Asset Tracker 2 with eDRX and PSM settings

We are developing a product that will use Asset Tracker 2 and Nordic's nRF Cloud.

We want to take advantage of the power saving offered by the LTE eDRX and PSM modes.  We are trying to sort out how Asset Tracker 2 sets either eDRX or PSM timer settings, and what the relationship is between those settings and the 'activeWaitTime' that is part of the device's configuration.

So, specifically, our question is:

  1. Does Asset Tracker 2 use a default eDRX or PSM setting?  Which one? What are the timer settings?
  2. How can those settings be changed?
  3. What, if any, is the relationship between whatever either the eDRX or the PSM settings are and the 'activeWaitTime' time setting that is part of the device configuration?
  4. How can we determine which setting, either eDRX or PSM that the device is currently set to?  For example, if I use the AT+CPMS? command the response is: +CPSMS: 1,,,"11000001","00001010", indicating the PSM mode is enabled, and if I use the  AT+CEDRXRDP command I get the response +CEDRXRDP: 4,"0001","0001","0001" which indicates the eDRX is enabled.  But they can't both be on at the same time on the same device.  So how do I know if eDRX is currently being used on the device or PSM?

Thanks in advance for your help!

  • > But they can't both be on at the same time on the same device. 

    Depends on the definition of "same time". A device may have negotiated both. In RRC connected, the device will be able to communicate (high energy consumption, 50-250mA). After a quiet phase (or usage of RAI, if supported), the device will switch to RRC idle (low energy consumption, 1mA). If eDRX is negotiated, the device will then use the extended times to enable the receiver (lower energy consumption <1mA). The device the stays in RRC idle until the PSM active timer expires. If eDRX is active, that time extended time must also be considered. e.g. 4s PSM active timer, 20s eDRX => the device will stay for at least 20s in RRC idle before it switches to sleep (lowest energy consumption 4µA + quiescent current of the rest of your device + self-discarge of the battery).

    So, a device may negotiated both at the same time, but actually effective at the really same time is only one, switching to the other, when the timers indicates it.

    I guess, in the end you will not save too much energy. Though you use MQTT, that seems anyway not to be the highest priority. And a tracker will also need energy for GNSS (50mA) and also some for "more network searches" (that may take the multiple amount of energy than just sending data requires), when moving around.

  • Thx Achim.

    Here's how I understand the difference between the the LTE-M eDRX vs PSM modes.

    Both modes have an identical idle/listen cycle when they are connected to the network and either sending or receiving.

    The difference as I understand it is that in the eDRX mode, the device is never allowed to go into the very deep sleep PSM mode.  The reason someone might want this is because once the device goes into PSM mode it is totally disconnected from the network and is unreachable until the PSM timer expires, which could be anywhere between 16 minutes to 413 days depending on the timer setting.

    When the device does finally wake up and come back online, it must go through a network sync and SIM sync and TAU update to reestablish itself back onto the network.

    In eDRX mode however, the device is never allowed to go into PSM mode and so is never totally disconnected from the network.   Because of this it never has to go through the syncing and TAU update process, and is reachable during the iDRX period.

    Here's a link to a short presentation I prepared recently to explain the difference to other people on our team: eDRX vs PSM.pdf

    Maybe my understanding is incorrect, but the way I understand it is that the device can be set to honour either the eDRX behavior rules, or the PSM behaviour rules, but not both at the same time.  The reason is that I may not want the device to ever go into the unreachable PSM deep sleep mode.

    However, if I'm wrong, and the device can, without my permission, go into the PSM deep sleep mode, how can I know that it's in that state?  Is there any way to query its state via the Asset Tracker 2 shadow state?

    Thanks again for your help.

  • My understanding is mainly based on power consumption analysis done with a PPK II.

    As you found out, it's easy to configure the device to negotiate both, PSM and eDRX. If you use a "larger" PSM active time (e.g. 20s eDRx, 60s  PSM active time), you will see the difference with eDRX only, PSM only, and eDRX with PSM.

