Hurricane Ian recently made its way through my region of Florida, leaving me without power and a primary Internet connection. The power was restored in about eight hours, which I consider very decent due to my previous experiences with hurricane-related power outages from living in Florida for more than 20 years. However, my primary Internet connection was not restored along with the power. My wife and I work remotely full-time, so this situation is less than ideal during business hours.
Currently, I have access to only a single commodity broadband provider, which is still fairly typical for the majority of the U.S. Additionally, I am on the very edge of coverage for my mobile carrier. I am lucky to get one bar of signal outside of the house. If I stand in precisely the right spot outside and face the right direction, I might occasionally get two bars of signal. One bar is practically useless, and two bars provide basic connectivity but not enough to do full-time remote work, especially when video conferencing is involved.
Due to my network engineering roots, I always consider backups and redundancy. Unfortunately, cost justification always plays a role in the final decisions, whether for business or personal reasons. While having a backup was always on my mind, my cable Internet was extremely reliable before the hurricane, so having a usable backup decreased in priority over time. That priority changed after Ian, though, as I have been experiencing issues much more frequently, including multi-hour outages during business hours. Having some form of backup is now a necessity.
With only a single commodity broadband provider and a weak LTE/5G signal, that usually leaves some form of satellite Internet as the only affordable backup. Unfortunately, satellite Internet is still very expensive! Traditional (geostationary) satellite Internet also has the latency tax, which makes it unacceptable for meaningful video conferencing for work.
The potential lifesaver in this situation is a connection from Starlink. Starlink uses low earth orbit (LEO) satellites, which significantly reduces the latency tax and makes it acceptable for use with video conferencing. The expected speeds seem to be generally good enough for regular remote work as well, as long as you don’t need to transfer lots of very large files back and forth. Unfortunately, my location is on the waitlist.
Starlink recently introduced their “RV” plan where I could get the service immediately at my location, but the tradeoff is an extra $25 per month, plus the traffic is always deprioritized over regular customers. After reading several reviews and considering the overall price, I could not justify spending that much ($600 for the equipment, plus $140 per month for the service) for a backup Internet service. If I lived in an area with no primary Internet option, it would be an automatic purchase for me. After spending a lot of time reading about the caveats of the service, I wouldn’t feel comfortable yet making it my primary method of connectivity unless I had no other choice.
So what am I left with? Although it might be fun to do, I am not ready to start my own business by operating a local Wireless Internet Service Provider (WISP). One other option available to try: an LTE/5G signal booster. An active signal booster works by receiving whatever signal is available, cleaning it up, and re-broadcasting the amplified version locally. I was highly skeptical of this working (garbage in, garbage out, right?), but after reading more about how they work, along with several positive reviews, I gave one a try.
I could not believe it. The signal booster actually worked! The model I got has an outdoor directional antenna, an indoor antenna, and an active (powered) amplifier/signal processor. All components connect with regular RG6 coax cable. Before the booster, I was lucky to get one bar of LTE signal with an average of fewer than 1 Mbps down and 100 kbps up. After the booster, I get 3–4 bars of signal with an average of 25–30 Mbps down and (more importantly) 3–4 Mbps up.
As mentioned, my wife and I both work remotely full-time, so my first test was a simulated conference call with two video streams. I used two laptops with two different hotspots simultaneously with a Webex session recording the call. The video stream connected via the LTE hotspot was flawless for the entire recording. The video stream connected to the 5G hotspot had some occasional delays and dropouts, but still very acceptable for regular work use. So with that test completed, I knew I had a working backup solution that also justified the cost of the signal booster.
On a side note, the signal booster still was not cheap. At just over $500, it approached the cost of purchasing a Starlink dish. However, unlike Starlink, this is a one-time purchase, and it boosts both LTE and 5G signal bands from all major carriers, so it will have some longevity. If there were an additional monthly service cost associated with this, that would have pushed me more toward a Starlink subscription. Likewise, if my mobile plan was more expensive than it currently is for the amount of hotspot data I receive, then that would also factor in the combination of getting a cheaper plan and getting Starlink. Ultimately, Starlink would probably be the better-performing choice for a backup service, but it is more expensive than the route I decided on. I believe it would be worth it if you could justify the cost.
Now that the signal is boosted and acceptable, it’s easy enough to just connect my laptop to my phone’s hotspot capability when necessary. But what fun is that? What about all the other devices in my house? And can I have everything switch over to the hotspot automatically when my cable Internet goes down?
Right now, I am using a Meraki MX appliance as my primary router. All physical Meraki MX appliances support dual WAN uplinks. While some MX appliances support LTE through a USB adapter, I can’t simply plug my phone into the MX and expect to use it as a tethered hotspot. Likewise, the MX cannot wirelessly connect to my phone. (Another side note: in a business context, there are many supported ways to handle this situation, including Meraki’s own MG series of wireless WAN gateways.)
My solution? I have an older MikroTik hAP AC² router facilitating the connection between the MX and my phone to use as an automatic backup WAN connection. MikroTik’s RouterOS is fantastic because it does not limit what features are available on your MikroTik router. If you know what you are doing (and maybe even if you don’t, if you get lucky), you can do just about anything in the realm of networking with RouterOS. This extreme flexibility allows you to devise creative solutions for networking problems. The WLAN Pi project also works for this solution if you have a spare Raspberry Pi.
In my case, I created two bridge devices on the MikroTik. I assigned one of the WLAN interfaces to one bridge, which connects to my phone’s hotspot as a regular wireless client (station mode). I assigned one of the Ethernet interfaces on the router to the other bridge, which connects to the WAN2 port of the MX appliance. The MikroTik obtains an IP address from the hotspot via DHCP and uses the hotspot for the default route. The MX’s WAN2 uplink uses the MikroTik’s Ethernet bridge IP address as its default gateway. The MikroTik also performs NAT outbound so that the phone’s hotspot doesn’t need to know about the MX.
The MX is set to use the cable Internet connection while it is available and automatically fails over to the MikroTik when it is not. I tested the failover by rebooting the cable modem. Automatic failover takes about ten seconds, which is perfectly acceptable for a better-than-nothing home Internet backup solution. All devices on my network simply experience a temporary interruption, and then the traffic resumes. Performance is as expected on a hotspot, with latency slightly increased and bandwidth reduced, but still entirely usable for regular remote work. Fail-back, when the cable Internet becomes available again, is nearly instant.
The Starlink RV service would almost certainly be the way to go if I could justify the additional monthly cost. As mentioned, I was genuinely amazed that the LTE/5G signal booster even worked at all. I would not have thought that a single bar of signal could be cleaned up and actually made usable. However, the signal booster is not a miracle device. It isn’t going to work when there is no LTE signal whatsoever. The Starlink, though, probably will.
When you have limited choices available for Internet access, sometimes you have to get creative, especially for areas that are more rural or underserved. Although I spent a decent amount of money on this project, it could certainly be done for much less cost. If I didn’t have the Meraki MX appliance, I could use the much (MUCH) cheaper MikroTik as my primary router. There are also less-expensive signal boosters available online. Ultimately though, as technology and connectivity improves, these kinds of solutions will become less necessary. In perhaps as little as five years, we may see nearly-global ubiquitous high-speed connectivity, especially if Starlink gains a viable competitor.