The run time is 37 minutes.
Additionally the full transcript is below. To distinguish between us I've indented David.Transcript
Good afternoon, David, how are you doing?
Pretty good, and yourself?
I'm all the more fantastic for being able to eventually connect with you, after a few tries there. You are dealing with an open Source BTS. Can you tell me something about what we mean by open BTS? If we imagine a BTS, we're imagining an antenna. We are imagining large, physical, costly hardware, and something which propagates radio signals. Can you tell me what open BTS is?
Yes, I guess what you're describing there, in terms of the big antennas and what they call the civil infrastructure - that's the site. That's the BTS site, but the core of the BTS site is a piece of equipment, the Base-station Transceiver Substation. That is the piece of equipment that actually generates the network air interface, the interface for the cellular network. Everything else around the site exists to support the BTS.
What open BTS is, is it's an open source implementation of the cellular air interface, for GSM. It's a piece of software that runs, uses a piece of open source hardware, called the USRP, the Universal Software Radio Peripheral, and with the USRP and this open source software, we can generate an air interface, that to a cell phone, looks just like any other GSM cellular network.
On the network side, it looks like an Asterisk server. It is an Asterisk server on the backside. What we're doing is putting a cellular air interface on an Asterisk PBX.
Okay, so this air interface, I want to say, "Hell, is it complete?" Are you actually, in software, supplying the whole DTAP or BSSAP app stack?
Yes
The whole signaling stack. I mean, you're supporting GSM so you're supporting the Base Station Subsystem Mobile Application Part (BSSMAP) and DTAP at our transfer application part?
Well, we are supporting a functional equivalent to that, but that's not how we're doing it.
I just wonder what you're doing because if I send a text message from a mobile phone, then it's going to go over the interface and it's going to send a DTAP message, or when I make a call, it's going to send a DTAP message over the air interface, a set up signal similar to Q.931.
That's right, so at layer 3, GSM call control is very much like ISDN. Most services, at layer 3 and GSM are very ISDN-like. What we've implemented is GSM air interface in layer 1, GSM air interface LAPDm in layer 2, and then what we have in layer 3 is very much like ISDN SIP gateway. We take all the ISDN-like services that GSM performs in layer 3 and we translate them into corresponding SIP services.
Okay, an air interface site. Since it's a standard mobile, it's going to be your ISDN-like layer 3 signaling. It is going to conform to the GSM spec on the air interface site, right?
That's right.
So, you're supporting your layer 3 setups. Is it full support that you have? Do you even support the text message signals?
We do, now, yes. The two things we support, right now, are speech telephony, Q.931-type call control, and text messaging. Those are the two services we support, right now.
Okay, how many people are behind this? The reason I jumped into the technical deep end a bit quickly, there, is I'm surprised at that degree of engineering, and with the name open. It's pretty costly to engineer those. How many people are behind this, and why?
It's not that complicated. That's the big myth [laughter]. The hard part of GSM is understanding the specification. Once you understand the specification, it's not that complicated. We have had three people working on this project for about two years, well, for about eighteen months of actual coding. We started coding this stuff in August of 2007.
We had all had prior experience with GSM systems, all in a signals intelligence context, not in a telecom context. The system we have right now, we do call controls, text messaging; we have a 3-layer stack, as I said. The third layer is like an ISDN SIP gateway. We let Asterisk do everything else.
To do that, we wrote about 10,000 lines of C++. It's not a terribly complicated system, once you have a good mental picture of how it fits together. We used a flow-based programming model. We coded it using a nice well-planned object hierarchy. Granted, the object hierarchy was informed by many years of prior experience in intelligence systems. It wasn't that complicated. That's the funny part. The hard part is understanding the spec. Once you understand the spec and you have a clear mental image of how these systems - how these protocols fit together and how the layers interact with each other; it's actually not that hard to build one.
You've kind of just hit the nail on the head. That's where I'm finding it amusing. It takes quite a long time to get to the position that you're speaking about. It's known as a high degree of expertise. Then, you're kind of giving the word "open". To me, it sounds pretty revolutionary.
