Software synthesis from dataflow schedule graphs

Abstract The dataflow-model of computation is widely used in design and implementation signal processing systems. In dataflow-based processes, scheduling—the assignment coordination computational modules across resources—is a critical task that affects practical measures performance, including latency, throughput, energy consumption, memory requirements. Dataflow schedule graphs (DSGs) provide formal abstraction for representing schedules processes. DSG allows designers to model as separate dataflow graph, thereby providing formal, abstract (platform- language-independent) representation the schedule. this paper, we introduce methodology based on explicit specifications application cooperating models. We also develop new techniques tools automatically synthesizing efficient implementations multicore platforms from these coupled demonstrate proposed synthesis through case study involving real-time detection people vehicles using acoustic seismic sensors.