Memory optimization techniques for multithreaded operating system on wireless sensor nodes

Multithreaded operating system (OS) is essential to be used for many wireless sensor network (WSN) applications. However, the RAM consumption of the multithreaded OS is high. Thus, it is infeasible to apply the multithreaded OS on many high RAM constrained wireless sensor nodes (WNs). To address this challenge, several memory optimization strategies are investigated in this paper. On one hand, the subfunction-granularity thread switch and hybrid OS scheduling model are proposed to decrease the RAM consumption of the thread stacks. On the other hand, different dynamic memory allocation mechanisms are presented to reduce the heap memory size. With the implementation of these optimization techniques, the total RAM consumption of the multithreaded WSN OSs can be brought down greatly. In result, the multithreaded WSN OSs become suitable to be used even on the high RAM constrained WNs.