I was thinking of ideas for students to do their bachelor or master degree thesis with the Xilinx ZYNQ device.

Development of A Windows CE BSP for the ZYNQ:

Currently, there is a company, Adeneo which is providing a board support package (BSP) for the ZYNQ. This is however not coming free. In my idea it would be very useful to assess the possibility of creating a BSP for the ZYNQ device using what Microsoft and Xilinx provide as free material. Later this BSP can be provided to all other people for free and even enhanced versions can be sold at some cost. This task might be more suitable for students interested in programming and software development for embedded systems.

Currently, to the best of my knowledge there already exist a complete Android platform running on the ZYNQ, but for Windows CE, there is no freely available sources. One reason can be the Microsoft is not giving the Windows CE development environment for free (which is absolutely a wrong decision) but in any case for educational purposes it can be easily obtained.

Development of a 2D/3D GPU for the ZYNQ:

Currently there exists a set of IP cores from logicBRICKS but they are being presented at high prices and may be they are not really suitable for research. The goal of the project here is to begin development of freely available and open-source IP core for a graphics processor. The final target is to be able to bring up the Android graphical interface using this developed graphical processing engine. As we have the IP developed, we can do a bunch of explorations and experiments on the power consumption and performance of this block which can lead to scientific papers. Furthermore, enhanced versions of the core can be sold as commercial products.

Wearable Computing using ZYNQ:

This might be a crazy idea since wearable computing is always constrained to extremely low power consumptions. Here the idea is we improve the power consumption of the required processing algorithms in such devices by performing hardware/software partitioning. Indeed we try to run those parts of the algorithms which have the potential of being directly implemented in hardware on the Programmable Logic (PL) part of the ZYNQ instead of the ARM cores. We then evaluate how such approaches to processing can improve the overall power consumption of the system.

The idea can be generalized to Wireless Sensor Networks where you have a large number of sensors producing streams of data which should be processed all together and at the same time.