"By His Genius Distant Lands Converse And Men Sail Unafraid Unto The Deep."
Epitaph on the memorial tablet of the grave of Canadian radio pioneer Reginald Fessenden
The Communications Research Centre Canada (CRC), the federal government's primary laboratory for advanced telecommunications research and development (R&D), has been at the forefront of new developments in mobile digital broadcasting technologies since their inception in the late 1980s. During this time, digital replacement technologies have been standardized in an effort to rationalize spectrum use and enhance broadcasting applications with datacasting services and associated program information. Eureka DAB was the first all-digital mobile broadcasting technology to be conceived, developed and deployed. It became a widely adopted standard for digital radio in many countries around the world in the mid-1990s. DAB was officially launched in Canada in 2000.
This paper presents CRC's attempt to increase collaboration and innovation in the field of mobile broadcasting by developing and offering complete end-to-end Free, Libre and Open Source Software (F/LOSS) toolsets for the transmission and reception of DAB and FM/RDS applications and services.
Background
Radio has always been mobile. AM and FM receivers were introduced in cars in the 1930s, and later, with the emergence of transistor receivers, into portable devices. Radio was one of the first wireless telecommunication applications available to the masses.
Since then, there has been a growing interest in mobile broadcasting services. This has resulted in the proliferation of new broadcaster-led standards as well as standards developed by the mobile industry. These industries recognize that physical layer broadcasting (as opposed to pseudo-broadcasting like multicast on the Internet) is efficient, both spectrally and in terms of infrastructure, for delivery that targets large audiences. This is particularly attractive for media-rich content like radio and TV, which are very expensive to deliver through 2.5 or 3G infrastructures. To mobile network operators (MNOs), this translates into lower delivery costs and relief on their one-to-one networks, which remain dedicated to high-margin personal communication services.
For broadcasters, mobile broadcasting is a natural extension of their traditional mandate and expertise: cultural content, radio, public services like weather forecasts, traffic conditions, emergency information and so on. New opportunities like mobile TV and datacasting are also of great interest.
Open Eureka DAB Transmitter
Broadcast transmission equipment is expensive due to low sales volumes and relatively high technical complexity. This creates a high barrier to entry that limits the potential for innovation in broadcasting.
At CRC, we realized that application innovation for Eureka DAB was difficult when using typical commercial transmitter equipment. The earliest issues appeared with the service multiplexer. In DAB, multiple services are combined in real time to form a single bitstream or multiplex. All services available are announced through a specific sub-channel in the multiplex so that receivers know where to find a service and how to decode it. When we created non-standard applications, there was no way to signal their presence in the multiplex. Application signalling was hard-coded in commercial multiplexers. This was the motivation to develop a more R&D and innovation friendly multiplexer. Over the years, this project evolved to become a full-blown F/LOSS DAB multiplexer, now known as CRC-DABMUX.
Another key component of the DAB transmission chain is the modulator. Again, these types of devices used to be quite expensive. However, with the emergence of software defined radio and accessible platforms, the implementation of a DAB software modulator has become straightforward. The USRP and the GNU Radio open source framework were invaluable F/LOSS tools that permitted the integration of our CRC-DABMOD modulator in a very short time frame.
Together, these two components can now generate DAB-compliant signals at an unprecedented accessibility level. We pushed this one step further by putting most of our DAB transmission tools on an Ubuntu-based live CD which we distribute for free from our projects' Web sites. The CRC mmbTools Live CD can be launched without prior installation on the host PC to produce DAB signals in real time with a simple mouse click and a connected USRP.
We also used our software to provide DAB audio encoding and multiplexing functionality over the Web. These tools replicate typical functionality of physical devices and are known as Web appliances or WAPPs. For example, our WAPPs provide the functionality of DAB audio encoders and multiplexers. CRC-DABMUX is used here to produce multiplexed bitstreams according to parameters and content provided by the users. Since their launch, the WAPPs have been used steadily by various members of the industry and this process turned out to be an excellent mechanism to test our own software. Sometimes, users inform us about issues that they have. Often, we are able to fix the problems and upload updated software components to our WAPPs server in just a few hours.
Openmokast
The context for the democratization of mobile broadcasting at the receive end is different. In a time when independent developers and users are empowered to create mobile apps on iPhone and Android platforms, broadcast apps development remains in the hands of a few players. More importantly though, there are still no clear signals that digital broadcast chipsets will be integrated into mass market mobile devices.
The stagnation of mobile broadcasting technological advances could be explained by the innovative and competitive wireless communications ecosystem that is thriving today. Several kinds of new wireless communications technologies are emerging in the quest to reach mobile users, wherever they are, with maximum throughput. It appears as if mobile broadcasting is standing at a juncture between broadcasters and MNOs who are driven by distinct objectives and business models. Broadcasters are naturally inclined to pursue and extend their current free-to-air services which are monetized by public funding, licensing fees and advertising. MNOs, on the other hand, plan to deploy mobile broadcasting services to generate new revenue streams through cable-like subscriptions and pay-per-view models.
Openmokast was developed at CRC to address the lack of support of mobile broadcasting on emerging open handset platforms and to catalyze an application-driven ecosystem. As with Android, third party developers would be able to create new apps long before compatible receivers actually reach the market.
Openmokast is a complete software stack that provides a high-level application programming interface (API) for the control of a DAB receiver connected to a generic computer or smartphone. It demultiplexes the received bitstream and forwards selected raw sub-channel content to upper-layer decoding and rendering applications. It currently offers decoding libraries for standard applications like DAB and DAB+ audio, DMB video, Slideshow and Visual Radio. It is compatible with the Openmoko FreeRunner and the GNU/Linux operating system.