    My interpretation of that power consumption is, that eDRX goes during RRC idle to sleep as well, but for shorter times (several seconds to minutes vs. hours for PSM) and wakes then up frequently at that shorter interval enabling the receiver (in my measurements about 10mC). The difference between the eDRX and the PSM sleep was with my SIM-cards the state and consumption of that SIM-card (about 20-25µA at 3V battery). If the device enters the PSM sleep mode, the SIM-card gets disabled (that saves these 20-25µA).

    If the device wakes up from PSM, that takes a "tiny bit more energy", but all in all in my experience, a CoAP/DTLS 1.2 CID exchange (request/response) takes about 100mC "all in". (I don't remember the eDRX number, with the SIM-card and the 10mC per receive window, that is simply far more than my approach with PSM and a 1 hour alive message.) 

    But again, all in all:

    You use TCP, that will drain your battery. Lot's more radio messages than you expect, and you will see, how frequently you need new connections. If you chose eDRX or PSM depending on your communication schema, e.g. 1 message every 2 minutes (eDRX) or every 1 hour (PSM) that will save energy. The savings from additional eDRX will not be relevant. And you need to consider, how frequent you need GNSS positions. That will additionally require a lot of energy. Even with A-GPS, 10s with 50mA are already 500mC. That's the energy my device runs 5 hours from. So, I can only recommend, that you start with own measurements and then you will see pretty easy, what consumes the energy and may be worth to optimize. And what is anyway only a minor part and savings will not really change the battery-runtime.

    Just, that you get an impression about the energy in the different states:

    LTE-M / PSM / no RAI / no eDRX

    That shows the initial peek when exchanging the message. That's followed by the quite high consumption during RRC connected (about 10s), followed by low consumption with frequent peaks during RRC idle (each 2.5s), followed by PSM sleeping.

    If eDRX is negotiated, then the peaks in RRC idle will be less frequent, eg. every 20s instead of the 2.5s without eDRX. This RRC idle time could also be disabled with a PSM active time 0.

  • Thanks.  Yes, we understand the different current draws in the different modes.

    We have used the PPK II to analyze current consumption as well. That's where the data for the slides in the presentation came from.

    I think we have a different understanding of the terms.

    I don't understand '...  negotiate both PSM and eDRX..." the same way you do.

    It seems that you are using the term eDRX mode for the period of time that the device is going through the idle/listen cycle.

    I understand the term differently.

    I agree that the idle/listen cycle behaviour is identical between the eDRX mode and the PSM mode.

    The difference is that in PSM mode the idle/listen cycle is determined by the PSM active timer, whereas in the eDRX mode there is no such thing as a PSM active timer, becuase in that mode the device is never allowed to go into PSM.

    The terms are confusing because the term 'PSM' is used to describe both the 'deep sleep' state (4.7 micro amps) and to describe the mode where the device is allowed to go into the deep sleep state.

    Whereas, my understanding is the term eDRX is used to describe a mode where the device is never allowed to go into PSM.  Instead it stays in the idle/active cycle.

    That's why I say that the device can't be set to both modes at the same time.  Either it's allowed to go into PSM or it isn't.

    I agree with you that by going into long periods of PSM it's possible to save battery life.  In that case the extra current consumption used by coming out of PSM and reestablishing on the network is negligible compared to the savings you get from being in PSM for long periods of time.

    The disadvantage of that is that the device is unreachable while it's in PSM, and that can be for very long time periods.

    That's why some users will want to set their devices to eDRX mode and eliminate PSM alltogether.

    Thanks again for your help.

  • > I don't understand '...  negotiate both PSM and eDRX..." the same way you do.

    "negotiate" means, the device requests it and the network grants it. It may be also something as "successful requested".

    The eDRX times will be active, when the device is in RRC idle.

    If you enable PSM and eDRX using a pretty large PSM active time, you will see your slide 5, followed by the PSM sleep of slide 4.

    See Modem current consumption

    PSM: 2.7µA

    eDRX: down to 9µA (655s interval, without considering the SIM)

    So, what do you think, 2.7µA and 9µA. Does the device sleep in 9µA or what do you expect to get "calculated" for that 9µA?