After Brenna Berman’s keynote address, four Harvard SEAS professors gave lightning talks about their research on the Internet of Things. Scot Martin, Gordon McKay Professor of Environmental Chemistry, spoke about how “Drone-Based Sensing Opens New Frontiers in the Environmental Sciences: Possibilities for Amazonia.” Martin discussed his project to use drone technology to “smell” the atmosphere of the Brazilian Amazon to quantify biodiversity and detect environmental change. Traditional efforts at monitoring Amazonia suffer from two fundamental problems: stationary sensor towers yield data that is specific to too small a region, while sensor data captured by aircraft produces extremely coarse-grained data that is inadequate to the research questions it must answer. Drone-based monitoring has the potential to bridge these two extremes, providing granular data that captures the diversity of microclimates in the rain forest. Martin’s project harnesses IoT to revolutionize a number of stalled questions about biodiversity and ecosystem change, and if successful, will have a crucial impact on the fate of Amazonia.
Na Li, Assistant Professor of Electrical Engineering and Applied Mathematics, gave a talk about “Distributed Energy Management in Power Networks.” Her research lies in the design, analysis, optimization, and control of power grids. She spoke about the emergence of a “smart grid,” informed by distributed sensing, computation, communication, and actuation systems. She envisions a future in which this “smart” power grid distribution system will be a large-scale network of distributed energy resources (DERs), each introducing random and rapid fluctuations in power supply, demand, voltage and frequency. These DERs provide tremendous opportunity for sustainability, efficiency, and power reliability. Li addressed the major challenge at the heart of her research – how intelligent devices and independent producers can respectively change their power consumption/production to achieve near maximum and reliable efficiency for the power network. Her talk focused on developing scalable, distributed, and real-time control and optimization algorithms to achieve system-wide efficiency, reliability, and robustness for the future power grid.
David Brooks, Haley Family Professor of Computer Science, talked about “Taking Machine Learning to Edge Devices.” He discussed the ways in which deep learning methods have transformed numerous aspects of computing. These methods – which have unseated traditional approaches to computing – are driven by an unprecedented influx of data, new machine learning algorithms, and the emergence of cheap, high-performance computing through the cloud. Brooks, who conducts research into wireless sensor networks for environmental, seismic, and healthcare tracking, spoke about cognitive IoT as a natural evolution of machine learning, a product of enormous amounts of new sensor data and a demand for automated analysis. Brooks discussed the severe limitations that IoT devices currently face. They are typically small computing devices on the periphery of the network. They suffer from low compute power, poor communication bandwidth, and intermittent connectivity. As a result, today’s machine learning applications typically require shipping data to a remote cloud system. However, due to privacy concerns, users frequently resist having their data processed in public or private cloud systems. Professor Brooks summarized the efforts of Harvard SEAS faculty to build new machine learning hardware platforms that enable autonomous, cognitive IoT systems to overcome their present limitations.
Jim Waldo, Gordon McKay Professor of the Practice of Computer Science, gave a talk on “Some Ethical Considerations of the Internet of Things.” He asked participants to consider the ethical implications of their work in IoT, suggesting that they move from a paradigm in which they think only about what can be done to one in which they think about what should be done. Because the Internet of Things is creating a flood of new data about the world and the people who live in it, we have data about everyone at resolutions never before possible. Professor Waldo’s talk explored what privacy means in a fully instrumented world, and why current methods of privacy protection are inadequate. Waldo stressed that technologists have a moral obligation to the people whose data they gather. Instead of understanding themselves as “simple engineers” who can leave policy to the lawyers and philosophers, they must understand that they are making policy decisions every time they design a new product. Policy can be decades behind technology, so the design of a product inevitably overrides any policy that is implemented down the road. Waldo implored attendees to collaborate with ethicists and policy makers to justify their design decisions and to take privacy into account as they collect, aggregate, and draw conclusions about their data. He asked questions such as: how much data should we gather and record? What should we process locally and what should we send over the wire? Where do we draw the line between what is processed by a human versus by a robot? Is it really sufficient to de-identify data, when we know how readily re-identifiable most data is? These are ethical decisions that cannot be made retroactively, and Waldo stressed the importance of intentional, ethical IoT design.
This symposium marked a new beginning for CRCS, which is poised to launch a multi-disciplinary initiative on the Internet of Things. In the coming years, we intend to create a sustainable consortium including academia, industry and government organizations. If you would like to contribute your expertise to this collaborative endeavor, or be added to our mailing list so that you can enjoy similar events in the future, please email firstname.lastname@example.org.