Open source hardware for telephony [LWN.net]

October 21, 2009

This article was contributed by Robert R Boerner Jr

Applying open source principles to hardware, specifically hardware for telephony, can lead to lower-cost telephone service, which may well be a boon to those in developing countries. Several projects are working on devices and software that can dramatically reduce the cost of providing phone service, particularly in rural areas or those with less infrastructure to support it. In addition, those projects can also potentially bring service to places where telephone service is currently unavailable.

The precepts of open source hardware (OSH) are very similar to those of open source software (OSS). The ideas that make up an object, whether they be schematics for a circuit board or CAD files for a clock enclosure, are available to view, copy and modify. As such, many OSH projects have adopted some of same licenses traditionally used by OSS projects, such as the GPL and MIT licenses. Some have even adopted Open Hardware specific licenses, such as the TAPR Open Hardware License.

One person, in particular, has created an OSH project that could change the face of telephony. David Rowe, an engineer hailing from Adelaide, South Australia wants to make the ability to make a phone call a right, and not a privilege, for every person in the world. And he has designed the hardware (running Linux, of course) to do just that.

The Free Telephony Project was started by Mr. Rowe in 2005. Like many OSS developers he had an itch to scratch, and the process of scratching that itch led him to develop the IP04 PBX: "a low cost phone system that can switch phone calls from analog phones or phone lines over the Internet using VoIP".

The IP04 is an embedded device powered by Analog Devices Blackfin processor and running Linux as the operating system with Asterisk software serving as a Private Branch Exchange (PBX). The IP04 is designed to bring the cost of deploying telephone systems down to the point where most anyone could deploy them in developing nations. The concept of such a device is not new, in fact Mr. Rowe had actually started and exited a company that provided hardware for telephony on Linux. What makes the IP04 different is the relatively low cost (starting at approximately $300 USD), low power requirements (the unit can be powered by solar power and/or batteries if need be), and the fact that the designs of the all the hardware and software are open source.

Mr. Rowe envisions possible deployments of the IP04 as seeds of entrepreneurship in developing areas. A budding local businessperson could set up a device and provide services to people in his or her local area; in essence becoming a small telephone company. Mr. Rowe believes that with the right help for the initial deployment, the IP04 presents an opportunity to find the right franchise model that would allow for "self-funded, viral growth of telephony in developing communities. Business is a much more powerful way to roll out a service than continual donations and first world support."

Mr. Rowe blogged his progress during the IP04 design phase, and his work caught the eye of Atcom, a Chinese manufacturer of telephony equipment. Atcom contacted Mr. Rowe to thank him for the open designs he had published and offered assistance if he ever needed anything to be manufactured. When the IP04 reached prototype stage, he took Atcom up on its offer. "Three weeks later DHL rang my doorbell and there were two assembled prototypes on my doorstep." Final production hardware started rolling off the line in July, 2007. It only took about 18 months to go from idea to a finished product.

The help from Atcom is but one example of how making the project open has helped the IP04's progress. Mr Rowe is quick to acknowledge everyone that has helped along the way, "I stand on the shoulders of giants. Thanks to all the people who have contributed and whose work I have built on. In no particular order: Atcom, Analog Devices Blackfin team, the Asterisk community, and the Astfin & BlackfinOne teams."

The IP04 has spawned other devices, such as the IP01, IP02 and IP08, differentiated mostly by the number of possible connections to either analog phones or analog lines in the absence of VOIP service. Atcom produces units for sale, and Mr. Rowe also sells the devices via his website (in addition to bare printed circuit boards for those adventurous enough to assemble a unit by hand). The IP04 has even been put through a gamut of certification testing, winning FCC certification in the U.S. and A-tick certification in Australia. There is an active forum where users can go for support, and many have helped extend the device either through software add-ons or by helping chase down bugs. One company has even started a successful business selling and supporting customers with a range of devices through the addition of custom firmware, as well as starting its own community forum.

The possibilities of low cost and open communication for the world are many. The IP0x series of devices seem to be but the first building block. This fact was recognized by Steve Song. As the Shuttleworth Foundation's Telecommunications Fellow, Mr. Song was involved with the creation of the Village Telco Project which had many of the same ideas that Mr. Rowe envisioned in his Free Telephony Project. Mr. Song invited Mr. Rowe and several other like-minded individuals to a workshop to come up with ideas for extending the concept of a low-cost telephone company toolkit.

