Wireless image/video communication networks are proliferating rapidly in applications like smart camera systems, military and civilian surveillance, green buildings, immersive computing, bio-telemetry, and autonomous robots/vehicles. Low-power embedded systems for multimedia sensing and communication are essential for their effective deployment. This research explores principles to minimize power dissipation of embedded systems for real-time imaging, high-volume multimedia processing, and wireless communication in time-varying noisy channel under Quality-of-Service (QoS) constraints. The heterogeneous integration of image sensor, memory, digital, and RF in a 3D-integrated circuit (IC) is explored to achieve this goal. The functional and physical interactions of components, algorithms, and environment in the 3D heterogeneous IC is investigated to create system design, physical analysis, and real-time control principles for power reduction. The principles are applied to design a low-power 3D wireless image sensor node for high-quality image/video communication network. The educational plan of the project includes creating a new undergraduate course, "Physics of Computation", a visual VLSI learning tool, and a "toy-sensor" chip design project for undergraduate courses; and fostering undergraduate research. This plan develops the pedagogical methods and tools for delivering an integrative learning of VLSI, instead of isolated skills in device, circuits, and systems, to next generation students. The outreach activities include engagement with the Summer Undergraduate Research in Engineering/Science program at Georgia Tech; recruiting under-represented students through the Facilitating Academic Careers in Engineering and Science for African-American students program; and participation in the First Lego League for the middle school students in the State of Georgia, hosted by ECE, Georgia Tech.
