Power Management of Non-Virtualised Devices

TEP:115
Group:Core Working Group
Type:Documentary
Status: Draft
TinyOS-Version:2.x
Author: Kevin Klues, Vlado Handziski, Jan-Hinrich Hauer, Phil Levis
Draft-Created:11-Jan-2006
Draft-Version:1.1.2.4
Draft-Modified:2006-09-19
Draft-Discuss:TinyOS Developer List <tinyos-devel at mail.millennium.berkeley.edu>

Note

This memo documents a part of TinyOS for the TinyOS Community, and requests discussion and suggestions for improvements. Distribution of this memo is unlimited. This memo is in full compliance with TEP 1.

Abstract

This memo documents how TinyOS 2.x manages the power state of physical (not virtualised) abstractions.

1. Introduction

The energy resources on a typical TinyOS platform are limited, so every effort should be made to put all devices on a given platform into their lowest possible power-consumption states as often as possible. Depending on the device type, selecting the correct power-consumption state could be as simple as switching between on and off states, or could involve selecting the optimum state from among several. Choosing the best power-consumption state requires making tradeoffs between various factors such as power consumption, fidelity, performance, and wake-up latency.

Because of the large difference in the number of supported power- consumption states that could potentially be present on any given device, as well as the complexity in keeping track of the state information needed for safe execution of any sort of system wide power-control algorithm, a unified power management strategy for all devices on all possible platforms is simply infeasible. Developing such a solution would most likely be overly complex and in some cases even suboptimal. TinyOS 2.x, therefore, takes the approach of defining several different classes of devices for which several different power-management strategies have been optimized. [TEP112], for example, details how TinyOS 2.x manages the multiple power- consumption states for microcontrollers. This document, in turn, details the support for managing the physical and dedicated and physical and shared device classes as defined in [TEP108]. Devices belonging to these classes often have only two power consumption states (on and off), and can therefore be treated differently than devices that have multiple power consumption states. Additionally, it turns out that, in practice, TinyOS software often just uses the two most useful of the many power states in these cases.

Various policies can be implemented on top of these two classes of devices that decide exactly when a device should be powered on or off. For physical and shared devices, for example, a power management policy could be written that explicitly powers a device on and off whenever the higher level component that uses it no longer requires it. Another policy, however, might decide to defer the power down of the device for a a few milliseconds if its wake-up time is large enough and it is likely that the device will be needed again sometime in the near future. A similar set of policies could be implemented for physical and shared devices, except that an attempt to power down such a device would not occur until all components using it had signalled that they no longer needed it. Providing these sorts of policies cause devices belonging to one of these two classes to become virtualized in the sense that their power states are automatically handled. Any higher level components connecting to one of these virtual devices need only signify when they require the use of the device and when they don't. The time at which the device actually becomes powered on or off, is left up to the power managment policy being used.

In order to provide the building blocks for implementing such power managment policies on top of physical and dedicated and physical and shared devices, TinyOS 2.x defines two different power management models that devices belonging to one of these two classes must adhere to: the explicit power management model and the implicit power management model.

2. Power Management Models

The explicit power management model provides a means for explicitly controlling the power state of a physical device by some higher level component. Whenever this higher level component tells the device to power up or down it does so without delay. This model can be particularly useful when the control information driving the selection of the proper power state of a device is external to that device.

The implicit power management model, on the other hand, provides a means for allowing the power state of a device to be controlled from within the device itself. Devices following this model are never explicitly powered up or down by some external component, but rather require some policy to be defined that decides exactly when their power states should be changed. This policy could exist natively on the hardware of the physical device itself, or be implemented on top of some lower level abstraction of the physical device which adheres to the explicit power management model.

Just like in TinyOS 1.x, StdControl and SplitControl interfaces have been defined by TinyOS 2.x (along with a third interface, AsyncStdControl) in order to control the on and off states of devices that follow the explicit power management model. One of these three interfaces SHOULD be provided by any component wrapping a hardware device that supports switching between an on and off state. The selection of the right interface depends on the latencies involved in changing between these two power states as well as the nature of the code (sync or async) executing any of the interfaces commands.

