Real-time Memory Scheduling Based on Middleware
Jun Fang Wei · 2007
The high data intensity of real-time monitoring systems makes it difficult to ensure that subtasks can maintain sufficient memory to store and analyze the data gathered form the sensors.This paper describes the design and implementation of a memory scheduler for these types of monitoring systems.The memory scheduler mediates between multiple real-time tasks and the facilities of a real-time operating system——using feedback and other inputs,it adjusts memory allocations over time to ensure that high application-level QoS is maintained.The scheduling connects to its monitored tasks in a non-invasive manner based on and integrated with standard middleware,and implements an open architecture for new memory management policies.The test results show that the overloads of scheduler are reasonable and allow the monitoring system to function in the face of run-time memory resource contention.