[tlrobinson]

For this purpose you could just have the cellular module wake up every x minutes to check to see if it should begin tracking.

Or use BLE and an accelerometer. If your phone isn't in range and the bike is moving then starting tracking.

I haven't done the math but you could probably get away with recharging every 6 months or so.

[dpark]

I have to be honest. I'm seriously doubting that math you didn't do. In my experience GPS murders small batteries. And putting large batteries on bikes seems like a nonstarter for bikes that aren't electric assist.

[mikepurvis]

But the GPS only needs to be active when the bike is a) moving, with b) the BLE phone connection absent. There are ultra low power motion sensor ICs that literally have a GPIO for "motion detected". Put that on the reset pin of your CC2541, and you'll be set for months of battery life.

[vostrocity]

Bluetooth LE and motion sensors are actually brilliant, but I think we're treating this too much as an individual problem when we can eliminate bike theft in the first place and not worry about installing trackers.

I've always thought about doing stings from the other direction: with bait bikes. For a bait bike, you just need a bike and a GPS tracker with a battery life of a week or so (a poorly-locked bike will be stolen in a matter of days). The GPS tracker is inserted in the seat tube, and if you glue/oxidize the seat in the seat tube, the thief can't remove the seat to check even if they are smart enough to check for a tracker. You can even add security cameras where you set up the bait bike (outside your apartment or home is a great place) to get additional information. The GPS tracker then gives you a location of either where the thief lives or where they store their stolen property, which lets you bust them pretty easily.

Even in a bicycle-friendly town with 100,000 in population (ala Berkeley), I can't imagine that there are more than a few hundred bike thieves. It would take only a dozen individuals doing bait bikes to bring down bike theft drastically. Combine that with "Bike Batman" style vigilantism identifying stolen bikes on the used market, and I think it's possible to reduce bike theft to near zero in a local area.

[Someone]

You don't care where your bike went, you care where it is, and you want a way to get that data remotely. The GPS only needs to be on when you call your bike, and that's where the problem lies: you have to find a way to decrease power usage of the GSM part.

Switching it on for a minute or so every hour if outside BLE range could help, but would make it harder to locate your bike fast. Maybe, it should start at a 100% duty cycle, and drop off over time? (Clock drift shouldn't be a problem. The device has GPS, so it has an accurate clock)

And you need to filter that 'motion detected' thing, or your bike will keep resetting itself, even before it can check the BLE connection. Or is that something that the sensor ICs can do for you, too, nowadays?

[walrus01]

I am pretty sure you could make a very, very low power watchdog circuit that only wakes up a GPS receiver and 2G/3G modem when it detects 60+ seconds of continual motion. Otherwise it'd sleep at very low power. This would fit in a $50 CF seatpost.

[Natsu]

Couldn't you power something like this from the bike itself somehow?

[dpark]

I am still doubtful but frankly neither informed nor interested enough to debate it further. Maybe BLE is more impressive than I've been assuming.

[tlrobinson]

There are a number of BLE beacons that last for a year on a single CR2302 battery. They only transmit for a fraction of a second at a time, then sleep for 2+ seconds.

[anc84]

That sounds like a great Kickstarter project. ;)

[rorykoehler]

I thought about turning this into a hardware startup. GPS stickers for anything like bikes, guitars etc. Energy consumption was the big issue. Didn't pursue it further because of that.