Multi-device powersaving

摘要

A control system reduces energy consumption in a multi-device system comprising a plurality of devices. The control system includes at least one processor. The processor is programmed to receive a job to be executed, as well as a selection of one of the plurality of devices for executing the job and a transfer cost for transferring the job from the selected device to each of the plurality of devices. A device to execute the job is determined through optimization of a first cost function. The first cost function is based on the device selection and the transfer costs. The job is assigned to the determined device and a time-out for each device in the multi-device system is determined through optimization of a second cost function. The second cost function is based on an expected energy consumption by the multi-device system. The devices are provided with the determined time-outs.

主权项

1. A control system for reducing energy consumption in an associated multi-device system comprising a plurality of devices, said energy control system comprising:
at least one processor programmed to:
receive a job to be executed;receive a selection of one of the plurality of devices for executing the job and a transfer cost for transferring the job from the selected device to each of the plurality of devices;determine a device from the plurality of devices to execute the job through optimization of a first cost function, the first cost function based on the device selection and the received transfer costs;assign the job to the determined device;determine a time-out for each device in the multi-device system through optimization of a second cost function, the second cost function based on an expected energy consumption by the multi-device system; and,provide the devices with the determined time-outs; wherein at least one of the device and the time-outs are determined by modeling a control problem as a Markov Decision Process (MDP), the MDP including the first cost function and the second cost function, wherein at least one of:
a) the first cost function in an optimal regime is computed according to the expression:
Vhσ,z=minkRhk+(1−σk){circumflex over (b)}kVσ+(1−σk)ek,z  (19) where k is one of the plurality of devices; h is the selected one of the plurality of devices; Rhk is the transfer cost for device k; {circumflex over (b)}k is the cost of transitioning device k from a first mode to a second mode, the second mode consuming more power than the first mode; a is a control state vector of the multi-device system, each k-th component indicating a state for device k, where a value of 0 corresponds to the first mode and 1 corresponds to the second mode; z is a demand state of the multi-device system; σ,z is a state of the multi-device system; ek is a vector of Os for each component except the k-th component, which is 1; and V is the second cost function; and
b) the second cost function in an optimal regime is computed according to the expression:V⁢⟨σ,z⟩=min(τk)⁢k⁢⁢ε⁢⁢σ⁢∫xl⁢[∑k∈σ⁢gk⁡(x,τk)+γ⁢⁢∫z′⁢Vl⁢⟨∑k∈σ⁢II⁡[x<τk]⁢ek,z′⟩⁢ⅆQ⁡(z′|zxl)]⁢ⅆP⁡(xl|z),(5) where k is one of the plurality of devices; gk is a cost of maintaining device k in a second mode during part or all of a time between jobs x given a time-out τ; γ is a discount factor; l is a selected device; σ is a control state vector of the multi-device system, each k-th component indicating a state for device k, where a value of 0 corresponds to a first mode and 1 corresponds to the second mode, the second mode consuming more power than the first mode; z is a demand state of the multi-device system; σ, z is a state of the multi-device system; ek is a vector of Os for each component exception the k-th component, which is 1; Vl is the first cost function; II[condition] returns 1 if condition is met and 0 otherwise; P is a probability distribution of time of a next job and its device given a demand state; and Q is a probability distribution of a next demand state given a previous demand state and time since a previous job and the selected device.