Parallelization in software systems used in wireless sensor networks and Internet of Things: Case study: Middleware systems

In the Internet of Things (IoT) architecture, middleware is the component that creates a bridge between the heterogeneous sensor nodes and the software applications. One aspect among many others that determine the efficiency, performance, and functionality of middleware is parallelizability. Parallelizability refers to how ready the software is to utilize multicore architecture. The problem in general has been studied recently. This paper presents a study that examines the middleware software systems that have been widely used in IoT and wireless sensor network (WSN) applications. The examination gives us an idea of how well the software takes advantage of the hardware's multicore technology. The study is conducted on the source code of three open source middleware systems containing 1.07 million lines of code and almost 5k files. Each system is analyzed and the inhibitors to parallelization are detected and presented. The most prevalent inhibitors are pointed out to show the developers of IoT middleware systems what they should focus on when refactoring and improving their products. A historical trend over the last five years concerning the presence of inhibitors in the source code is also presented. The study suggests some techniques for refactoring the source code to improve the parallelizability of the systems as well.