Critical Scheduling for Soft Real Time Operating System

Abstract

The scheduling approaches have largely been analyzed so far, but the attention given to multi-processor environments and real-time operating system is very less. Mostly all the approaches which are optimal are appropriate for the general-purpose operating system which cannot accomplish the real-time systems requirements as the RTS is closely concerned about the constraints of timing. The correctness of output is extremely affected by the deadlines means the output must be generated within a given time-bound. Nowadays, the real-time systems are multi-tasking and multiprocessing, so the scheduler plays an important role for controlling the functioning into the system. Different researchers had designed numerous techniques for real-time task scheduling to meet the varying needs of different environments like uniprocessor systems and multiprocessor systems. The selection of scheduling approach is an essential task as the output is greatly affected by which approach is being used in which environment. Furthermore, existing schedulers designed for supporting RTS aren t absolutely optimal for multiprocessor system. The schedulers designed earlier had focused on priority assignment which decides only the execution order of a task in the task set. newlineNowadays, we are entering into the era of the high-performance computing, which demands for the multi-processor environment. The uniprocessor system and the multiprocessor system; both are having their individual requirements and challenges as per their architecture. In the multiprocessor system the scheduler has to address both the problems like process selection (Priority assignment) as well as processor selection (set Processor affinity).All the existing approaches for multi-processor real-time system described in literature are divided in two categories; 1) Global scheduling approach 2) Partitioned scheduling approach and they all are application dependent approaches. newline The concept of affinity is used widely in the non-real-time environment to improve the cache performance