All eyes on code: Using call graphs for WSN software optimization

Efficient code is essential for Wireless Sensor Networks. Limited computational resources and low memory capacities require a disciplined and provident programming style. However, optimizing code requires tools to provide a deep insight into where the code may have potential for improvement. In this paper we present a way of generating call graphs of software for standard Wireless Sensor Nodes. We execute the software on the actual nodes to collect profiling information and visualize this data on a PC-based host system. The call graphs are enriched with information about function execution time, execution count and visualize the call chain of the program to allow the programmer to identify room for optimization.