FMeXtra: Another On-Channel Solution
Derek Kumar of Digital Radio Express Explains the Development of a Digital System Using Subcarriers

By Michael LeClair

Down load a PDF copy

At the NAB Broadcast Engineering Conference in April, a new digital radio system debuted on the show floor from Digital Radio Express.
Offering a simple implementation and relatively low costs, DRE demonstrated working digital transmissions and receivers.

In this issue's interview we speak with Derek Kumar, vice president of engineering for Digital Radio Express. Kumar received his BSEE and MSEE from the University of Illinois at Urbana-Champaign in the 1980s. He has been active for many years in the development of digital radio systems.

What drew you to broadcasting?
While I was a National Science Foundation graduate fellow at the University of Illinois, I met Professor Bill Hunsinger. He had invented a new microwave signal processing technology called acoustic charge transport [ACT]. Professor Hunsinger was already well known in the field for his contributions to surface-acoustic wave [SAW) filter theory
SAW filters are used extensively in satellite and mobile phone systems. ACT technology allowed for a digitally programmable SAW fitter. It had all sorts of applications that were previously thought to be too complex or too power-hungry to implement by conventional circuits.

The professor spun the research group out of the university and founded the company Electronic Decisions Inc. We were funded by the then-largest single grant from the Defense Advanced Research Projects Agency
It was an incredibly exciting time. We had a generous equipment budget, and we were researching both fundamental technology issues and high-level system applications, which is a rare opportunity today

How long have you been working on the digital radio system that is now being offered as FMeXtra?
In the late '80s, a group of broadcasters approached EDI to see if there was any way to use ACT technology to make an FM IBOC system. That was my first encounter with Tony Masiello of CBS, Paul Donahue of Gannett and Glynn Walden of Westinghouse.

They had formed a consortium, eventually named USA Digital Radio, to come up with an alternative to Eureka-147. The objective was to come up with a system that preserved the existing transmitter infrastructure, relative signal coverage and station branding.

EDI built the first workable FM IBOC system. At that time, the receiver was a large rack of equipment, and it took a gasoline-fired generator to keep the whole thing running.

I have many road warrior stories, most of which involve sitting in the back of rental vans with Tony, rambling through downtown Chicago expressways at midnight with a homemade aerial tied to the front bumper, and the generator mounted and running on the lift gate. When we'd stop for gas, the locals thought we were federal agents.

Bottom line, the system worked, and it didn't rely on blend-to-analog. Unfortunately, the audio codec technology, MPEG Layer 2, was inefficient compared to what's available today, so we were trying to achieve data rates well over 200 kbps in the IBOC "shoulder" sidebands. You can't do this without going into the adjacent channels or causing compatibility problems with the analog host.

There is no significant difference in spectrum occupancy between the "extended hybrid" mode of IBOC today and these earlier systems, which were deemed by the NRSC and others to be incompatible with the host analog FM signal.

After the initial industry rejection, USA Digital Radio basically went into hiatus for several years until the audio compression technology improved. I had left EDI and gone to National Semiconductor for a short time, but I preferred to be an independent consultant.

Soon I recognized that new, more efficient audio codecs like AAC could be combined with redundant IBOC sidebands, and the combination could overcome the objections to the earlier IBOC systems. So I filed a series of patents and teamed up with my former manager at National Semiconductor, Norm Miller, and Mr. Dwight Taylor, to form Digital Radio Express. We founded DRE in the heart of California's Silicon Valley to have access to the latest technologies and the region's tremendous engineering talent.

Eventually DRE asked the NRSC to reactivate the DAB subcommittee. Early on, we saw that IBOC was going nowhere as long as there were multiple proponents, and even in the best estimates, it would be many years before there would be any return on investment.

So we decided to license our patent portfolio for use in IBOC to USA Digital Ridio, which eventually merged with Lucent's, IBOC group to form lbiquity We are an lbiquity shareholder.

We decided to turn our attention to the FM SCA because it didn't have the long political process of IBOC, so we felt it had a much better near-term return on investment. The FM SCA also is applicable worldwide, and it appeared that IBOC would have uncertain international acceptance in view of competing open-standard digital radio systems.