Normally, you are talking about the likes of Alcatel or Nortel, and so on, who are offering such stacks and facilities and at high cost because it's considered high engineering talent. I like the way you say, "Hey, it's not that difficult, once you understand the spec," but the whole difficulty sn getting to that level of where you do understand the spec - if you take the spec we're talking about, the layer 3, it means you have to understand everything else. To understand the DTAP, the Q.931-type signaling, you had to strip it out from the SCCP, for example.
Well, that's the funny thing. I guess that maybe part of it is that coming from the signals intelligence background, you tend to look at the rest of the network as a black box. All that's exposed to you is the air interface. That's all you see. Since all you see is the air interface, you tend not to think about all the complexity in the network.
As I said, the people involved in this project, my business partner and myself [0:08:27.5 unclear name], we both started working in cellular signals intelligence in 1999. We've been doing this for about ten years, now. For the first seven of those years, I never even gave much thought to network structures. I built a mental model of how the air interface worked, but wasn't really informed by a lot of understanding of things like HLRs and MAP and TCAP and all those types of things.
When we started to first take a crack at building our own BTS, and we started thinking about network integration, and we started looking at all the components; even in a phase 2 GSM network, which is really simple compared to a lot of stuff that's getting rolled out right now; we were horrified. We were horrified by the complexity because a lot of that junk really didn't seem to need to be there.
You ought to be able to build a peer-to-peer telephone system [0:09:50.0 audio breaks].
Based on what? You broke up, slightly there. You should be able to build it on what?
Ironically, I said based on VoIP protocols. [Laughs] A lot of that stuff - if you rethink the way the telephone network works, a lot of that stuff you don't really need. You can work around it; you can build much simpler peer-to-peer mechanisms; you can leverage tools like Asterisk and simple email servers, tools like [0:10:40.1 unclear], and it just doesn't need to be as complicated as it's made out to be. We implement in about 10,000 lines of code.
Just a quick tech question, there. Are you going to offer a CDMA equivalent, ANSI- 41?
Not in the foreseeable future, no. We're concentrating on GSM. There are some very specific reasons that we really like GSM. We're not terribly interested in CDMA because our original goal was low-density areas in the developing world. For those areas, CDMA is not an appropriate air interface. It's also complicated to implement. The computational load to implement CDMA in a software radio is much higher than for GSM, so we're not really looking at that.
Okay, now this is the first time I've spoken with you. I'm beginning to wonder if you saw one of the talks on the debut Emerging Communications Conference, the one with our Swedish speaker. Did you see that at all, where he was speaking about peer-to-peer GSM networks?
No, I didn't. I wasn't aware of that. I'd love to.
The overlap here is pretty incredible because you're talking about developing nations and that's where he was seeing the demand. Anders Carlius spoke at '08. His video is actually online.
I will have to go see it. I wasn't aware of this; that sounds great.
I remember, I was looking at the BBC news and I saw a company which was developing peer-to-peer GSM. Obviously, it would be illegal in developed countries. I thought, "Hell, that technology could be stolen and somehow used in developed countries, in some other ways." I'll link to it when we do the transcript of this, and I'll also ping you offline about that.
What is it you actually want to do with this project?
Our goal right now is to find a sponsor for a pilot deployment, of maybe six to ten cell sites, in a rural area, in the developing world. We've talked to people in Africa, in India, and in Central America. We still haven't had anything come together yet, but that's really - our goal right now, is to set up a pilot program. Once we can set up a pilot program, we believe we can overcome some of the carrier resistance to this approach.
Initially, when you present this kind of thing to an incumbent GSM carrier, they would consider it as something for green field deployment, but the truth is; there is not a lot of green field left out there. Only in the most desperately poor areas, and they've pretty much given up on a lot of those markets. Areas that don't have GSM service right now, probably won't get [0:14:47.3 audio skips] deployment.
To get a carrier that is accustomed to running conventional, hierarchical MSC-oriented network, to get that carrier to look at the kind of network we're talking about, it's sort of a leap of faith, on their part. We're hoping that by setting up a good pilot deployment, we can prove this is a viable technology and we can get carriers to come and see the pilot network and maybe give this kind of technology some more serious consideration.
What I want to know, at this point, is I can see what you're doing is really technically - I think it's awesome what you're doing, just from the little I've heard from you. I think it's very nice; I'm trying to be polite in my choice of words, here. [Laughs] Apart from seeing it technically very neat and very hackerish, what I don't understand is what is the value proposition, speaking to incumbents? Why would the ever want to go near it?
The gist of it is that the type of network we're talking about is deployable at much lower costs than conventional GSM systems. Because it can be deployed and operated at much lower costs, you have the potential to push the cost of service down into the $1 a month range. We know, from public records and interviews with former employees of cellular carriers in Africa, we know the actual cost of service in Africa is something on the order of $5 or $6 per month, per subscriber. We believe in those same environments, the type of network we're talking about can be operated at an order of magnitude lower cost. You could have profitable operations at around $1 per subscriber.
What that does is it allows you to address the last remaining market, which is the three billion poorest people on Earth, who will not get telephone service otherwise. They will not get telephone service at $6 a month because they simply cannot afford it. There are too many of them to do it as a charity.
From a commercial standpoint, is that a great value proposition? It probably depends on how hard your regulators are leaning on you for universal service.
It sounds to me like not only can you address the issues in poorer markets, but going by what you're dealing with there, this open source air interface, which appears on the other side as an Asterisk server, to me, it sounds like you could build some very neat applications and have some innovative development around it, let alone just cost saving. It sounds to me like it could be some nice infrastructure, if we call it that, or platform to build some neat applications around. Don't you see potential that way, as well?
Yeah, we can see the opportunity that you could do a lot of creative things with text messaging and with calling services that you might not be able to do in an ISDN-type world, standard GSM-like ISDN-type world. Also, the technologies also are applicable - there are other places outside the developing world where this technology could be applicable, isolated sites like oilrigs, and ships, and those kinds of things.
Another aspect of this technology that is attractive is that it is very rapidly deployable in the absence of any meaningful infrastructure. If you need to deploy temporary cellular service in the wake of a natural disaster, like a hurricane or major earthquake, it lends itself to that very readily. There are other [0:19:46.2 unclear] applications, services on airliners, campus-type... there are a lot of niche markets, too. These are all places where we're actually more likely to get development money than going for the developing world. In the end, our big vision is still the developing world. Those other applications are just a means to an end.
When you say it's open BTS, does that mean somebody can go and download the source code?
Yes, there is a distribution of open BTS available from GNU Radio, right now, available for download from GNU Radio.
Why did you decide to make it open source?
From the very beginning, we had always wanted an open source element to our project. I think part of the thinking of that, frankly, was that we wanted to create a sort of movement mentality around this technology. We were a little inspired by other open source projects that we've seen in this area, like Asterisk and OpenMoko. We were really kind of inspired by those projects and we were hoping to create a kind of movement sort of mentality around Open BTS, as it evolves.
How long has open BTS become available? I only became aware of it, myself, about seven months ago.
That's about when it became available. We made our first open source release in September of 2008.
So you guys are pretty fresh.
Yeah, like I said, we started actually building the software a year earlier than that, but we didn't actually start releasing things in open source until we had enough of a system to be useful and functional. We didn't bother releasing anything in open source because your technical nature of the software is such that if it's released in an incomplete state, unless you're signal processing engineer with a lot of experience, there is nothing useful you would be able to do with it. It's not like something somebody could tinker with at layer 1. Nobody can tinker with it until it's complete - until it's talking to Asterisk, it's not really a tinkerable project. [Laughs] That's why we didn't release anything until it was fairly close to complete.
