Recently Nokia strongly communicates its commitment to Open-RAN. They want to be like Tesla or Toyota (in case of hybrid-drives): so good that competition is unable to keep up even after opening their patents. Once telcos grasp the benfits of open-interfaces infrastructure (eCPRI and stuff), there is no coming back to closed ecosystems.

The "radio protocol" is "open" since GSM, anyone can download standards from 3GPP and implement it accordingly. But the mere amount of knowledge and specialized hardware required to do this, even for single layer like L1, is tremendous. I think this is the real reason why we don't already have open-source implementation of the full stack.

EDIT: an afterthought - maybe the O-RAN is really a chance for open-source here. In the future, once O-RAN is accepted and widely deployed, we could work on implementing the stack piece-by-piece, layer-by-layer, filling the gaps with commercial software/hardware as we go, instead of doing everything at once...

The 3GPP standard leaves a lot of implementation freedom in many areas and there is a huge number of patents on everything. So in practice licensing is a big issue and keeps new entrants at bay.

Nokia thinks it is in good shape because without Huawei there is indeed not many threats. Starting from scratch is hugely costly and takes many years. In any case Nokia has no choice but to be "committed" to Open RAN since that's what telcos want.

IMHO, Open RAN is a push by telcos to commoditize the infrastructure and to avoid being locked in because key interfaces are proprietary.

In the UK where Huawei is on a phase-out, Open RAN is a bet by operators to ensure that they have more than 2 vendor options. They are required right now to use 2 vendors, so they are not solely reliant on any one company for their equipment.

Most operators used 2 radio vendors, in order to keep things competitive when buying more. All have exposure to Huawei to some extent in their networks.

If there are only 2 permitted credible radio vendors, then pricing on the "new" second vendor is hardly going to be competitive - where else will you go? You can't go with one vendor, and if you did, their pricing would be ramping up as well, as they know you only have one other choice.

Open RAN presents an interesting insurance policy for operators, and Vodafone UK has even announced it will build a couple of thousand sites using less prominent vendors' equipment. This will get them better pricing going forwards from their existing vendors (as there's an onboarded alternative supplier), as well as give them a new option they might wind up preferring.

Right now though, with the proprietary nature of X2 (in the real world - the standards suggest it's not proprietary, but it absolutely is proprietary), you can't deploy interoperable radios, and that keeps your vendor choice limited. If operators can run a procurement and buy a box from one of five companies, that feels a lot better for their shareholders.

WSB on reddit is pumping it now too. Makes sense.

5G is more of the same telecom monster complex design, but open RANs and open packet core designs are a good step forward.

Also: you know your network has lost its mind when there are 20 extra headers on an IP packet.

Amarisoft and Nexedi also launched a joint company called Rapid.Space last year. It combines cloud and 4G/5G infrastructure operation and is "HyperOpen" = using open-source software, open hardware and open service.

Press release: https://handbook.rapid.space/RS-Hyper.Open.Converged.Cloud.D...

More info on 5G technology: https://handbook.rapid.space/evangelist/RS-Presentation.For....

(disclaimer: I work for Nexedi and help out on Rapid.Space)

as an open source fan I really like your Nexedi products. I find that https://nayuos.nexedi.com/#install-the-image-on-your-chromeb... is missing.

I work at a regional govt in Germany - we want to transition to local cloud - and think this is great product. If these 'NayuOS" can by synced to our own nextCloud installations. Could you help? Thnx

> Amarisoft LTE and NR network software suit is a unique full software LTE and NR solution ... Our binary licenses can be fixed and bounded to a single hardware, floating on a USB dongle, or floating using a license server.

I suppose it's possible that their product includes some Open Source Software, while still using node-locked licenses. But it doesn't sound like it.

This press release [1] from 2020 makes it sound like Amarisoft is working on OpenRAN.

[1] https://apnews.com/press-release/pr-businesswire/56f3c23c33a...

From reading HN it seems like there needs to be open source radio hardware as well to break the dependency on qualcomm, broadcom and whomever else is supplying chips. At least if the idea is to have something auditable.

From what I understand (someone correct me if I'm wrong) part of the problem is that baseband hardware and firmware requires expensive FCC approvals that's more compatible with the profitability of closed-source intellectual property.

It's not so much that it is more aligned with profitability per se, but the fact that each modification you put in the part of the device controlling the radio also needs said approval before being used to transmit - and FCC has been tightening the enforcement of that. It has been on HN before [0]...

Usual caveats about Part 15 vs Part 97 use apply.

[0] https://news.ycombinator.com/item?id=11267443

Do we really need control over the radio hardware? Or can we treat them as dumb pipes, and use something like Tor to shield us from whoever can track who is using a radio channel?

Tor can not prevent a denial of service.

I'm dealing with intel AX200 chipsets (deployed to hundreds of sites only to find out they do not follow intel's own specification for PCIe (it's a 1X board that fails on any socket >1X) and that chipset handles a lot. From AP/Frequency/Channel negotiation one would expect, all the way to WPA3 which one would think to be on the software side, specially because we should trust OS kernel/drivers updates for generics such as WPA to be updated and validated faster than closed source firmware blobs.

Now, I do not know if that is truly done on the firmware or the closed source drivers, because well, the drivers are closed source. But after you accept one binary blob, you might as well threat it all as integral part of the hardware. Who cares where it runs?

i'm getting off topic :) but that is one tiny aspect of a consumer device. One would think that the routing/DPI/etc aspects of 5G hardware for tel Co to be an amplified version of that.

I came exactly to promote this. Thank you.