> I only recently understood that my Coax cables are basically like an RF spectrum similar to Wifi except inside the cables. It appears this is taken as obvious because I had a really hard time finding someone explain it like that.
Back in the days nearly everyone had the experience of connecting an outdoor antenna to a coax cable.
BTW, you can even run Wi-Fi over a random, abandoned RG-58 TV coax in the wall by simply connecting a suitable coax adapter. Some will mention the problem of impedance mismatch, but it's negligible [0]. The downside is that MIMO is not possible, only 1T1R is possible, and 150 Mbps is the best possible speed for 802.11n (but I wonder if it's possible to use a mixer at both ends to convert the second antenna port to a different frequency, so both bands can operate simultaneously in a single coax cable with a duplexer for MIMO).
> I had trouble understanding "how is this secure" when my modem is also connected to the same coax and this is clearly not behind my firewall. Well, of course, you filter out the signal from leaving through the Coax which provides you cable!
A Coax-to-Ethernet modem and a Cable modem doesn't even have the same analog front-end or use the same modulation, software-only attacks are practically impossible. It can only be a problem if the attacker has physical access to the coax cable, or if the coax leaks (which always occur). But then, G.hn uses AES encryption with Diffie-Hellman key exchange [1][2], you shouldn't worry after all. Man-in-the-Middle attacks can still be a problem, it's more secure to set the password (a.k.a. Device ID) manually via Ethernet instead of using the automatic pairing. But still, it's unlikely that anyone is injecting traffic during the initial pairing.
[0] A 50-ohm transmitter into 75 ohm ideal coax has a VSWR (a metric of impedance mismatch) of better than 1.6:1 at 2.4 GHz, real cable loss can make the apparent VSWR even lower.
>> I only recently understood that my Coax cables are basically like an RF spectrum similar to Wifi except inside the cables. It appears this is taken as obvious because I had a really hard time finding someone explain it like that.
>Back in the days nearly everyone had the experience of connecting an outdoor antenna to a coax cable.
>BTW, you can even run Wi-Fi over a random, abandoned RG-58 TV coax in the wall by simply connecting a suitable coax adapter. Some will mention the problem of impedance mismatch, but it's negligible [0]. The downside is that MIMO is not possible, only 1T1R is possible, and 150 Mbps is the best possible speed for 802.11n (but I wonder if it's possible to use a mixer at both ends to convert the second antenna port to a different frequency, so both bands can operate simultaneously in a single coax cable with a duplexer for MIMO).
This reminds me of a hack we did at work. We needed to connect the networks of two of our labs without going through the official IT equipment (don't ask). The difficulty was one of the labs is a Faraday cage, so essentially no accessible ways of routing cables in and out. However, we do optics research so we had some fiber connections between the labs, they are a limited resource though. What we ended up doing was connecting a media converter to use the fiber connections. But ethernet over fiber uses separate fibers for duplexing and the fibers between labs are a limited resource. We ended up using a circulator (which essentially separates the forward and backward transmission from a fiber into two fibers), and used a single fiber for full-duplex without issues.
Take home message, if you have enough SNR (the media converter bridges are designed for 15 km) you can get away with all sorts of hacks.
> We ended up using a circulator, and used a single fiber for full-duplex without issues.
Great hack.
BTW, I always want to see if a circulator can be used as a Wi-Fi booster. A typical Wi-Fi chipset has a noise figure around 6 dB, in principle, if a high-performance LNA is used, connected via a circulator (so the Wi-Fi doesn't transmit into the amplifier), 3 or 5 dB improvement for RX should be possible (and without violating any radio regulation!). But I don't have to tolerate the horrible campus network since then, I don't have any motivation to try it anymore...
> Take home message, if you have enough SNR (the media converter bridges are designed for 15 km) you can get away with all sorts of hacks.
It's said that all communication accomplishments are ultimately a problem of adding a bunch of positive and negative numbers in your link budget. ;-)
Back in the days nearly everyone had the experience of connecting an outdoor antenna to a coax cable.
BTW, you can even run Wi-Fi over a random, abandoned RG-58 TV coax in the wall by simply connecting a suitable coax adapter. Some will mention the problem of impedance mismatch, but it's negligible [0]. The downside is that MIMO is not possible, only 1T1R is possible, and 150 Mbps is the best possible speed for 802.11n (but I wonder if it's possible to use a mixer at both ends to convert the second antenna port to a different frequency, so both bands can operate simultaneously in a single coax cable with a duplexer for MIMO).
> I had trouble understanding "how is this secure" when my modem is also connected to the same coax and this is clearly not behind my firewall. Well, of course, you filter out the signal from leaving through the Coax which provides you cable!
A Coax-to-Ethernet modem and a Cable modem doesn't even have the same analog front-end or use the same modulation, software-only attacks are practically impossible. It can only be a problem if the attacker has physical access to the coax cable, or if the coax leaks (which always occur). But then, G.hn uses AES encryption with Diffie-Hellman key exchange [1][2], you shouldn't worry after all. Man-in-the-Middle attacks can still be a problem, it's more secure to set the password (a.k.a. Device ID) manually via Ethernet instead of using the automatic pairing. But still, it's unlikely that anyone is injecting traffic during the initial pairing.
[0] A 50-ohm transmitter into 75 ohm ideal coax has a VSWR (a metric of impedance mismatch) of better than 1.6:1 at 2.4 GHz, real cable loss can make the apparent VSWR even lower.
[1] https://en.wikipedia.org/wiki/G.hn#Security
[2] https://en.wikipedia.org/wiki/X.1035