I seem to recall it was about three months ago, or maybe two; time flies and I can't remember. Someone pointed out to me a lawsuit going on against - I don't know if it was you, named personally, but it was certainly to do with open BTS. The quick look I took at it - are you able to comment on what this lawsuit is about?
I can tell you some. The lawsuit is being pressed by a former consulting client of ours. When I say ours, I mean Kestrel Signal Processing, Inc., which is the consulting company that my partner and I have, under which we've done open BTS, and a lot of prior consulting work.
Kestrel had built a GSM stack for an intelligence product for this small defense contractor. Later, when we started writing open BTS, he made assertions that our work in open BTS was somehow based on proprietary intellectual property and trade secrets from this prior intelligence project. Beyond that, I just have to let the declarations in the case speak for itself, and just say we deny any wrongdoing and will maintain a vigorous defense, and our declarations in that case just have to speak for themselves.
Okay, I certainly hope that clears up for you, in the near future. I know from other people who found themselves in similar positions, it can drag on a little bit and it's not exactly what I would call one of the things which make life better, having even these little things hanging over you.
We're hoping this case can be resolved. We're all hoping this case can be resolved as quickly as possible.
When I think of developing countries, I see quite a few of them aiming at using WiMAX, for example, in Sri Lanka - Sri Lanka Telecom is deploying WiMAX and connecting it to an IPv6 optical network. It's not some kind of advanced economy, next generation network service-type scenario. They're just saying, "We're connecting people for the first time in their life because we can't roll copper out through the jungle, so there are people who have never had a dial tone. With WiMAX, we're beaming that somewhat into jungle territory and we're giving people their first dial tone".
This is the kind of context I've seen WiMAX being used for in the developing world. Why is it you're some kind of GSM evangelist?
There are a couple of things. First of all, if you deal with a system like WiMAX, I'm not familiar with the system you described in Sri Lanka, but I'm going to assume that the customer equipment for this system in Sri Lanka is probably a rooftop antenna and a router?
It's what they call it. It's quite funny what they call it. They call it a "fixed mobile phone". It's meant to be a fixed phone but it has a radio antenna and technically, it's not meant to leave the home. It's a little bit bulky for that, but technically, you could leave the home with it.
What do those things cost?
I'm not aware of the cost, at all. I could certainly find out and let you know, by the time I see you, in a few weeks at the eComm Conference.
I would be interested. One of the reasons we like GSM is that the developed world produces a steady stream of recyclable handsets that can be purchased at very low cost, but are still perfectly functional. Beyond that, there are tremendous price pressures in the GSM handset market, to drive the prices of customer equipment absolutely as low as technology will allow.
Given all of the factors being equal, the radio for GSM is so much simpler than the radio for WiMAX or Wi-Fi, or CDMA. It will always be cheaper. All other factors being equal, it will always be cheaper. It will always have lower power consumption. It provides the basic service that people really need, which is basic speech telephony, maybe data service, its efficient rates for email. That doesn't sound very exciting, but when you talk again, about the poorest places on Earth, the people who have never had a dial tone, that's a very big deal. GSM can provide that service with probably the least expensive customer equipment that late twentieth, early twenty-first century technology is going to produce, the basic GSM handset.
So, while we are all buying new iPhones and discarding older phones, you're hoping that these can be sort of $5 a piece to the developed world, our old handsets.
Yes, I do. In fact, you say $5 apiece; right now, there are companies in the United States, where you can buy refurbished, repackaged handsets, by the case - some of them for less than $5, depending on the model. Unfortunately, those handsets tend to be North American handsets, which aren't compatible with GSM in the rest of the world, which is a whole different story. There is no reason you couldn't do the same thing in Europe. Again, people are discarding these things in huge numbers and replacing them with more complicated equipment. They're still perfectly functional.
What is your grand vision, then? Is it to provide dial tone to those who have never had?
Yes, over very large, sparsely populated areas, at very low cost.
What's in the way of this grand plan?