We decided to call our new digital SCA system FMeXtra to emphasize its connection to the existing analog FM signal - the "FM"- and to emphasize the new content our technology made possible - the "extra."

How do you see FMeXtra in relation to Me HD Radio system from lbiquity?
We view the only NRSC-tested and FCC-approved form of IBOC, which is "hybrid" mode IBOC, as a complementary service. Bits are bits. More bits are better. However, our business plans diverge when it comes to utilizing the bandwidth closer to the main channel signal, and broadcasters will ultimately make that choice.

Our vision of the future is the transition from analog stereo to analog mono main channel with a corresponding increase in ir-jection and bandwidth for the digital subcarrier. There is a certain naivete that digital signals need very low power levels for proper operation, but in other digital radio systems such as Eureka-147 and Digital Radio Mondiale, significant increases in the recommended power levels over the original designs were required in order to achieve reasonable coverage.

We want to transform the existing highpower analog FM transmitters into highpower digital FM transmitters. There is nothing inherently digital or analog about a transmitter. They are simply either classC or linear transmitters. Digital modulation methods can be made to work efficiently with class-C amplifiers. GSM mobile phone technology, which represents more mobile phones than any other competing technology, is an excellent example of a successful class-C digital transmission technology.

Are there ways in which your digital system is superior to that offered by lbiquity?
Performance-wise, I think all in-band systems that inherently have to coexist with existing signals in a relatively narrow bandwidth, including both IBOC and FM SCA, have certain limitations when compared to clean-slate designs.Talk about the basic technology concepts underlying FMeXtra.

The FMeXtra system implements bestin-class across many technologies. We use a multicarrier modulation like COFDM to achieve tight spectral shaping and minimize intersymbol interference. Each of the carriers in the multicarrier composite is modulated using hierarchical modulation [HM] to provide a variable information throughput.
Our error correction is based on "turbo" iterative decoding, the most powerful form of error correction known. We incorporate an adaptive RF channel equalizer based on waveform properties to mitigate multipath, and IF digital filtering for adjacent-channel rejection.

The channel equalizer is a crucial part and is one of the big advantages over previous SCA receiver implementations. It is similar to the adaptive equalizers used by Motorola - now Freescale - in its Symphony digital radio chipset and the Blaupunkt Twinceiver.

A side benefit of all of this processing is that it cleans up the received analog FM signal as well. We use the latest audio compression technology, aacplus version 2, which I'll discuss later. When MPEG surround sound implementations are available, we'll offer that as well.What is the payload data rate for FMeXtra?
The payload depends upon the FMeXtra encoder configuration. Our system is very flexible. Almost everything is controlled by software switches, and we let broadcasters make their own choices about important parameters like delay and data rate.

There also are three different signal constellations, from QPSK to 16-QAM,and four different code rates, so there many combinations. Minimum payload is around 20 kbps for a partial subcarrier, up to 64 kbps for a full subcarrier with stereo main channel, and well over 128 kbps when the main channel is mono.
Everything is implemented in a single IU rackmount server with custom PCI interface cards. Connect your audio sources to the PC and connect the PC to your existing FM exciter. Plug and play You can be on the air with FMeXtra in only a few minutes.

FMeXtra system can be used for a variety of multicasting applications. We provide good stereo audio quality with 15 kHz bandwidth and 90+ dB dynamic range at bit rates as low as 20 kbps. Our system supports arbitrary sample frequencies and does not have strong synchronization requirements, so we have a lot of flexibility for incorporating alternate audio and even video codecs specifically optimized for low bit rates, unlike other digital radio systems.

We also support closed or subscription services with built-in conditional access. Foreign language clients, reading services for the visually impaired and other private radio networks are natural candidates, and we've seen tremendous interest from them. Some public radio stations have expressed an interest in having memberonly FMeXtra channels, immune from pledge breaks.

The ability to partition the bit capacity into arbitrarily small slices makes it possible to service a lot of niche interests that simply weren't economically feasible before, and we are just now starting to see the entrepreneurs emerge. But we also have a lot of interest from existing broadcasters who simply want to multicast in the most economical manner.What about the legal aspects of using your system, with regard to how it is regulated by the FCC.