Devices implemented according to the implicit power management model, on the other hand, MAY expose one of these three interfaces, but they are not required to do so. Virtual abstractions of physical and dedicated devices that implement power management policies according to this model, will indeed provide one of the three derivatives of the StdControl interface. The interface provided will no longer be used to explicitly power the device on and off, but rather allow the power management policy to determine whether the dedicated user of the device requires the device to be powered or not. It will then make its own decision as to which state to put the device in.

For physical and shared devices, the information required to determine if one of its many users require it to be powered or not can be inferred through information provided by the Resource interface they provide. Because of this, no StdControl-like interface needs to be provided, and the power management policy can act accordingly whenever users request of release the device.

2.1 Explicit Power Management using StdControl

When a devices powerup and powerdown times are negligible, they SHOULD provide the StdControl interface as defined below:

interface StdControl {
  command error_t start();
  command error_t stop();
}

An external component MUST call StdControl.start() to power a device on and StdControl.stop() to power a device off. Calls to either command return SUCCESS or FAIL.

Upon the successful return of a call to StdControl.start(), a device MUST be completely powered, and calls to commands of other interfaces implemented by the device abstraction CAN succeed.

Upon the successful return of a call to StdControl.stop(), a device MUST be completely powered down, and any subsequent operation requests through calls to commands of other interfaces implemented by that device abstraction MUST return FAIL or EOFF.

If a device is not able to complete the StdControl.start() or StdControl.stop() requests for any reason, they MUST return FAIL.

Based on these specifications, this matrix describes the valid return values of calls based on the compoenent's power state:

Call On Off
StdControl.start() SUCCESS SUCCESS or FAIL
StdControl.stop() SUCCESS or FAIL SUCCESS
operation depends FAIL or EOFF

Devices adhereing to this power management model would provide this interface as shown below:

configuration DeviceC {
  provides {
    interface Init;
    interface StdControl;  //For Power Management
    ....
  }
}

2.2 Power Management with SplitControl

When a devices powerup and powerdown times are non-negligible, the SplitControl interface MUST be used in place of the StdControl interface. The definition of this interface can be seen below:

interface SplitControl {
  command error_t start();
  event void startDone(error_t error);
  command error_t stop();
  event void stopDone(error_t error);
}

An external component MUST call SplitControl.start() to power a device on and SplitControl.stop() to power a device off. Calls to either command return one of SUCCESS, FAIL, or EBUSY.

Successful calls to SplitControl.startDone() MUST signal one of SplitControl.startDone(SUCCESS) or SplitControl.startDone(FAIL).

Successful calls to SplitControl.stopDone() MUST signal one of SplitControl.stopDone(SUCCESS) or SplitControl.stopDone(FAIL).

Upon signalling either a SplitControl.startDone(SUCCESS) or a SplitControl.stopDone(FAIL) event, a device MUST be completely powered, and operation requests through calls to commands of other interfaces implemented by the device abstraction SHOULD succeed.

Upon signalling either a SplitControl.stopDone(SUCCESS) or a SplitControl.startDone(FAIL) event, a device MUST be completely powered down, and any subsequent calls to commands of other interfaces implemented by the device abstraction MUST return EOFF or FAIL.

If a device is not able to complete the SplitControl.start() or SplitControl.stop() requests for any reason, they MUST return FAIL.

Calls to either StdControl.start() or StdControl.stop() while a start or top operattion is pending MUST return an EBUSY.

Call On Off Starting Stopping
SplitControl.start() SUCCESS SUCCESS FAIL EBUSY EBUSY
SplitControl.stop() SUCCESS FAIL SUCCESS EBUSY EBUSY
operation depends FAIL EOFF EOFF FAIL EOFF SUCCESS FAIL EOFF

Devices adhereing to this power management model would provide this interface as shown below:

configuration DeviceC {
  provides {
    interface Init;
    interface SplitControl; \\ For Power Management
    ....
  }
}

2.3 Power Management with AsyncStdControl

The commands and the events of the StdControl and the SplitControl interfaces are synchronous and can not be called from within asynchronous code (such as interrupt service routines, etc.). For the cases when the power state of the device needs to be controlled from within asynchronous code, the AsyncStdControl interface MUST be used in place of the StdControl interface. The definition of this interface can be seen below:

interface AsyncStdControl {
  async command error_t start();
  async command error_t stop();
}

All of the semantics that held true for devices providing the StdControl interface also hold for this interface.

