Heterogeneous Parallel Embedded Systems

Embedded systems are driving an information§revolution with their pervasion in our everyday §lives. With more and more different types of FUs§available, same type of operations can be processed§by heterogeneous FUs with different costs, where the§cost may relate to power, reliability, etc. Due to§conditional operations, some tasks may not have fixed§execution time. Therefore, how to assign a proper FU§type to each operation of a system and generate a§schedule that minimize the total costs while§satisfying timing constraints with guaranteed§confidence probabilities becomes a critical issue.§This book, 1) proposed several efficient algorithms§to solve the problem; 2) combined Dynamic Voltage§Scaling (DVS) and soft real-time to solve the Voltage§Assignment with Probability (VAP) Problem. 3) applied§our efficient algorithms to dynamic adjust the§working mode of sensors to save energy for sensor§networks; 4) designed rotation scheduling algorithms§for real-time applications that produce schedules§consuming minimal energy; 5) combined data mining and§prefetching to reduce energy consumptions; 6)§improved performance and energy consumption for§processors with multi-bank memory.