For lightly loaded multicore processors that contain more processing cores than running tasks and have dynamic voltage and frequency scaling capability, we address the energy-efficient scheduling of periodic real-time tasks. First, we introduce two energy-saving techniques for the lightly loaded multicore processors: exploiting overabundant cores for executing a task in parallel with a lower frequency and turning off power of rarely used cores. Next, we verify that if the two introduced techniques are supported, then the problem of minimizing energy consumption of real-time tasks while meeting their deadlines is NP-hard on a lightly loaded multicore processor. Finally, we propose a polynomial-time scheduling scheme that provides a near minimum-energy feasible schedule. The difference of energy consumption between the provided schedule and the minimum-energy schedule is limited. The scheme saves up to 64 percent of the processing core energy consumed by the previous scheme that executes each task on a separate core.

The proposed scheme maximizes the communication rates of multiple sender sensors to a terminal sensor while assuring a given connection time without connection abortion caused by the depletion of available battery energy. Each wireless sensor chooses one of two radio transmission ranges with disparate energy consumptions when sending a packet. The proposed scheme generates an edge-capacity graph from a sensor network and finds the routing paths by applying an advanced version of the existing maximum-flow algorithm to the graph. Finally the scheme determines the transmission rate of packets from each sender node to the terminal node, which is the number of transmittable packets in the found routing paths divided by the given connection time.

Hierarchical scheduling frameworks provide ways for composing large and complex real-time systems from independent sub-systems. In this paper, we consider the imprecise reward-based periodic task model in a compositional scheduling framework. Thus, we introduce the imprecise periodic resource model to characterize the imprecise resource allocations, and the interface model to abstract the imprecise real-time requirements of the component. The schedulability analysis of mandatory parts is analyzed to meet the minimum requirement of tasks. In addition, we provide a scheduling algorithm for guaranteeing a certain amount of reward, which makes it feasible to compose multiple imprecise components efficiently.

This paper addresses an energy-saving scheduling scheme of periodic real-time tasks with the capability of dynamic voltage and frequency scaling, on the lightly loaded multi-core platform containing more processing cores than running tasks. First, it is shown that the problem of minimizing energy consumption of real-time tasks is NP-hard even on the lightly loaded multi-core platform. Next a heuristic scheduling scheme is proposed to find an energy-efficient schedule with a low time complexity while meeting the deadlines of real-time tasks. The scheme exploits overabundant cores to reduce energy consumption using parallel execution and turns off the power of unused or rarely used cores. Evaluation shows that the proposed scheme saves up to 64% energy consumption of the existing method executing each task on a separate core.

We propose a novel scheduling scheme that determines the instant operation modes of multiple tasks. The tasks have probabilistic execution times and are executed on discrete operation modes providing different utilities with different energy consumptions. We first design an optimal offline scheduling scheme that stochastically maximizes the cumulative utility of the tasks under energy constraints, at the cost of heavy computational overhead. Next, the optimal offline scheme is modified to an approximate online scheduling scheme. The online scheme has little runtime overhead and yields almost the maximum utility, with an energy budget that is given at runtime. The difference between the maximum utility and the output utility of the online scheme is bounded by a controllable input value. Extensive evaluation shows that the output utility of the online scheme approaches the maximum utility in most cases, and is much higher than that of existing methods by up to 50% of the largest utility difference among available operation modes.

In this paper, we consider the mesh with a single bus as a multi-computer topology that enhances the communication capability of the mesh and show that the mesh with a single bus has more salient properties than the mesh, the hypercube, and other mesh variants. These properties are small diameter, relatively small degree, small average distance, suitable for broadcasting, small initial data distribution time, etc. We propose a dynamic load distribution algorithm to utilize the enhanced communication capability of the mesh with a single bus. Also, asynchronous bus control and arbitration logic is designed to support the proposed algorithm efficiently. It is shown through simulation that the proposed dynamic load distribution is superior to the previous receiver-initiated diffusion method known as the best to-date. The proposed algorithm shows better total execution time of tasks and better processor utilization with a smaller number of task migrations.