CSR-CSI: An Update-conscious Compilation Framework for Energy-Efficient Code Dissemination in Wireless Sensor Networks

University Of Pittsburgh, Pittsburgh PA

Investigators

Abstract

The wireless sensor network (WSN) has been emerged as a promising embedded computing platform for many nontraditional applications. In many cases, the computation and data acquisition processes of these applications are dynamically correlated, which poses the need for code changes after the deployment. For example: code patches to handle new phenomena, or software bugs. A naïve way of performing code changes in a WSN is to send the code or code differences in form of data packets, and let the sensor nodes compute the new code. Since energy consumed by data transmission is much higher than that consumed by local instruction execution, it is important to be energy-efficient during the code dissemination in WSNs. This project proposes update-conscious compilation for achieving energy-efficient code dissemination. The integrated compilation framework consists of a set of sink-side update-aware compilation techniques, and a sensor-side software-decoder/binary-rewriter. Specifically, the intellectual merits of this project are: 1) update-aware register allocation techniques; 2) update-aware data allocation techniques; 3) update-aware code placement techniques; 4) tools for enabling update-aware compilation. The proposal also contains interesting problems that are simple to understand for students at their early research stages. This will attract more students into the traditional fields from a new angle. At the same time, this project provides students with opportunities to connect classic fields with the latest prevailing areas. Through this research, the PIs are committed to cultivating a strong interest and a positive attitude among students towards embedded software development.