Devices adhereing to this power management model would provide this interface as shown below:

configuration DeviceC {
  provides {
    interface Init;
    interface AsyncStdControl; \\ For Power Management
    ....
  }
}


.. Note::

Determining exactly when to use the AsyncStdControl interface
instead of simply posting a task and using the StdControl interface
can be tricky.  One must consider the advantages and disadvantages
of using one interface over the other.  How complex is the code
being I am trying to execute?  Do I really want to execute all of
my startup code in async context?  Do I gain some performance in
terms of startup latency or code size if I run my startup code in
async context?  These are just a few of the factors that need to be
considered when making this decision, and unfortunately there just
is no easy answer. Note that a component with an AsyncStdControl
cannot call SplitControl or StdControl. In practice, AsyncStdControl
is used for low-level hardware resources.

3. Power management policies

While the explicit power management model provides the basic means for controlling the power state of the device, it is void of any policy about who, when, or how the power of the device should be managed. This does not represent a large problem for the simple case of physical and dedicated devices, but can become crucial for non-trivial cases involving complex interdependencies between devices being controlled by multiple users.

For example, if component A is using two devices, B and C, what happens with B and C when one calls StdControl.stop() on the top component A? The above problem has its dual in the case of shared devices. Assuming that device A is shared by components B and C, the question exists as to when device A can be powered down?

The complex nature of the problem is evident from the number of unexpected behaviors in TinyOS 1.x involving StdControl. On several platforms, one of the SPI buses is shared between the radio and the flash device. On some of them, issuing StdControl.stop() on the radio results in a cascaded SPI bus becoming disabled, rendering the communication with the flash impossible. Of course, the right policy would involve tracking the users of the SPI bus and powering it off only once both the radio and the flash devices wer no longer using it. Conversely, the SPI bus should be powered on whenever there is at least one active user.

The selection of the right policy is a complex task that depends on the nature of the devices, their interdependency, as well as on the application requirements. For the cases when some of these features are known a-priori or are restricted in some sense, it is preferable that the system provide architectural support for enforcing a meaningful default power-management policy instead of simply passing the task to the application programmer to be solved on a case-by-case basis.

TinyOS 2.x provides two contexts of restricted resource interdependency where such a default power-management policy can be offered. For physical and shared resources (defined in [TEP108]) (and covered by the Hardware Abstraction Architecture [TEP2]), TinyOS 2.x provides a flexible implicit power management model that is tightly coupled with the arbiter concept and uses the basic mechanisms offered by the explicit power management scheme.

4. Implicit Power Management for Physical and Shared Resources

The physical and shared resource class defined in Section 2.3 of [TEP108], provides a well defined component interdependency, where a single resource is shared among multiple clients. This relationship enables definition of a simple default power-management policy that powers the resource off when no potential users are waiting for access to the resource. Conversely, the resource is powered on whenever a client requests its use. When the resource being controlled happens to be an actual physical device, the implicit power management model for physical devices as described in section two applies.

The realization of a specific power-control policy for shared reources does not need to be resource specific. Just as generic arbiters are offered in TinyOS 2.x to provide the arbitration functionality required by shared resources, generic power management policies are also offered to allow the power management of devices to be automated.

The PowerManager component implementing one of these polices acts as the lowest-priority user of the shared resource. In contrast to "normal" clients, the PowerManager interacts with the resource arbiter using the richer ResourceController interface:

interface ResourceController {
    async command error_t request();
    async command error_t immediateRequest();
    event void granted();
    async command void release();
    async event void requested();
    async event void idle();
}

Acting as a lowest priority client, the Power Manager waits for the ResourceController.idle() event to be signalled before trying to gain ownership over the resource. It does so using the ResourceController.immediateRequest() command in order to gain control of the resource as quickly as possible.

