Scheduling of Datacompression on Distributed Systems with Time- and Event-Triggered Messages

详细信息    查看全文

• 关键词：Scheduling ; Distributed systems ; Network on chip ; Time ; triggered services ; Real ; time systems ; Data compression ; MILP ; Optimization
• 刊名：Lecture Notes in Computer Science
• 出版年：2017
• 出版时间：2017
• 年：2017
• 卷：10172
• 期：1
• 页码：193-204
• 丛书名：Architecture of Computing Systems - ARCS 2017
• ISBN：978-3-319-54999-6
• 卷排序：10172

文摘

The compression of messages can improve schedulability by decreasing network latencies and contention at the cost of computational overhead for compression and decompression. Existing scheduling models do not consider compression as required for the deployment in distributed real-time systems. This paper presents an MILP model with decision variables, constraints and an objective function for selectively compressing messages as required for minimizing the system’s makespan, thereby optimizing the trade-off between communication time and computational overhead. We consider multi-hop communication in systems with multiple routers and computational nodes. The algorithm is evaluated using example scenarios and the results are compared to previous work without compression support.