2
\$\begingroup\$

I am working on a project where I will have 2 or 3 sensors inside a house, all sending data back to a central sensor/hub in a star network configuration. This central sensor would then buffer all the data from others and itself and send it across a LoRa network to a central server once a day. The whole system has to be ultra low-power.

At the moment my plan was to use a non-LoRa wireless network within the home (with a CC1310 or nRF51822) and have one device with LoRa capability (SX1272) to act as the hub. I wondered though, would it be possible to just use LoRa as the main network within the house too and replace the wirelss SoC with a different non-wireless low power microcontroller?

\$\endgroup\$
2
\$\begingroup\$

Yes, it is possible to use a SX1272 or SX1276 to receive data sent by sensors and then use it to transmit data to a LoRaWAN public network while staying very low-power.

All LoRa transceivers have a "channel activity detect" (or CAD) function that can be used to save power and wake a receiver only when an emitter wants to send a message. Unlike traditional FSK transceivers such as the CC1310 that use RSSI, the CAD function is very selective and triggers only when a LoRa signal, with the proper bandwidth and spreading factor, is emitted in the channel. Because it doesn't wakes up every time there is noise in the channel, it can be quite power-efficient.

If you do a CAD detection every N seconds, each transmitter will have to send a packet with a N seconds preamble in order to wake up the receiver. This is a energy trade-off.

\$\endgroup\$
  • \$\begingroup\$ Thanks for the useful insight. With LoRa, however, is there a way to 'pair' the devices so that the gateway device only listens out for a select few other nodes where there may actually be hundreds transmitting? \$\endgroup\$ – Sensors Sep 21 '16 at 8:05
  • \$\begingroup\$ @Sensors in addition to the frequency, a LoRa receiver is selecting for SF, BW and polarity (normal or inverted modulation). If any of this 3 parameters is not matched between RX and TX, the PHY will not detect a reception. It's way more selective than any FSK system. For additional selectivity, use a "group" address at the beginning of the frame, an do some MAC filtering on that. Unlike CDMA spreading, chirp-based spreading doesn't offer a lot of possible orthogonal "codes". \$\endgroup\$ – Sylvain Sep 21 '16 at 16:13
  • \$\begingroup\$ @Sensors moreover, keep in mind that if you use the 868 MHz band, each sensor has a maximum 1% duty cycle, so if you have hundreds of them emitting simultaneously, it means you're in range of tens of thousands of them... not really a problem you should encounter in practice. \$\endgroup\$ – Sylvain Sep 21 '16 at 16:18
1
\$\begingroup\$

Given that LoRa stands for "long range" then adding a few extra metres (size of a house) shouldn't really pose a problem. This website (radio-electronics.com) states: -

There are several key elements of the LoRa wireless system. Some of its key features include the following:

  • Long range: 15 - 20 km.
  • Millions of nodes
  • Long battery life: in excess of ten years

So it does appear to be ideal but the devil is in the detail and you would have to research what size battery would give you the "ten years" and under what environmental conditions. This is crucial I suspect and you may find that the 3 sensors you require will give far more life if they transmit their data (say) once per hour to a LoRa modem/interface that collects that data and sends it on to wherever.

The reason why I suggest this is that LoRa requires has a receiver that consumes several mA and this may need to be activated quite a lot and this will be the main flattener of your battery.

Dumb transmit-only-devices that blindly "send" a few times a day can certainly last for years on very moderate battery sizes.

\$\endgroup\$
  • \$\begingroup\$ The big power cost is keeping a receiver on when you don't know when transmission will come. \$\endgroup\$ – Chris Stratton Aug 30 '16 at 14:49
  • \$\begingroup\$ I was hoping to maybe come up with some sneaky low power timing involved to wake the receiver up at the correct time. I suppose if I was able to have a way of giving all sensors within one house the same unique identifier then it might actually end up more power effective. That might be hard though since they'll be made in bulk and have to be 'plug-and-play', so to speak. I worry that the area receiver/gateway may not handle a few thousand sensors in a square KM too well... \$\endgroup\$ – Sensors Aug 30 '16 at 15:18
  • \$\begingroup\$ I don't share your last concern especially if you use frequency modulation. Because those closest ie those that are relevant will drown those further away. Maybe produce a typical diagram of how you see several houses panning out. \$\endgroup\$ – Andy aka Aug 30 '16 at 16:07
1
\$\begingroup\$

Take a look at the CC1310 Wireless Sensor Network examples (http://www.ti.com/tool/cc13xx-sw). The sensor/nodes remain in sleep mode until they need to wake to send data, so RX power for the sensors is not an issue, there is also the added benefit of the CC1310's Sensor Controler Engine for reducing the system power of monitoring sensors. The RX current will be the driving factor for the Hub/Concentrator, here the CC1310 has a lower RX power than the LoRa transceiver.

You also have the option of using the Wake On Radio feature of the cc1310 (SW example also available on above link) to allow the concentrator to sleep and wake when the nodes send sensor data.

\$\endgroup\$
  • \$\begingroup\$ I didn't know about the wake on radio feature of the CC1310, that's pretty great. Doesn't help me keep the BOM cost down so much using this part instead of what I described trying to make LoRa work as, but I think power is more important. \$\endgroup\$ – Sensors Aug 31 '16 at 8:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.