living on the same planet, e.g., natural disasters, climate changes, magnetic field changes, species migration patterns, etc. Such PLAnet Scale problems can not be monitored with a single sensor system deployed at a local area. As with various applications of environmental sciences, geological sciences, oceanography, etc., there is a significant demand to correlate data from various information sources for reliable data analysis. For instance, interpretation of particular solvent amounts in water depends on the current weather conditions, dissolved oxygen at different levels of the lake, and possibly the behavior of species around. In short, we need data from various systems, e.g., sensor systems deployed at distributed sites, remote sensing satellite data, etc.

When we consider sensor networks at this scale, we face novel research challenges. Foremost, we are faced with an enormous amount of data being generated. For large-scale deployments, it is highly uneconomical and also infeasible to collect all raw data from the network since most sensor nodes have resource constraints such as energy, memory, processing, storage, and communication limitations. On the other hand, even if it was possible to retrieve all the information, the amount of the collected data would be beyond the analysis capacity of the end users. For this reason, we exploit data integration and digestion in the network during data collection such that data processing is pushed in the network itself. This necessitates a close interaction between network communication and information systems.

In this project, we study self-organization and location management. Self-organization allows dynamic adaptation to network and application changes, but it comes at a cost of additional energy consumption. As energy limitations in unattended environments is a major concern, we proposed Adaptive Energy-Efficient Registration and Online Scheduling (AEROS) protocol. Our protocol exploits asymmetric data flow characteristics to select routes, and to formulate an organized transmission schedule. Our results suggest that AEROS's transmission schedule allows the minimum number of data message exchanges and guarantees a collision-free communication. Another concern for data analysis is the correct time and place labeling of collected observations. In cases where the precise location and timing of the observations are not known due to resource constraints at sensor nodes, position estimations are used. We proposed a novel location estimation algorithm that maintains a representative overall network topology using confidence levels to reflect possible errors.

  People:
    Demet Aksoy (faculty)

    Chi Chen
    Chi Nguyen
    Bo Won Kim
    May Wong
    Saravanan Balasubrahmanian
    Weiwei Cao
    Ryan Norton
    Tufan Demir

  Publications: