5 ms·

Definitely, and I think there are even 30 km links at E-band. What concerns me is fiber build-outs are stalling, thinking that wireless is going to save costs.

Definitely, and I think there are even 30 km links at E-band. What concerns me is fiber build-outs are stalling, thinking that wireless is going to save costs. It’s not going to happen, at least not the performance of a fiber to every home. Electricity and twisted pair could be run to every home, but now it’s too costly.

walrus01 8y ago
Realistically you can't go even 8km reliably in e-band, no way for 30... 30 at traditional frequencies like 11GHz, sure. But the atmospheric attenuation is extreme at 71-86GHz, as is the rain fade. Even with +15-18 tx power and 52db gain antennas.
- madengr 8y ago
  I swear Facebook claimed 10G at 30 km, and recently Bridgewave was claiming the same. Facebook was using a custom 33 dBm PA MMIC. Quorvo (or maybe HRL) presented a 33 dBm GaN MMIC a couple years ago, but it was ~90 nm GaN process which isn’t out yet. Once that’s out, I’d expect >1W at 80 GHz to allow some cool stuff.
  - walrus01 8y ago
    The hard part is that increased tx power beyond about +18 also increases noise - 33dBm might work for OOK or BPSK modulation with a 5000MHz wide channel each way, FDD, but the commercial 10Gbps radios available now use from 64-128-256QAM to achieve 10GbE in reasonable channel sizes. Using all 5000MHz works, but limits the number of radios you can Colo on one roof. Ever seen the first generation of Bridgewave and gigabeam 71-86GHz radios? OOK modulated for 1GbE. It's also possible FB's tests are using channels even wider than 5GHz with STA from the FCC.
rayiner 8y ago
How long will it be before typical households need the performance of fiber? I’ve got 4 gig symmetrical to my house (3 gig Comcast, 1 gig Verizon). I had a wiring issue where my desktop’s ethernet port was syncing at 100 mbps, and I didn’t even notice. The Web can’t even really use the bandwidth—the TCP windowing algorithm takes too long to ramp up. 100 mbps is plenty to stream 4k on the TV plus 1080p to a couple of iPads. Even games download plenty fast at 100 mbps. There is the cloud, but I can back up my entire iPhone to the cloud in less than an hour at 100 mbps—doing it incrementally, it doesn’t even register. There is a cost to doing fiber build outs before there is a real demand for it. If people don’t need more right now than what cable and wireless can offer, you might wait a very long time to recover your investment.
- madengr 8y ago
  The cost is in the horizontal drilling, not the actual fiber equipment. Why it could be done “then”, with electricity and twisted copper pair, but not now, is what irks me. I could use symmetrical 10G right now for software defined radio experimentation. I presently have 100/5, and it’s the 5 that is the issue. Without fiber, we’ll be stuck with horrible asymmetry.
  - rayiner 8y ago
    > with electricity and twisted copper pair, but not now, is what irks me Electricity and telephones brought immediate, game changing benefits, and there was no alternative that was almost as good. Fiber still doesn’t offer “must have” benefits over existing alternatives for typical consumers. Maybe in the future with VR and whatnot, but not yet. The demand for faster wired connections has stalled even on the LAN. Apple added GigE to the PowerMac in 2000, a year after the standard for 1000base-t was ratified. It’s been more then a decade since 10Gbase-t was ratified, and Apple just added it to a workstation this year. There has been almost no uptake outside the server room. Lack of demand means that it’s hard to draw consumers away from existing options. FTTH uptake rates are below 50% (meaning less than half of homes who can get it actually subscribe, or put another way you have to build fiber past two or three homes to get one paying subscriber). On the supply side, anything involving human labor is relatively more expensive compared to other inputs than it was then. Minimum enlisted military pay at the time of the rural electrification act was $9,000 per year (adjusted for CPI inflation). It’s double that today.

walrus01 8y ago

Realistically you can't go even 8km reliably in e-band, no way for 30... 30 at traditional frequencies like 11GHz, sure. But the atmospheric attenuation is extreme at 71-86GHz, as is the rain fade. Even with +15-18 tx power and 52db gain antennas.

madengr 8y ago
I swear Facebook claimed 10G at 30 km, and recently Bridgewave was claiming the same. Facebook was using a custom 33 dBm PA MMIC. Quorvo (or maybe HRL) presented a 33 dBm GaN MMIC a couple years ago, but it was ~90 nm GaN process which isn’t out yet. Once that’s out, I’d expect >1W at 80 GHz to allow some cool stuff.
- walrus01 8y ago
  The hard part is that increased tx power beyond about +18 also increases noise - 33dBm might work for OOK or BPSK modulation with a 5000MHz wide channel each way, FDD, but the commercial 10Gbps radios available now use from 64-128-256QAM to achieve 10GbE in reasonable channel sizes. Using all 5000MHz works, but limits the number of radios you can Colo on one roof. Ever seen the first generation of Bridgewave and gigabeam 71-86GHz radios? OOK modulated for 1GbE. It's also possible FB's tests are using channels even wider than 5GHz with STA from the FCC.

rayiner 8y ago

How long will it be before typical households need the performance of fiber? I’ve got 4 gig symmetrical to my house (3 gig Comcast, 1 gig Verizon). I had a wiring issue where my desktop’s ethernet port was syncing at 100 mbps, and I didn’t even notice. The Web can’t even really use the bandwidth—the TCP windowing algorithm takes too long to ramp up. 100 mbps is plenty to stream 4k on the TV plus 1080p to a couple of iPads. Even games download plenty fast at 100 mbps. There is the cloud, but I can back up my entire iPhone to the cloud in less than an hour at 100 mbps—doing it incrementally, it doesn’t even register. There is a cost to doing fiber build outs before there is a real demand for it. If people don’t need more right now than what cable and wireless can offer, you might wait a very long time to recover your investment.

madengr 8y ago
The cost is in the horizontal drilling, not the actual fiber equipment. Why it could be done “then”, with electricity and twisted copper pair, but not now, is what irks me. I could use symmetrical 10G right now for software defined radio experimentation. I presently have 100/5, and it’s the 5 that is the issue. Without fiber, we’ll be stuck with horrible asymmetry.
- rayiner 8y ago
  > with electricity and twisted copper pair, but not now, is what irks me Electricity and telephones brought immediate, game changing benefits, and there was no alternative that was almost as good. Fiber still doesn’t offer “must have” benefits over existing alternatives for typical consumers. Maybe in the future with VR and whatnot, but not yet. The demand for faster wired connections has stalled even on the LAN. Apple added GigE to the PowerMac in 2000, a year after the standard for 1000base-t was ratified. It’s been more then a decade since 10Gbase-t was ratified, and Apple just added it to a workstation this year. There has been almost no uptake outside the server room. Lack of demand means that it’s hard to draw consumers away from existing options. FTTH uptake rates are below 50% (meaning less than half of homes who can get it actually subscribe, or put another way you have to build fiber past two or three homes to get one paying subscriber). On the supply side, anything involving human labor is relatively more expensive compared to other inputs than it was then. Minimum enlisted military pay at the time of the rural electrification act was $9,000 per year (adjusted for CPI inflation). It’s double that today.