Once it owns the resource, the PowerManager is free to execute its power-management policy using the mechanisms provided by the resource via the explicit power-management model. Different managers can implement different policies. In the simplest case, this would involve an immediate power-down via one of the .stop() commands. When the power-state transition involves non-negligible costs in terms of wake-up latency or power consumption, the PowerManager might revert to a more intelligent strategy that tries to reduce these effects. One strategy might involve using a delayed power-down timer to defer the powers-down of a resource to some later point in time unless some normal-priority client requests the device in the meantime.

Regardless of the power-off policy, the PowerManager remains owner of the resource as long as the resource is not requested by a normal user. When one of these "normal" users finally makes a request, the PowerManager component will receive a ResourceController.requested() from the arbiter it is associated with. Upon receiving this event, the PowerManager MUST power the resource back on (in case it was powered-off) through one of the explicit power-management interfaces provided by the lower level abstraction of the physical device. The PowerManager can release the ownership of the resource (using the ResourceController.release() command) ONLY after the resource has been fully powered-on.

Modeling devices as shared resources and allowing them to be controlled in the way described here, solves the problems outlined in section 3 regarding how to keep track of who, when, and how the powerdown of nested resources should proceed. The PowerManager component answers the question of who, the combination of the power management policy being used and the reception of the ResourceController.idle() and ResourceController.requested() events answers the question of when, and through one of the explicit power-management interfaces provided by the lower level abstraction of the physical device answers how. As long as the device resource at the bottom of a large set of nested resource users is released at the proper time, the power mananger will ba ble to power down its device appropriately.

Using the model described above, a device resource that follows the implicitly powere management model could be built as shown below:

module MyFlashP {
  provides {
    interface Init;
    interface SplitControl;
    interface Resource;
    ...
  }
}
implementation {
...
}

generic module PowerManagerC(uint8_t POWERDOWN_DELAY) {
  provides {
    interface Init;
  }
  uses {
    interface SplitControl;
    interface ResourceController;
  }
}
implementation {
...
}

#define MYFLASH_RESOURCE "MyFlash.resource"
configuration MyFlashC {
  provides {
    interface Init;
    interface Resource;
  }
}
implementation {
  components new PowerManagerC(MYFLASH_POWERDOWN_DELAY)
           , FcfsArbiter(MYFLASH_RESOURCE)
           , MyFlashP;

  Init = MyFlashP;
  Resource = FcfsArbiter;

  PowerManagerC.ResourceController -> FcfsArbiter;
  PowerManagerC.SplitControl -> MyFlashP;
}

This example implementation is built out of three components. The first component (MyFlashP) follows the explicit power management model for defining the interfaces to the physical flash device. The second component (PowerManagerC) is the generic powermanager component that will be used to implement the specific power management policy for this device. The third component (MyFlashC) is the configuration file that wires together all of the components required by the implementation of device adhereing to the implicit power management model. It includes the MyflashP and PowerManagerC components, as well as an arbiter component for managing shared users of the device. Notice how the powermanager is wired to both the resourcecontroller interface provided by the arbiter, and the SplitControl interface provided by the flash. All normal users of the resource are directly connected to the resource interface provided by the arbiter. As outlined above, the PowerManagerC component will use the events signalled through the ResourceController interface from the arbiter to determine when to make calls to the commands provided by the SplitControl interface coming from the FlashP component in order to power it up and power it down.

5. Sample Power Managers: PowerManagerC and DeferredPowerManagerC

TinyOS 2.x currently has two sample power management policies in lib/power. Each policy has three implementations: StdControl, SplitControl, and AsyncStdControl. The two policies are immediate and deferred. T

6. Author's Address

Vlado Handziski
Sekr FT5
Einsteinufer 25
10587 Berlin
GERMANY

phone - +49 30 314 23831

Kevin Klues
503 Bryan Hall
Washington University
St. Louis, MO 63130

phone - +1-314-935-6355

Jan-Hinrich Hauer
Sekr FT5
Einsteinufer 25
10587 Berlin
GERMANY

phone - +49 30 314 23813

Philip Levis
358 Gates Hall
Stanford University
Stanford, CA 94305-9030

phone - +1 650 725 9046

6. Citations

[TEP2]TEP 2: Hardware Abstraction Architecture.
[TEP108](1, 2) TEP 108: Resource Arbitration.
[TEP112]TEP 112: Microcontroller Power Management.