Re: [core] sleepy to sleepy communication I-Ds

[Jari Arkko <[e-mail]>]

Good points, Bruce.

I will add that for many applications and platforms, the ability to communicate is the primary issue and waking up the CPU is secondary. In addition, on many link layers the ability to communicate depends on the network as well as the device, as it is the network that does the final synchronization, granting of time slots, etc. These may not matter on a coarse time scale (I'll wake up to listen for messages from 00h00min0s to 00h00min5s, but it may matter for smaller time scales (it definitely does for a millisecond).

Jari

On 01.03.2012 23:56, Bruce Nordman wrote:
> In some contexts it has been helpful to identify "light sleep" vs. "deep sleep" when
> there are two sub-states with different characteristics.  Beyond two, the terminology gets trickier.
>
> It is probably already in the various drafts but it is important to distinguish the link
> state from the device power state.  Energy Efficient Ethernet (IEEE 802.3az) created
> a sleep state for Ethernet with quick sleep/wake times (as opposed to a link disconnected
> state with very long renegotiation of the link).  On PCs, the link state and the system
> power state are completely independent.  Turning off the radio on a duty cycle sounds
> like a link sleep state that enables faster reestablishment of the link than from a no-link state.
>
> I have found that it is difficult to impose detailed requirements on the specifics of
> power states across a wide variety of devices, but CORE's clarity on this for CoAP
> low-power devices will be of great help in future.
>
> Thanks,
>
> --Bruce