Out of this first meeting, known as the First Village Telco Workshop and held in June 2008, the idea of the next generation of OSH telephony device came to life: The Mesh Potato. Essentially a WiFi router with a FXS (Foreign eXchange station, a connection for a traditional analog phone) port running a mesh network, the original concept was devised by colleague of Mr. Rowe's, Kristen Peterson during a conference in 2007.

The concept of the device is simple. A small unit the size of a standard WiFi router (running OpenWRT) that would cost approximately $50 USD and has a connection for a traditional cheap analog phone (the FXS port). This device would operate in a mesh network with other, similar devices, much in the same manner as the the OLPC's XO laptop, in essence creating an ad-hoc telephone network with no other infrastructure required. The devices can operate standalone, or could be connected upstream to a local provider. Mr. Rowe offers, "Many people in the developing world already spend a large proportions of their income on cell phones (up to 40%). They are getting ripped off by the sort of business models that cell phones seem to attract. We aim to introduce a little competition using service running on unlicensed spectrum."

Use in developing nations is not the only potential use for the Mesh Potato device. Mr. Song has envisioned uses in a crisis situation. After a major disaster occurs, if all cell communications and landline communications are shut down, a number of Mesh Potato devices could be deployed in a very short time. Though still a concept at this stage, Mr. Song has laid out a interesting scenario in one of the Village Telco blog posts.

The Mesh Potato has hit prototype status and the first devices are being readied to hand out to testers all over the world. When asked to define what would mark success for the Village Telco Project, Mr. Rowe answers, "Six months in operation in some township while making a profit for the Village Telco Entrepreneur. Making $ is the best way to prove the technology is working."

The OSH telephony innovations don't stop with Mr. Rowe or the Village Telco Project. The Astfin project, a uClinux Asterisk distribution, not only supports the IP0x series of devices, but they have also produced hardware that offer different connection options such as ISDN's basic and primary rate interfaces (BRI/PRI).

Of course, the community has pushed the envelope even farther. The OpenBTS project is a "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 software PBX to connect calls. 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 in greenfields in the developing world." In essence, they have taken an OSH device (the USRP) combined it with some OSS, and provided a means to create a wireless network compatible with many mobile phones throughout the world. They recently put the system to the test at the Burning Man Festival and have detailed blog postings about what worked and what did not.

Whether the Open Telephony Project, Village Telco Project, or the OpenBTS Project are successful remains to be seen, but in all three instances, the decision to make the not only the software but hardware open has already paid dividends in terms of time to market and fostering of the community. No one project would have a chance to succeed if other open software or open hardware projects did not already exist. Just as the world has benefited from OSS, the future of OSH seems to hold many possibilities for the future.

(Log in to post comments)

Open source hardware for telephony

Posted Oct 22, 2009 16:55 UTC (Thu) by iabervon (subscriber, #722) [Link]

One of the things that makes it exciting to get hardware mass-produced is that, once you've designed and built your prototype and you send the design off to a manufacturer, you find out that they won't make what you specified. They'll tell you that certain components have cheaper equivalents, some things are too close together to be easy to assemble, some components aren't available in the quantity you'd need (or are going to go out of production). It's like if you built some software, and then tar decided you have too many zeros in your executable. If you've got a good manufacturer, they'll be right about exactly what's wrong with your design (for production right then, because it will change) and what they can do to fix it. If you don't have good communication with the manufacturer, they'll just make the changes.

Of course, if you're doing anything more clever than they understand, your devices won't actually work. In general, you've specified some things more exactly than necessary (e.g., you have to specify a part number for your LED, because half of the LED brands need to be installed in one orientation and half in the other, so it won't work to say "random LED, oriented like this"; but it does work to switch brands and install it backwards); but other things need to actually be exactly like you say.

And, traditionally, you're trying to keep other organizations from being able to figure out your design and release something similar, so you've obfuscated your design; and you don't trust the manufacturer and you're half pretending that they aren't reverse-engineering everything you've done so they can redo it for you, so you're not going to give them all the important information.

Now, if you've got an OSH design, you obviously aren't trying to make it hard to reverse-engineer, and you've got all of the information in the design. You've probably got a page of schematic for each major part with the components that are for it, and the manufacturer can just look up whether they can swap values and what tolerances actually matter. If you have the relationship in place early, you could probably even avoid doing the work they're just going to redo anyway. You usually think of open source as allowing other individuals to modify your code and run it, and hardware isn't so much like that because there's a substantial cost to producing a working device from a design (if I change the design of my cell phone, I have to spend a ton of money getting a new cell phone built to my new design). But there is the aspect that the people producing your physical objects will do a better job if the source is available, in a way that doesn't apply to people pressing CDs of software.