Are there any STAs or special permissions required?
FM SCAS, both analog and digital, are not regulated by the FCC. Provided the SCA system complies with the bandwidth and injection limits, any form of modula tion is permissible without any additional authorization. No paperwork, no hassles.

However, we certainly see a future path to using more bandwidth and more injection than the current rules allow, and when the time is right, together with broadcasters, we will approach the regulatory authorities. Similarly, we don't charge broadcasters license fees when they use our technology. You simply buy our box as you would any other piece of broadcast equipment.

Can you discuss the results you have from field-testing FMeXtra?
Coverage is a complex topic and we don't have all the answers yet. The number one issue is injection, which is limited by antiquated FM SCA rules.

The biggest issue is peak-to-average power ratio. Since we use a COFDM-type signal, this ratio is significantly higher than a single-carrier system, and the current rules penalize us even though there is no interference. We believe we can get coverage at the 60 dBu contour within the existing rules and our tests to date support that.
We've had the system on the air in the San Francisco Bay area, Los Angeles area, San Diego, New York, Minnesota, multiple states in Brazil and, by the time this interview is published, Israel. We can do much better than that with higher injection, what we call a "superinjection" subcarrier, which causes only a minor increase in the occupied bandwidth, much less than the bandwidth occupied by IBOC signals.

By superinjection, we simply mean SCA injection levels well beyond current FCC SCA limits. We have tested our SCA in the laboratory environment with the ir~jection set at over 50 percent peak modulation, and the increase in interference is very small, much less than the interference measured by the EIA and NRSC in the first- generation IBOC configurations. The next step will be experimental testing with an STA.

What type of audio coding is used in FMeXtra?
We use Coding Technologies aacplus version 2, which is an open MPEG4 standard, selected by 3GPP [Third Generation cellular Partnership Project], XM Radio, Digital Radio Mondiale and many others. We believe it is the best audio codec at these data rates.

Greg Ogonowski of Orban has posted a large variety of audio samples on Orban's Web site in the Opticodec product section, and I would encourage your readers to have a listen. We are currently working on a data link between our FMeXtra encoder and the Orban Opticodec. When that is finished, you'll be able to ship compressed aacplus streams directly into our system over TCP/IP from anywhere.

This is powerful because it eliminates transcoding artifacts, dramatically reduces STL costs on the transmitter end and it opens the door to all sorts of interesting DVR-like devices on the receiver end. Millions of cell phones today already have an aacplus decoder in them and that number is going to increase dramatically in the coming years. Free PC-based aacplus players like Winamp are readily available on the Internet. We don't believe there is room for another audio codec at this point.

Are receivers available that will work with FMeXtra?
Regarding receivers, right now we are concentrating on specialty receiver markets as a replacement for analog SCA receivers and other niche markets. It's a long road to get any significant consumer acceptance for general-purpose digital radios~ just look at the IBOC radio sales.

Rikei Corp. of Japan is overseeing the manufacture of several different model types - tabletop, car and boombox for low-cost FMeXtra radios. The simplest radio is less than $100. We are working with IBOC radio manufacturers too. For implementations that use the TI IBOC chipset, which is just a re-badged general purpose DSP, there is no additional hardware cost to add FMeXtra, and we expect to see multiple models of FMeXtra/IBOC combination radios.

However, we believe few IBOC radios will be sold at the current IBOC receiver price point.
Some analysts have criticized the DSP implementations of IBOC receivers because of their excessive power consumption, but we've implemented FMeXtra on these DSPS using only about lOOmA, which makes battery operation feasible.

In closing, how do you see the future of FM broadcasting over the next couple of decades? Is the conversion to digital inevitable?
We strongly believe that digital conversion is essential to long-term survival of the radio industry. We see competitive pressure coining from alternatives such as SDARS, 3GPP, WiNlax, you name it, the list goes on and on. It has nothing to do with CD-quality audio, though; it's all about the content choice. We believe our system accomplishes the goal of digital conversion to increase content choice without making hundreds of millions of radios obsolete.
FM radio is one of the few true worldwide communication standards, and we'd like to see it stay that way. Our roadmap from the existing stereo subcarrier today to a wideband digital "superinjection" subcarrier, while preserving analog mono indefinitely, with or without IBOC, is seamless.