Fabricd Bellard is a rare person who seems to do and understand everything. I often just grab a cup of coffee and read through the code for random projects[0] of his.
This weeks post covers the same thing as this post - how easy/difficult is it to actually set up a private 5G network compared to a similar WiFi deployment.
Thanks, yes I'm subscribed to the newsletter as well, what a surprise coincident when posting about open 5G on HN and receiving article exactly in the same subject later on the same day.
Really interested to see if this will get picked up by network operators and how.
In my experience, corporate buying at this level has always been the home turf for salesmen of entrenched companies, partly because they actually pitch better and partly because buying StandardCorpX's product comes with a level of delegating design decisions you could get blamed for to the supplier. That last one typically doesn't come to the same extent with open source.
Not that I'm condoning any of these corporate buying strategies, just stating the fact.
Network operators are the ones driving O-RAN - they have the most to gain: cheaper equipment & more supplier options. However interoperability is not as easy as you might hope, even with an open standard.
>However interoperability is not as easy as you might hope, even with an open standard.
Came here to say this. Rather than being all interoperable / plug and play, you wind up having to validate different combinations of hardware from different vendors... which then has to get revalidated when new versions of software are rolled out.
I’ve set up an LTE network in my house before and while it is a fantastic educational tool, that is pretty much the extent of the benefits offered vs WiFi.
With lower frequencies comes much lower bandwidth. 5G networks can transmit at much higher power than WiFi, so that's where you get your range at higher frequencie - however that requires licensed spectrum which makes all this a bit harder.
Maybe I've just been out of school for too long, but I can't think of anything particularly interesting a campus would want an IoT network for beyond maybe motion sensing for automatic light turn-offs and run-of-the-mill security systems.
I can't imagine a low-frequency IoT network being good for broadband access for people walking around (a la wifi) because of bandwidth limitations.
"campus" can mean lots of things... your handle is "vineyardmike" - what about irrigation / water / temperature / other sensors spread over a vineyard?
I used to attend O-RAN meetings for a major telecom, and this was always the major hurdle I envisioned for the adoption of these things. On the technical side, it's a fun problem and interesting work, but O-RAN and open source cellular lacks the organic growth other open source projects can enjoy. A comparison point was always the networking space. Open source routers/switches broke into the Juniper/Cisco market, but it doesn't work as well in cellular. Small businesses can buy 1 or 2 open source switches and in aggregate that can start to change the market. No small business is going to own their own spectrum and deploy an open source cell network. And moving to shared spectrum eliminates most of the advantages over WiFi.
It can be used at small scale like home wifi (small cell / femtocell) and it would probably be on par with regular wifi. Cellular networks don’t really stand out in home-sized few-user deployments though because they’re designed for serving large areas or high user densities.
The main advantage should be proper base station handoff. If you have a house that can't be covered by a single wifi access point (or an office), then the normal approach is multiple APs with the same SSID. It's a bit of a hack though, you can't guarantee that your device is connected to the one with the best signal, particularly if you move around, and transitioning involves a connection drop.
Proper cellular radio does measurement of signal to surrounding basestations, and seamless handoff. If would have some advantages over wifi in a corporate environment.
This is a bit outdated -- 802.11r fast transitions (FT) allow devices to roam between access points nearly instantly, and 802.11k and v allow access points and clients to exchange scan results, so that access points can move clients to the most appropriate access point.
This does require more than just having multiple access points with the same SSID, and used to be limited to expensive commercial AP brands, but it's quite accessible to consumers these days.
Doesn't 5G share some of the frequencies used by normal wifi? IIRC devices implementing one standard even cooperate with devices implementing the other to coordinate frequency use.
There is CBRS, which some phones are equipped to use. However, I believe it’s used only in conjunction with licensed bands rather than independently. It’s more of a capacity thing rather than a way to have a carrier using unlicensed spectrum (carriers are quite happy with the license barrier to entry).
short answer is no, but the frequencies aren't far off
wifi operates at 2.4Ghz most typically and also at 5Ghz (shorter range but faster)
5G will be ranges of frequencies, depending on jurisdictions, but typically between these two frequencies which are the sweet spot for domestic usage
in the US, 5G carries AFAIK have ranges between 2.5Ghz and just under 4Ghz
in the UK we have EE, O2 and Three operating ranges around 700Mhz besides also having ranges around 3.4Ghz and 3.6Ghz were Vodafone operates too, and those work a lot like US 5G
4G in the UK operates under 2.6Ghz, which is pretty close to wifi also, and LTE is just over 850Mhz or so
basically wifi, 4G and 5G are not strictly higher or lower frequency to each other, it depends on local operators
It always amazes me how pulsing some electricity into a wire at a specific frequency can get you arrested lmao. I really don't understand why there's any legislation at all covering anything above 1GHz where there's enough bandwidth for literally everyone to do whatever they damn well want and not enough range to do any damage at all.
Has anyone ever seen a software defined radio based phone/client out of interest?
Also I've not properly read the paper yet but it mentions a VCTCXO (Voltage Controlled Temperature Compensated Crystal Oscillator), used with some of the SDRs. A while ago I bought a rubidium clock, which I thought may have been decommissioned from a cell tower. I'm curious how important the clock used is, for a cellular system?
> I'm curious how important the clock used is, for a cellular system
Very, that's why they bother with things like rubidium holdover clocks. A lot of things depend on precise synchronization, both locally and between multiple towers.
Thanks, that makes sense, curious how they seem to get away with a VCTCXO in this case (or at least I couldn't seem to find mention of rubidium clocks in addition).
Different qualities: Quartz crystals are better than Rb when looking at short periods. Long-term stability is normally provided by GPS. Rubidium beats crystals when it comes to stability over hours - it is added as a fallback for when GPS fails. If you don't have a requirement that your system can work lets say 24 or 48 hours without GPS sync, you can skip the Rb.
I too have some kind of GPS-disciplined high accuracy clock (I don’t recall the exact tech but I think rubidium sounds right). From what I recall, precision timing is quite critical on the infrastructure side of things so that the user device can sync to it and accurately transmit and receive in the correct TDD windows.
One of the niftiest things was using the 10MHz reference clock as the clock source for my SDR, and tuning the receiver to the centre carrier of a nearby cell tower. I got a ~0.25Hz beat frequency out of it, which means this cheap eBay clock is accurate to about 1/3.2e9 Hz (cell tower was around 800MHz). Neat!
[1] https://www.amarisoft.com/ [2] https://bellard.org/lte/