Article Preview
TopIntroduction
Real-Time applications have the requirement of both fixed arrival patterns as well as random patterns in nature. The ‘fixed arrival sequence’ is termed as a periodic task, whereas term, ‘aperiodic task’ is used for a random one. For example, a mobile video phone application has stringent regular computing requirements for the number of frames received per second along with aperiodic requirements generated by a user such as a volume control and play-list editing. These user-generated requests are event-driven. Similarly, the flight control system executes pilot control commands such as control of sudden rise of temperature, and speeds are aperiodic requirements along with routine computation requirements. Besides these requirements patterns, time criticality classifies the system as hard and soft in nature. For hard real-time system failures to met one condition may lead to catastrophe; however, degraded quality is received even missing to meet many requests for the case of a soft real-time system.
The algorithms described in this paper determine when aperiodic tasks are executed in presence of periodic task set. They are solutions to the following challenges:
- 1.
Based on the execution time and deadline of each newly arrived aperiodic task, the scheduler decides whether to accept or reject the task. If it accepts the task to execute, it schedules the task so that the task completes its execution in time without causing periodic tasks and previously accepted aperiodic tasks to miss their deadlines. The problems are how to do the acceptance test and how to schedule the accepted aperiodic task set.
- 2.
The scheduler tries to complete each aperiodic task as soon as possible. The problem is how to do so without causing periodic tasks and accepted sporadic jobs to miss their deadlines.
The novel two-phase approach proposes for multi-level budget bandwidth preserving server (MLBBPS) is utilizing the concept of the multi budget with the construction of the budget sample on or after hyper-period. It consumes budget at a multi priority level for execution of aperiodic tasks while ensuring periodic one. The first phase performs offline feasibility analysis for periodic tasks and fixes up the static budget renovated at regular intervals. Apart from the static budget, slack based budget is computed, forming a budget pattern from hyper period to hyper period. Over the above budget pattern formed in the first phase, the calculated budget tuned in the second phase. The tuning has achieved through utilizing the online slack availability that arises due to execution of the task with less execution time (as compared to worst-case time considered at the time of feasibility analysis), and accumulating the small budget fragments into a larger one. Abeni.L. et al. (2015) proposed the improvement in the responsiveness of aperiodic tasks to enhances the acceptance execution ratio of aperiodic tasks, rejected by a deferrable servers. The complexity of the proposed novel algorithm is very similar to the deferrable server that intended to make sure of the correctness of the server. MemGuard: Memory bandwidth reservation system discussed by Yun, H. et al. (2013, April) gives efficient performance under heavy memory workloads in the multiprocessor system but lacking on throughput under time-varying memory workloads.
The significant contribution of proposed research work is to develop a novel approach to enhance the responsiveness of aperiodic task for multi-level budget bandwidth preserving server in Real-Time System. The proposed algorithm developed in two phases, the first phase is for the construction of budget1, budget2, and proposal for MLBBPS algorithm; the second phase is for refinement of budget to increase the responsiveness of aperiodic task. An illustrative example of a functional analysis of the algorithm shows the effectiveness of the proposed approach. The performance of the proposed algorithms evaluates for both synthesized task sets used in Erciyes K. et al. (2019) and data available for different applications in Hamann A. et al. (2018). The study indicates that the proposed multi-level budget bandwidth preserving server receives better responsiveness with an increased number of completed aperiodic tasks over a wide range of variations.