A new book just published provides a simple introduction on how to build your own GSM small cell, using freely available open source hardware and software. Aimed primarily for use in the lab or in remote/unserved rural communities, a simple GSM site can deliver voice, text and data services using standard handsets. We review the current state of open source projects and look at what's available today.
What exactly is Opensource?
The Opensource concept has been highly successful in many areas of software. This website, as do the majority of the web, runs on MySQL and Linux – both developed by volunteers from around the world. Source code is published and can be used by anyone, on the basis that any improvements made are also shared with the community. Most of the successful projects have a commercial business co-ordinator that is funded by providing support and/or more robust/complete for those organisations that want to pay for it. Well supported crowd-sourced developments can achieve high levels of functionality, security and maturity because they've been stretched, scrutinised and tested in many different ways. Smaller projects that haven't attracted critical mass can fall by the wayside leaving poor quality or incomplete designs.
Opensource also applies to other fields, and includes hardware, media (photos, videos etc.). Popular opensource licence agreements, such as GNU and Creative Commons encourage sharing by making it clear what the author intends.
This doesn't avoid the issue of patents and Intellectual Property Rights – there are many involved in all aspects of mobile networks, embedded in the standards. Many of the original GSM patents have expired since the system was originally developed more than 20 years ago. Others still apply.
Opensource for Mobile
There are several Opensource projects working towards a complete mobile network, including both the hardware and software, compatible with today's standard mobile phones.
OpenBTS is arguably the most well known, with an end-to-end solution for GSM. A more basic capability for 3G UMTS was added in October 2014. It builds on Asterisk, an opensource voice switch used in many PBX and Internet VoIP services, extending it with the GSM protocols. It's managed by Range Networks who own the trademark and strongly supported by others including Fairwaves.
Osmocom (Open Source Mobile Communcations) is arguably a more stable in production. It takes a very modular approach and includes both GSM and other radio technologies, such as DECT and TETRA. It also builds on Asterisk to offer the full scope of voice services and has GPRS data capabilities (with some limitations). Perhaps more fully open source and engineering led with less marketing than OpenBTS, it's overseen by Sysmocom of Germany and also supported with product by Fairwaves.
YateBTS was recently launched by one of the founders of OpenBTS. It has a long term vision to create a unified core network using VoLTE for calls for both 2G and 4G, and substantially reducing bandwidth for voice over satellite links compared to traditional SIP. The project is co-ordinated by Romanian company Legba.
OpenLTE is relatively new project to implement the core 3GPP LTE specifications. Today, code is available for test and simulation of downlink transmit and receive functionality and uplink PRACH transmit and receive functionality. This is very much research lab oriented and nowhere near ready for field use. Three other LTE opensource projects are also at early stages as described here.
These projects all benefit from and build on other Opensource projects, such as OpenSS7, Asterisk, GNURadio etc.
Although not nearly as extensive as a standard commercial product, these can be feasible for basic use with isolated service. Mobility, handover and roaming capabilities are included as are voice, SMS and data services. Looking a bit deeper, each cell/sector is configured as a completely separate Location Area so a full location area update is used to handover between cells. GPRS supports only two of the four coding schemes; manual configuration is required of many parameters, such as neighbour lists, timeslot allocation, RF power levels. In my view, this would hamper anything other than a small scale deployment.
The system can be connected to wholesale VoIP, SMS and Internet connections to provide inbound and outbound voice and data. One complication is that because different suppliers are typically used to provide wholesale voice and text services, each would require a different MSISDN (phone number) – definitely confusing for end users.
GPRS data does work but isn't as fast or mature as a commercial EDGE implementation. One company doesn't recommend using it at all, reasoning that Wi-Fi is cheaper/better in such low price markets for data only. However VoIP over Wi-Fi is considered far less attractive than GSM for voice.
The system can use existing SIM cards from an existing network, assigning a local number and automatically registering them for use. The full GSM security can't be used in this case (because the encryption key is hidden inside the SIM card), but a simpler form of encryption is offered.
Today, it's possible to program your own SIM cards manually. For larger quantities, a full production run can be bought with your own logo design using your own specified parameters.
Off the shelf hardware
Several vendors offer all you need to run a basic GSM service, including the core network, with off-the-shelf hardware for use in the lab or outdoors.
Just don't expect a fully automated SON solution that sits comfortably with any existing network on the same frequencies – you'd still need a commercially mature small cell solution for that. You'll also need some spectrum to use this legally, either a test licence for your lab or a fully blown one from the regulator. In a few countries, low power GSM is legally permitted in certain guard bands (eg at 1800MHz) without a licence.
Example products include:
- Range Networks starter kit based on OpenBTS from $3,300
- Fairwaves offer solutions based on Osmocom and OpenBTS software with their development board for just $850 and a packaged lab system for $2,500. Their equipment range is expanding to include outdoor 2x3W and 2x10W products.
- Sysmocom in Germany use Oscocom; products include a 200mW small cell and larger 10W outdoor product.
- Legba, the new company behind YateBTS, currently offer a GSM lab radio kit for about $2000 and an outdoor model for $12,000. Licences for their HLR/HSS and other modules run from $12,000.
Case study deployments
An example installation described here is of a remote Mexican village of 700 inhabitants 5 hour away from the nearest city. It has a simple network with two GSM transceivers that handle around 1000 voice calls and 4000 texts in a typical day. The antenna mast is constructed from a 6 metre bamboo pole. Another Mexican village of 500, San Juan Yaee, was connected for just $8,000 – about 15% of the cost quoted by the national operator – with ongoing monthly rates of $2. At those prices, nobody's going to get rich.
Software definable radio hardware can be used for a wide variety of different applications, ranging from detecting/decoding shipping and aircraft location beacons. This article outlines 10 different possibilities.
One application which I though remarkably innovative was used to locate stranded hillwalkers from a helicopter. The GSM basestation onboard takes several measurements of the walker's phone signal from different positions and triangulates to find where they are. Using simple GSM time-advance measurements, the results are displayed on an iPad inside the helicopter. It's not dependent on mobile data or the victim being conscious, as is needed for a similar app called SARLOC.
I'm enthusiastic about the use of Opensource projects to stimulate research and development into new pioneering new ways and means of improving and extending mobile technology. It should enable our academic institutions to demonstrate and prove their theories with limited budgets.
There may also be an opportunity to connect some of the most remote and unserved communities which commercial organisations haven't been able to reach. The scale of this would be limited by spectrum licences and IPR. The recent proposal by Mexican regulators to allocate some 850MHz spectrum for community use by unserved areas Mexico sends a signal to commercial operators that they can't simply ignore this demand.
In most cases, I believe it would be better to use commercially mature, mass market solutions managed by professional organisations. Only where those needs are not being served, and regulators support and encourage it, would we see this self-driven community driven approach adopted more widely. The lack of scalability and management features of these solutions limits their scope to very small and simple deployments. Commercial ventures could either develop their own products using proven software from companies such as Radisys or NodeH, or adapt and extend many of the existing proven small cell products already on the market (look in our vendor section for plenty of ideas!)
Fancy building your own small cell?
This recently published book (Getting Started with OpenBTS) provides a good explanation of how to go about setting up and configuring a basic GSM cell using OpenBTS, standard hardware and a Linux server. A complete commercial product for the lab costs around $2,500 but there are lower cost alternatives. Normally priced at around $15 for Amazon Kindle, you can download the PDF free here (in exchange for your personal details).