The main thing that's in the way of this grand plan is the fact that almost all of the cellular spectrum in the world is already licensed. Companies already hold those licenses. They are reluctant to let anyone else do anything with them. In a lot of cases, they're prevented by regulation. In some countries, you're allowed to sub-lease spectrum and do creative things. In a lot of countries, you're not even allowed to do that.
Carriers have no strong motivation to adopt a radically less-expensive technology. The reason they don't is because if you start deploying $1 a month telephone service, out in a rural area, and you're still paying for your $6 month network, which you installed in the urban area a few years ago, people who live in town are going to want to know why they can't get the cheaper telephone service. You can't afford to do it because you're still paying for that expensive network you already installed.
You can try rebranding; that's expensive. We've talked to a few cellular carriers about this, and it's always the question that comes up. "That's great; you can show me a network that costs less to operate. Now what do I do with the network I just installed? How do I leverage my existing SS7 infrastructure?" That's an exact quote from an executive we talked to.
The answer he wants to hear isn't "You don't; you scrap it and replace it with a packet switched network". They don't want to hear that. These companies have all the licenses. They have all the spectrum. Legally, you can't deploy these networks without their cooperation. The challenge is to present this to them in some kind of way that makes it acceptable to them.
In a few weeks time, I see on the conference schedule that you are marked down as a demo. Is this still going to be a demo you're doing in a few weeks?
That's a good question.
[Laughter] Good question - I have you on schedule as doing a demo, in a couple weeks time.
That's no problem. I can do a demo. In fact, if I have an Internet backhaul and no nasty firewalls getting in my way, we can connect calls to anywhere in the world.
What to do is ping me, off the call. I'll do my best to ensure you've got the connectivity you need because I would love to see a demo in action. A demo, to me, speaks louder than theory.
The thing about the demo of this system is that it's disappointingly mundane. You put a little black box on the table, you pass out a couple of GSM handsets, and you just place telephone calls. Unless you go to pains to do a few tricks to convince people otherwise, it's hard to tell that what they're using aren't just ordinary cell phones, talking to the local T-Mobile network. The demo is disappointing in its ordinariness, unless you really want to look under...
That doesn't sound disappointing to me. Hopefully, it doesn't sound disappointing to others, because that sounds like very good fun to place a call between two GSM handsets without using a real cellular BTS.
That's no problem; it's easy to arrange and I'll have all the equipment there and ready to go.
You mentioned spectrum. Are you aware that the Wireless Innovation Alliance has become a media partner to the conference?
No, I've seen it but I've never really known what it meant.
The Wireless Innovation Alliance is the most respected pushing for change in the spectrum space. People who work in conjunction with the WIA, for example, if you take Maura, who is on schedule, she is Executive Director of the WIA. You have others, like Richard Whitt, from Google; Sascha and Michael, both with the New America Foundation; these are all going to be spectrum-based talks. Actually, I think one of the panels is going to be called something like "Spectrum 2.0," certainly it will be the 2.0 again [laughs]. The panel will certainly be on spectrum. Maybe we will be able to find a way of getting you on that panel, if you're interested. I'll put you in touch with the relevant people.
That would be fine. I will study up on their policies and activities. That sounds interesting. There certainly needs to be a change in the way that spectrum is managed, in a lot of countries. There is a lot of licensed spectrum out there that is sitting fallow. The people who live under it can't afford the services that are being offered on it. It becomes useless.
I'm going to do my best to make the connections there. As one final, closing loop between people, are you aware of the work of Nathan Eagle?
No, I'm sorry; you got me there, too.
Don't be sorry, that's the whole good thing about the "community" - it all sounds like a love fest or something. [Laughs] He also spoke at the debut of 2008. I'm going to need to link you up with Anders Carlius, and I'll link you up with Nathan Eagle. We'll get the transcript of our talk we've being having about your Open Source Cellular BTS and I'll also insert the hyperlinks in there to try to close some loops for everybody.
I really appreciate the time that you've given me in describing that because I think I've learned a lot more from chatting with you, than anything I read on the Internet about the project.
Thank you for the opportunity.
