I've had a technology problem I've been unable to solve for almost a decade.
Our family holiday home is in an extremely remote location, an island to be precise, off the coast of New Zealand's South Island
. The property is completely off-grid as there is absolutely no infrastructure whatsoever.
The challenge -in which I have singularly failed- is to establish a persistent Internet connection and until now, the only viable option has been cellular. By precisely tuning a 24dBi antenna by hand to point at a rock at the top of the range opposite our place I can obtain just a slither of cellular connectivity from the mainland. Like this:
Which is good enough when I'm in residence but not enough to establish a stable connection when I'm not. To resolve this I've tried numerous amplifiers and different combinations of antennas, to achieve the outcome described here
. All of which have failed.
Because Internet connectivity is sporadic our systems have evolved by necessity to be largely autonomous. IoT sensors and actuators
operate according to the logic enforced by Home Assistant
. The solar electricity system is managed by the ICC
package on Raspberry Pi, utilising serial connections to our Inverter and Battery Monitor, and managing appliance consumption with Sonoff electricity control devices
. Time (one doesn't realise how important accurate time is until configuring unmanaged off-grid systems) is provided by GPS.
And so on. Right now the topology is supposed to look like this:
To be fair, whilst I'm in residence this is largely the case. The rest of the time, which is most of the year, only those autonomous components and systems operate nominally because I'm not there to manage the rest.
But all that is about to change. At least I'm hopeful it is.
Last Saturday Elon Musk's Starlink
service informed me that I'd been chosen to participate in the New Zealand beta programme, and I'm expecting delivery of the satellite dish next week. I'm extremely hopeful it is going to solve all of my persistent Internet connectivity problems and, appealing very much to my sensibilities, simplify my topology.
But the Starlink service is not without its challenges. Here are some of those I'm considering before the unit is even delivered:
- Cost. At $NZ150/month, Starlink isn't altogether cheap for a holiday home that isn't occupied all that often. The solution is to share the connection with those in proximity, by connecting a 13dBi RT2800 antenna/dongle such as this one to Raspberry Pi, then providing guest access with a Captive Portal solution such as NoDogSplash.
- Electricity consumption. The Starlink unit consumes ~100 Watts continuously, about the same amount as an old-fashioned fluorescent lightbulb. This doesn't seem like a lot unless your holiday home is off-grid and over-depletion of your battery storage becomes a serious issue after a few cloudy days. The solution is to monitor battery storage capacity with the above-mentioned ICC and power on/off the Sonoff switch Starlink is plugged into via the home automation software, Home Assistant.
- Inbound Connectivity. Starlink uses CG-NAT so there isn't a static IP address assigned one can connect to, in order to access our holiday home from elsewhere. This is trivially easy to work-around by establishing an outbound VPN connection to a server I control then routing inbound connections along that established tunnel.
- WiFi Calling. Conceptually, WiFi Calling is a solution whereby cellular providers recognise that a cellphone is outside the cellular coverage area but is connected to WiFi, so route calls and messages over that network instead. Unfortunately not all cellular providers support this type of handover and not on all cellphones. Not sure how I'm going to solve for this.
In summary I've expended a lot of time, effort and money endeavouring to get cellular to work for our particular use case and I've failed. I'm not a particular fan of Elon Musk but I'm cautiously confident his Starlink service is going to solve my headache.
-SRA. Auckland, 29/vii 2021.