A physical extension was built to seamlessly integrate a USB-based receiver and its antenna at the back of the FreeRunner. The resulting prototype was the first open programmable handset to integrate the reception of live digital radio, video and data services with typical smartphone functions such as mobile telephony, wireless Internet and GPS positioning. An Android application was developed to showcase the usability of Openmokast in a mobile Wi-Fi broadcast hotspot configuration. This application is offered on the Android Market, as a GPL open source project, and is included on the CRC mmbTools live CD.
Software-defined-radio demodulation of the DAB signal would represent another interesting approach to get mobile broadcasting reception on a mobile device without the need for specialized chipsets. However, today's smartphones still lack the low-cost wideband front-ends plus the processing power to perform heavy lifting tasks like signal demodulation.
Unencumbered Audio Codecs
Many of the tools provided on our CD are offered as F/LOSS to promote innovation and new developments in mobile broadcasting and DAB digital radio technology. However, some key components are missing as the free and unrestricted distribution of standardized DAB audio codecs is not possible. For these codecs, controlled distribution and payment of royalties are required.
Since, audio is important in a broadcast radio system, we decided to integrate a royalty-free (RF) audio option with our tools. Interestingly, new developments in the area of RF or unencumbered codecs were being launched at the same time. For example, the IETF had just created a new working group to standardize an unencumbered Internet wideband audio codec. There is also a growing interest in the integration of the unencumbered Theora codec within the various Internet browsers to provide a free default option for the new HTML 5 video tag. The Open Video Alliance promotes this solution.
We selected the CELT audio codec for our implementation. CELT is a new full-band audio codec being developed and optimized for low-latency Internet applications. Its technical features make it very attractive for broadcast applications. Like other free and unencumbered technologies promoted by the Xiph.Org Foundation, CELT requires no royalties and no special licensing.
We designed and implemented a new transport protocol for CELT over DAB and adapted the C library provided by the Xiph project to our live CD and Openmokast client applications. This library now compiles on the various platforms that we use for our projects: Ubuntu, Android/G1 and Openmoko/FreeRunner. On Ubuntu, our CELT library will run as a plug-in for GStreamer, the F/LOSS media framework. Of course, none of the commercial DAB receivers can decode CELT radio as CELT is not a DAB standard.
Unencumbered codecs have the potential to become widely available on most types of Internet media platforms, which happen to be the same platforms broadcasters are targeting for mobile broadcasting reception. If this happens, broadcast radio systems may have to support such codecs in order to remain competitive with Internet radio on handhelds or other media devices.
Enabling Hybrid Radio
In many regions of the world, where the transition to digital radio appears to be stalled, FM radio could prevail for many years. In this context, how will radio evolve to provide all those multimedia enhancements envisioned for high capacity radio systems like Eureka DAB?
One option would consist of delivering radio over Internet-streaming networks instead of specialized physical layer broadcast infrastructures. While this appears to be a reasonable scenario for broadband-connected locations, live and uninterrupted audio delivery over mobile Internet today simply does not compare to FM.
Consequently, a hybrid broadcast/broadband approach for radio could represent a more realistic evolutionary path toward its all-digital future. Current FM infrastructures would be leveraged to provide reliable and dependable audio-service components while optional multimedia elements would be provided out-of-band through the mobile Internet and the Radio Data System (RDS). RDS is a communications protocol standard for embedding small amounts of digital information, including program information, time and station identification, in conventional FM radio broadcasts.
In the logical continuity of our mmbTools and Openmokast work, we have integrated and developed new approaches and tools to foster the development of innovative hybrid FM/RDS/Internet radio applications and services. One important finding was that most of the required building blocks were already available as F/LOSS projects. A functional implementation was integrated in a few days using the same USRP and GNU Radio platforms that were used for our DAB projects.
While many new smartphones are already equipped with FM and RDS receivers, most of these platforms do not provide official APIs to third party and independent developers to control the broadcast receivers. For example, some new Android-based HTC products have FM and RDS receivers inside them, although no Android API provides access to this functionality and an official Android representative has mentioned that there are no plans to include an API. It seems that handset manufacturers will be the only ones able to provide FM tuning and RDS decoding apps.
We found some unofficial mechanisms to access FM and RDS functionality on some Windows Mobile handsets. Special libraries were developed by hacker communities who reverse-engineered the FM APIs. Thanks to these developments, we produced a portable RDS decoding software library that we will use to demonstrate hybrid radio application prototypes on commercially available smartphones.
Conclusion
The CRC mobile broadcasting F/LOSS projects have generated a lot of interest among the DAB community. Several dozen DAB industry players, including broadcasters, universities, manufacturers and application developers, have downloaded the CRC mmbTools live CD. Our online Wapps are visited steadily. An independent online community of enthusiastic users is actively building a knowledge base around our tools. Our compact end-to-end system triggers interesting discussions about the future of mobile broadcasting whenever we introduce it at international events and trade shows.
While the business models for broadcasting are blurred by new content distribution options and by new media applications, it is difficult to estimate the real impact od the CRC mobile broadcasting F/LOSS projects. We think that by democratizing mobile broadcasting technologies, we increase their chance to remain competitive and succeed in the future.
Recommended Resources
Live demo of CRC mmbTools and Openmokast prototype at eComm 2009
CRC mobile broadcasting team presentations
Video clips about CRC mobile boradcasting projects