6LoWPAN current consumption

The power consumption of a network is tricky to determine because it depends on so many different factors:

• The data rate of each node
• The number of nodes in the network
• The number of hops in the mesh
• The amount of routing changes in the mesh
• The amount of interaction with the network (i.e. outside data coming into the network)
• ... and many more

These all depend on the specific application and environment in which the system is to be used and are difficult to predict beforehand. This makes it difficult to give an indication of a typical power consumption.

That said, I may be able to provide some useful data that will give you a ballpark figure at least. Our (Thingsquare's) system uses 6lowpan and runs on the CC1310. While I don't have any power data on the CC1310 yet, we measured the current draw of a Sensortag/CC2650-based system (video example here).

An oscilloscope reading is below. The highlighted part is sensing four sensors and transmitting the data in four separate messages - that's the peaks in the graph. The four left peaks are the sensing and the right (wider and higher) peaks are the transmissions. The smaller peaks on the left of the highlighted area are channel samples. Channel samples are used by the sleepy router mechanism to be able to wake up when a neighbor sends a packet.

Based on the above measurement, we can compute a rough estimate of power consumption / lifetime on a Sensortag, with its built-in coin cell battery, that acts as a non-router device:

• 1 * 4 sensor readings/minute: 4.9 months lifetime
• 2 * 4 sensor readings/minute: 4 months of lifetime
• 60 * 4 sensor readings/minute: 11 days of lifetime

Note that these measurements are taken with the Thingsquare system and will differ from similar measurements taken with other systems on the same hardware. They are also done on the CC2650, which is slightly different from the CC1310 (for example, CC2650 has a higher data rate so consume less power when sending a packet). But they may give a ballpark number.

Hope this is helpful,

/adam, Thingsquare CEO

Hello Adam,

you got some interesting information.

As I can calculate for next cases we got next average current consumption:

•    1 * 4 sensor readings/minute: 4.9 months lifetime        47 uA
•    2 * 4 sensor readings/minute: 4 months of lifetime       58 uA
•    60 * 4 sensor readings/minute: 11 days of lifetime        641 uA

I calculate it by expression:

I ≈ (C)/(t*k),

where C - accumulator capacity (Ah), 0.22 Ah for CR2032,
t - time in hours (4.9 * 30 * 24 = 3528 for 1 case),
k - coefficient of partial discharging (1.3)

These values based on typical BLE connection point-point. There is no routing load on nodes, so we can think about analogy for Host in 6LoWPAN.
As we understand too this values including power consumption of sensors too.

Can you get us current consumption values for CC1310 without any sensors?

Current consumption for routers is very interesting too, because it's main question for solution.

Best regards.

Hello Adam!

I tried to replicate the experiment, for 6LoWPAN current consumption, on a CC2650 SensorTag. I ran the Very Sleepy Demo example, available in Contiki.
I used a SensorTag as a SLIP radio, Cetic's 6LBR, and Copper addon on Firefox.

The following figure is the waveform I get in response to a GET request, containing a payload of 49 bytes.

I am trying to have a better understanding of the power consumption during various activities and explain the various sections of the capture.

Is my understanding correct?

My main doubt is section D.

The similarity of the shape of the curve's rising edge, to that of Section A, makes me think that the CSMA/CA is trying to send data but failing.

I have also observed through repeated observation, across payload size 1-60bytes, that behavior of Section D is always similar, always 6 spikes.

What am I missing here?

Thank you for your time.

br/Spandan