We're experimenting with using a WiFi module as a battery driven beacon or RFID tag. The cheapest module out there with plenty of documentation (thank you Adafruit) is the Texas Instruments CC3000.

The big problem is the power consumption - 190mA on transmit is very high for a battery driven product - and the demand is for 1 year+ battery life transmitting maybe once every 10-15 minutes. The range is over 150m x 150m with only enough data sent to indicate presence - ie: a beacon.

My question is - how quickly can the CC3000 wake up, transmit and shutdown? The indications are that it takes 60ms to initialise from sleep. Connecting to the Access Point is generally a long drawn out process, at least it is on a Windows machine. I assume that once connected, you could send a broadcast rather than waiting 1000s of milliseconds for DHCP to finish but I don't know if the CC3000 gives you that control.

A posting on TI's forums suggested that it would take 6 seconds to connect to an AP. At 200mA, 4 times an hour you'd need an 11Ah battery for a year; which isn't what I'd call an RFID solution!

(My personal feeling is that this path isn't a reasonable one - if you have to be at 2.4Ghz then there are BLE modules (Nordic Semi) that are more suited but the boss likes WiFi and its range (+18dBm). Anyone have any experience with WiFi in this scenario?)

  • 1
    \$\begingroup\$ I think it's your boss who needs to be asking questions not you!! What range is your beacon intended to work over? What bandwidth etc? \$\endgroup\$ – Andy aka May 19 '14 at 9:15
  • \$\begingroup\$ (edited question to give range...) It's "18dBm is more therefore WiFi is better"! The range is supposed to cover a 150m x 150m space (with metal obstacles), which can be covered by raised Access Points. The bandwidth needs only to be enough to indicate presence which is why I'm thinking the DHCP process can be eliminated. That does assume the AP is a Raspberry Pi or something that can get presence just from a broadcast. I have seen WiFi locator beacons (www.ekahau.com) but they are rechargeable and used in hospitals. \$\endgroup\$ – carveone May 19 '14 at 9:31

Wifi is probably hopeless for this application. You should use a sub-1GHz ISM band radio instead.

The Texas CC1101 and Nordic nRF905 are both quite good. The CC1101 has an edge on power consumption I think. Texas also have some newer parts, such as the CC1120, that are on-air compatible but a bit more sensitive.

In any case using sub-1GHz bands will give you much better range than 2.4GHz. Depending on where you live 433MHz, 868MHz or 915MHz should be available. Modules can be bought cheaply on eBay or Amazon.

As an example a pair of lithium AA cells can run a CC1101 and Atmel microcontroller transmitting every ~5 seconds for about 5 years. To get that kind of performance you need to do a bit of work optimizing your code and hardware. For example, run at the lowest voltage possible with an ultra low quiescent current LDO.

Data rate is important to get right too. If you only need to transmit once every few minutes then the lowest possible (1.2kbaud is reasonable for a CC1101) should be fine. Obviously the lower the bit rate the longer the radio has to stay powered up. You can avoid having to do listen-before-talk by simply randomizing the transmission time by say +/- 1 minute, since you will only be on air for a few tens or maybe hundreds of milliseconds maximum.

Then you need to make a receiver for your PC, or perhaps a bridge to wifi.

  • 1
    \$\begingroup\$ I have both the CC1101 and nRF905 written down here with a preference for the latter, can't remember why. I worked out similar lifetimes using a PIC16 + nRF905 or CC1101 or Hope Module (digitalsmarties.net/products/rfm69cw-868-mhz) with a note to track bitrate vs range vs battery. WiFi bridges are what the RaspPi is for. Cheap as chips. You are right - WiFi is hopeless but I might have to show it's hopeless before anyone will believe me. \$\endgroup\$ – carveone May 19 '14 at 11:34

My personal feeling is that direct WiFi is not the best solution given that you need extended battery life and you are only transmitting minor amounts of data. Another aspect is that your system only appears to be transmitter ---> receiver. No bidirectional comms are needed (from what I can tell) and this will push-down the battery usage.

Given the range and the proposed operating frequency, the free space link loss will be: -

32.5dB + 20log(MHz) + 20log(kilometres) = 32.5dB + 68dB + (-16dB) = 84.5dB and with a transmit power of 18dBm, you would receive -66.5dBm - this is a big signal and even accounting for terrestrial problems it should be easy to receive. Having said that I'd consider going for a significantly lower frequency such as 900MHz - the link loss reduces from 84.5dB to 76dB meaning you can save on transmit power by nearly 10dB.

The amount of data you need to transmit is minor and could probably be achieved at 1kbps. A receiver that only has the bandwidth for receiving 1kbps will naturally have a much higher sensitivity than a WiFi receiver. Receiver sensitivity is governed by this empirical equation: -

Sensitivity = -154dBm + 10log(data rate)

At 11Mbps a WiFi receiver needs a signal that is -154dBm + 70.4dBm = = -83.6dBm

At 1kbps, a receiver needs -154dBm + 30dBm = -124dBm i.e. it will be 40dB more sensitive!!

I'd consider going in with a proposal that uses the lowest frequency band and a reasonably low transmit power to achieve the objectives. Have a dedicated receiver wired to your network and forget WiFi.

  • \$\begingroup\$ Thank you for the free space loss calculations, that's very very useful. I was aware that range is inversely proportional to frequency/datarate, but didn't know the actual numbers - very nice to have those. The dB increase is much larger than I'd expected. I totally agree with your conclusions - my original plan was the EU 868MHz band and 2400bps, TX only (or BLE as you can buy a ready to test solution from Nordic). But I'm being overridden as WiFi is the buzzword. I suppose the advantage of WiFi is inbuilt collision handling but that's doable with a propriety solution. \$\endgroup\$ – carveone May 19 '14 at 10:27
  • \$\begingroup\$ OK, but the problem with Wifi is that the receiver probably needs -84dBm (11Mbps) and it's going to see -66.5dBm in free space - personally, when accounting for terrestrial problems I don't think it will work reliably - I reckon you're going to be 20dB short of a decent system. \$\endgroup\$ – Andy aka May 19 '14 at 10:46
  • \$\begingroup\$ Also, will a standard WiFi box be able to send power sufficient to interrogate the modules? \$\endgroup\$ – Andy aka May 19 '14 at 11:02
  • \$\begingroup\$ From what I've seen reliability-wise, yes, I believe we'll have to shorten the range to 50m, nominally by using more Access Points. At least they are cheap these days. But we'll have to see. Thanks Andy. \$\endgroup\$ – carveone May 19 '14 at 11:18

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.