The OpenBTS Project
David A. Burgess, Harvind S. Samra
Kestrel Signal Processing, Inc.
Fairfield, California
October 31, 2008
2
Chapter 1
About The Project
1.1 What is the OpenBTS Project?
The OpenBTS Project is an effort to construct an open-source Unix application that uses the
Universal Software Radio Peripheral (USRP) to present a GSM air interface (“Um”) to standard
GSM handset and uses the Asterisk VoIP PBX to connect calls. This is in fact very different from
a conventional GSM BTS, which is a dumb device that is managed externally by a basestation
controller (BSC) and connects calls in a remote mobile switching center (MSC). Because of this
important architectural difference, the end product of this project is better referred to as an access
point, even though the project is called “OpenBTS”.
1.2 Why Build an Open Source GSM Stack?
The combination of the ubiquitous GSM air interface with VoIP backhaul could form the basis
of a new type of cellular network that could be deployed and operated at substantially lower cost
than existing technologies. Since these new hybrid networks are not readily compatible with legacy
networks, and since radical two-tier pricing would be disruptive for existing carriers, we are not
likely to see this kind of innovation from the conventional telecom community. This is the primary
motivation for starting this project: a vision of truly universal telephone service.
The inspiration for this project came from a simultaneous recognition of these elements from prior
experience
1
with GSM, software radios, VoIP and sustainable power systems:
• The USRP can be readily adapted as a GSM transceiver and the hardware can be reworked
to give a carrier-grade radio for use in a software BTS.
1
The founders of this project actually come from a signals intelligence background. They’ ve b een working with
wideband digital radios and looking at telecoms networks as “black boxes” for a total of 25 years, giving them a
different mental model of GSM than you might find within the cellular industry. For the most part, they’ve only
cared about the air interface, and with VoIP that’s all we really need anymore.
3
4 CHAPTER 1. ABOUT THE PROJECT
• We know from prior experience that a GSM protocol stack for basic telephony can probably
be implemented in less than 15,000 lines of C++, including the software radio. A useful
basestation can probably run in real time on a 1.5 GHz 32-bit CPU. L1 is textbook radio and
FEC algorithms. L2 is simplified HDLC. L3 is fully defined in GSM 04.08 and ITU-T Q.931.
It’s actually not that complicated, once you understand the specifications.
• Most telephone switching functions, calling features and even mobility management functions
can be moved into a software PBX like Asterisk, eliminating the need for most of the network
infrastructure (HLRs, MSCs, etc.).
• Plenty of functional GSM handsets are discarded in rich countries every year that can be
reused in developing countries.
• With economical software design and good hardware engineering, it should be possible to run
a low-capacity GSM cell from solar panels or micro wind turbines, drastically reducing the
cost of service in off- grid environments.
2
Early in this project s ome of us had a conversation with an executive from a well-known (but
NDA-ed and anonymous) cellular carrier. He asked, “Lowering costs sounds good, but how does
this leverage my existing network?” Our answer was, “It doesn’t. That’s not the point. This is
leapfrogging.” He then asked, “Has Ericsson put a horse’s head in your bed yet?”
1.2.1 GSM is Old and Boring. Why Not CDMA?
GSM is a good choice precisely because it is old and boring. Everyone knows it works and 80%
of the world’s carriers are still using it. It’s a proven technology that is well-suited to the target
application. The specification is publicly available and in a few more years most of the essential
patents will expire.
CDMA physical layers are too complex for an inexpensive all-software radio and do not scale well
for low-capacity cells. CMDA capacity comes in increments of 50 or more subscriber lines and
the lowest layers of your radio must process all of that bandwidth whether you intend to use it or
not. By contrast, GSM capacity comes in increments of 7-8 lines and a well-managed radio can
even ignore inactive parts of the signal. Beyond the technical issues, IS-95-s tyle CMDA (including
cdma2000) is tightly controlled intellectual property. You can’t even get a copy of the specification
without signing an NDA and paying several hundred dollars.
1.2.2 What About GPRS/EDGE and UMTS?
Future versions of the OpenBTS may well support GPRS and EDGE. GPRS, when available,
should be a software-only upgrade for any installed OpenBTS system. EDGE support may require
additional computational resources but the additional software is not complex, at least when com-
pared to the rest of the BTS. UMTS is a radically different CDMA-style physical layer and well
outside the current scope of this project.
2
One of the founders of the project runs most of his house on solar power.
1.3. LEGALITIES 5
All that said, let’s walk before we run by implementing basic voice services on a single-ARFCN
GMSK BTS.
1.2.3 What’s Wrong with WiFi, WiMax, WiWhatever?
There are a lot of people out there who would rather not blanket Africa with WiWhatever than
find a way to make GSM dirt cheap.
3
It’s sexier to talk about the newest air interface and that
talk gets a lot a buzz, but the truth is that that WiWhatever is po orly suited to mobile telephony.
WiFi range is far too short for mobile coverage in rural areas. For example, you’re not going to
cover 700 square miles with a single WiFi tower, but that’s exactly what GSM was made to do.
If access points are connected through different ISPs, handovers will be unreliable. The phones
are expensive and power-hungry, and compared to GSM they always will be. WiFi may become
a decent technology for semi-mobile telephones in dense urban areas, but it’s not a mobile phone
standard and it’s not well-matched to the rural cellular application.
WiMax has most of the problems of WiFi. To make matters worse, most WiMax bands don’t
penetrate structures very well. Most WiMax deployers are planning to solve this problem by
saturating large buildings with small access points. T hat’s fine in Manhattan and London, but
we don’t see anyone putting femtocells in a million houses when those households couldn’t afford
phones in the first place.
More important than all of that is to remember the goal of the OpenBTS: universal telephone
service. Our project philosophy is that it is much better to give people basic telephone se rvice
with an upgrade path to 250 kb/sec EDGE than to generate a lot of hype over a scorching fast
broadband technology that can probably never be truly universal. Don’t let the perfect be the
enemy of the good.
WiWhatevers do have a place in OpenBTS, though: backhaul. The standard OpenBTS backhaul
is likely to be redundant-path point-to-point WiFi or WiMax.
1.3 Legalities
Every project has the potential for abuse. Let’s avoid that.
1.3.1 Intellectual Property
We can bet this project will get a lot of intellectual property scrutiny from the vindictive employers
and ex-employers of every experienced telecom engineer who contributes. To limit their use of the
legal system as an instrument of harassment and abuse, and to prevent genuine bad acts, pleas e
observe these guidelines:
3
We say “not” because none of those people who talk ab out it are really going to do it because it’s not a realistic
goal.
评论0