Internet-of-things

Challenges

With computation capabilities becoming faster and more affordable, connectivity capabilities increasing in the era of 5G, and competition growing among various device makers, IoT networks are more feasible than ever to build. While these advances allow for enormous flexibility in a network’s design and implementation, the financial and time investments in building such a network are still substantial. The performance of the network—both at the individual sensor level and the network level—deserves careful analysis.

In particular, there are several important questions to ask when designing and maintaining an IoT network. 

  • Where are the gaps and bottlenecks in our sensor network coverage? 
  • Could we maintain coverage with a more strategic deployment of fewer sensors? 
  • How robust is our coverage to device malfunctions or hostile weather conditions, and can we better design our network to protect against these issues? 
  • Could we get more from the sensor data we collect, and can we improve battery life in the process?

Our Solutions

On the edge, GDA’s signal processing capabilities enable energy and computational resource conservation with domain-specific signal compression and efficient device-to-device communication. In learning environments, our fusion of data from heterogeneous sensor networks, including data from video, audio, motion, temperature, and other sources, improves machine learning inference over a single data source.

In the cloud, GDA’s shape analytics and sensor modeling capabilities enable optimal placement of devices. Furthermore, in situations where robustness guarantees of coverage become critical, both in terms of probability of coverage and handling device failures, our network design capabilities allow for the construction of IoT networks that can minimize the risk of gaps and bottlenecks in coverage while still minimizing the cost of building and maintaining the network.

Case Study

IoT over the Ocean

The complexity of the ocean environment requires advanced systems to understand maritime dynamics and activity. Improved maritime analysis is needed to provide a detailed understanding of the ocean environment, inform regulatory commitments to protect natural resources, and enable the U.S. military to operate more effectively. The time and effort required to deploy floats over the geographic scale of the ocean make addressing the standard IoT challenges---like maximizing inference capabilities and preserving battery life---particularly important.

To address these IoT challenges, GDA works closely with float manufacturers to apply our topological signal-processing toolkit, which operates at the edge to extract parsimonious but information-rich summaries of massive amounts of mission sensor data and transmits key findings from maritime edge to an on-shore hub. These summaries allow powerful on-shore machine learning to take place while respecting the communication constraints imposed by satellite data transmission.

GDA’s shape-analytics toolkit is well suited to address the complexities of ocean dynamics. Our Float Field Geometric Health software suite ingests GPS reports from sensors and outputs a holistic and user-friendly set of multi-scale geometric summaries of sensor field health, allowing users to better understand and predict holes and bottlenecks in sensor coverage. The suite also offers recommendations on improving the robustness of sensor coverage, as well as placement and quantity of future deployments.