A network for free?

This article is from the CW Journal archive.

Someone asked: "What will happen when anyone can build their own mobile network for nothing, how will the mobile operators react?" I must say my initial reaction was very sceptical, how could you build a network for nothing? The premise was that much of a network's function depends on software; almost any software nowadays can be found somewhere in open source and will run on a PC; and PCs cost almost nothing. For example you can find an open source LTE nodeB at https://bellard.org/lte/ and core network at
https://github.com/acetcom/nextepc. So why not? Let's explore why a network for free isn't as easy as it sounds.

Suppose first of all that we get that open source software, running on cheap PC hardware and using off the shelf SDR radio platforms… oops. First hurdle, actually those radios, such as the USRP (https://www.ettus.com/) will set you back a couple of grand each. And they don't come with vital hardware needed to make a base station, such as a power amplifier all the necessary duplexing filters, and power supplies. And they are designed as a lab tool for rapid prototyping and research and don't meet the necessary EMI standards to be allowed to radiate significant power into free space.

Then we need sites, suitably located for reasonable coverage, secure accommodation for equipment, mains power, and backhaul communications. Once upon a time sites were fairly easy to get, but with the sucess of cellular and need to build thousands, landlords and agents realised they had a goldmine – nowadays they are much harder to find, and the charges for access and leasing much higher. Even having found the site its preparation will cost money.

Then there's backhaul. Time was when you could use a consumer-grade ADSL cable when you only needed voice; but for LTE you will need probably 50 Mbit/s both ways, and we know that doesn't come easily or cheaply in the UK. And unless a cable is already provided to the site the operators will charge business tariffs to put the line in and to rent it.

Making networks and keeping them going is a difficult and expensive enterprise

Then you need somewhere for the core network components – all those PCs running free software. This might be easier, but remember you need lots more bandwidth, definitely not a consumer broadband connection. You will need network management, to keep an eye on the network's performance, restart nodes that crash, and update their software.

Oh, and that open source code – how reliable is it? What happens when bugs appear when traffic load builds up – who fixes those? It's no good relying on the open source community to do it for free – they need fixing, testing, and rolling out quickly! So you need some good software engineers at your command – they don't come cheap. And then some new features get specified by 3GPP that Apple and Android phones start to use, and you have to upgrade and test to support those (preferably before the new phones or software updates are launched).

Did I just mention test? Oh yes, so you need to maintain an automated test lab with a model network kept in step with your operating network where terminals can be tested to make sure they interoperate – network vendors and operators do this routinely, testing each upgrade for compatibility. That needs space, equipment, money, and skilled people.

Just when you've got all this in place a busybody from your local spectrum regulator starts asking hard questions about your right to use the spectrum (which has of course to use the standard 3GPP bands to work with the handsets). There is some spectrum available (the old DECT guard band for example) in some countries, but it's far from universal; and it's also licensed.

And of course, since your customer base will be quite small compared to the incumbents, you will need to interconnect to their networks – and you may need roaming agreements since your coverage will be limited in comparison especially when you start. Negotiating those needs skilled people and time, and will attract significant charges once done.

Oh, and talking about licences, you start to get rude letters from IPR lawyers in large network vendors pointing out that your "free" radio and network equipment is infringing a large stack of essential patents – you are of course welcome to carry on but they would like you to pay a "fair reasonable and nondiscriminatory" royalty.

All this of course is just to point out that there is much, much more to building a mobile network than just the basic network costs; and though software is critical to their function, "free" software makes little difference. Making networks and keeping them going is a difficult and expensive enterprise and much of the effort has to be made in the real world of hardware, buildings, roads, organisations, and people.