Dynamic memory allocation prediction algorithm based on Markov chain

Abstract

When the Operating System processes memory allocation, if the linked list of free memory blocks does not contain any memory block that meets the current requirement, the system needs to apply for a larger linked list of free memory blocks or enable the memory reclamation mechanism, which consumes time and affects the memory allocation efficiency. Based on the analysis of classical partner algorithm and Markov chain prediction principle, this paper proposes a dynamic memory prediction allocation algorithm based on the Markov chain. The algorithm can predict the dynamic memory allocation of the system by updating the state transition matrix based on the state transition information of memory blocks. By adding a memory prediction module in the kernel to predict the size of the next memory block, the time consumption caused by the above problems in memory allocation is reduced. The experimental results show that the dynamic memory allocation prediction algorithm based on the Markov chain is more stable and has lower allocation delay.