Pattern Recognition based Adaptive Real-time Scheduling

Unmanned and autonomous real-time system generally run in uncertain, highly dynamic environments. Currently, there is no easy way to model such kind of systems. This paper presents a Pattern Recognition based Adaptive Real-time Scheduling (PRARS) framework for adaptive real-time systems. The usage of Pattern Recognition Theory provides a scientific underpinning on PID control. Through processing feature information, establishing character mode collection, pattern recognizing, and building control rule collection, we implement the PRARS. This enables us to fulfill more precise and efficient QoS and admission control, and guarantees the dynamical requirements of resources. Thus complex modeling methods could be avoided. The algorithm ensures robust performance of real-time tasks.