Data Partitioning and Asynchronous Processing to Improve the Embedded Software Performance on Multicore Processors

Nowadays, ensuring information security is extremely inevitable and urgent. We are also witnessing the strong development of embedded systems, IoT. As a result, research to ensure information security for embedded software is being focused. However, studies on optimizing embedded software on multi-core processors to ensure information security and increase the performance of embedded software have not received much attention. The paper proposes and develops the embedded software performance improvement method on multi-core processors based on data partitioning and asynchronous processing. Data are used globally to be retrieved by any threads. The data are divided into different partitions, and the program is also installed according to the multi-threaded model. Each thread handles a partition of the divided data. The size of each data portion is proportional to the processing speed and the cache size of the core in the multi-core processor. Threads run in parallel and do not need synchronization, but it is necessary to share a general global variable to check the executing status of the system. Our research on embedded software is based on data security, so we have tested and assessed the method with several block ciphers like AES, DES, etc., on Raspberry PI3. The average performance improvement rate achieved was 59.09%.