One common desire of all broadcasters is to extend the station's coverage. Within the framework of our licensing system, that isn't easy to do; however, having your coverage optimized within your allotted coverage area is a realistic goal. The idea of the Single Frequency Network (SFN) is nothing new, and has been implemented in both the AM and FM bands. Interestingly, iBiquity's HD Radio system, and the open source DRM system both lend themselves readily to SFN implementations.
Research has been done in the implementation of SFNs with HD Radio. I'm going to talk about three different examples. I have used two papers in research for this article, and I encourage you to read them for even greater details. The first was prepared by Anders Mattsson of Harris and John Kean of NPR Labs. The second was published by NAB Fastroad.
My own experience with SFNs goes back to an earlier part of my career working in the San Francisco Bay Area. The station group I worked for had three boosters operating on the combined system on Mt. Diablo, which was designed to serve the east bay portion of the greater Bay Area. (This booster system was developed in 1989 by several well-known chief engineers plus the consulting group Hammett and Edison. More details can be seen by studying FCC facility IDs 86911, 28878, or 59990.) This system provided frequency and modulation synchronization for the boosters, and thus minimized the effects of multipath in locations where the desired to undesired ratio was fairly low (between 10 and 20dB). Later, as the chief engineer of KIOI, I developed a booster for the east bay that relied on off-air reception, amplification and re-broadcast (see, for example, FCC facility ID 4085).
What I learned through the experience of building boosters (we didn't refer to them as SFNs) and listening to them for countless hours driving around is that, despite your best efforts in synchronizing the frequencies, and modulation, that terrain shielding was actually the key to their success. I would even go as far as saying (and this is purely anecdotal) that 90 percent of a booster's technical success is due to the terrain shielding factors. The geography of the Bay Area lends itself to the success of VHF booster systems.
It turns out that HD Radio (and DRM) can both be used successfully in the construction of SFNs, also by taking advantage of terrain shielding, frequency and modulation synchronization. According to Anders Mattsson (in the first article I mentioned) digital radio in general has a distinct advantage, not in the transmission mode, but in the design of the receivers: "This poor performance (of analog SFNs) is not due to any inherent limitation in SFNs. In fact, it is because of the lack of equalizers in traditional FM receivers. At the moment the receiver can handle the multipath, SFNs offer many potential advantages. Since all digital systems such as HD Radio and DRM already have equalization, the old limitations are gone."
Even so (as one would expect) there are practical limits to what can be done with HD Radio SFNs, and they mainly relate to what is known as the guard interval in OFDM. Recall that the IBOC transmission scheme uses Orthogonal Frequency Division Multiplexing (OFDM), where parallel data streams are transmitted by way of multiple low-level carriers, themselves modulated in a standard fashion (such as QAM or QPSK). In the specific application of IBOC, hundreds of subcarriers are used; the modulation scheme is QPSK. One of the big advantages to OFDM is its robustness, and that is in part attributable to a long symbol length. Symbol length is the amount time that the individual carriers are in a state that is detected by the receiver. Part of the symbol length, though, is called the guard interval. This is the amount of time that needs to elapse prior to the next symbol. In other words, there is not a continual flow of symbols; between changes, a certain amount of time expires -- the guard interval.
- continued on page 2
This is important because of the mitigation of multipath on the receiver end. If the receiver, in the field, encounters two IBOC signals that are on the same frequency, with synchronized data, there are two characteristics that will determine whether or not there is a destructive multipath effect:
■ One signal is considerably stronger than the other (say 15dB or more) in which case the weaker signal doesn't have any effect on the receivers detection of the subcarrier state changes on the stronger signal, or
■ Both signals are of nearly the same strength, and they are delayed by some amount of time because of different distances between the receiver and the various sources.
If that difference in time (also called the delay spread) is within the guard interval, the receiver can correlate the symbol changes in both received signals, and essentially they add. If the delay is outside (or longer) than the guard interval, then the picture gets blurry; the symbol changes are too far apart, and intersymbol interference (ISI) is caused. From the Harris/NPR article, Anders Mattsson states that the symbol length for HD Radio is 2.9ms, the guard interval is 156µs, and that "it seems that multipath delays of up to the guard interval of 156µs, should be acceptable. Ideally it should be less than 78µs." This leads to a practical limit of about 15 miles in difference between the two IBOC sources; considering that the receiver could be located directly between the two, that leads to a practical limit of about 30 miles between transmitter sites (based on 78µs).
Let's look at some real world examples now of HD Radio boosters that have actually been built and performance-tested. The first will be the KCSN booster noted by John Kean in the Harris/NPR article. It's interesting for several reasons, not the least of which is that the booster has a higher ERP than the main signal. It's also more of an old fashioned booster design in that it takes advantage of terrain shielding. Kean notes that the two transmitters made use of modulation synchronization and frequency synchronization.
Combined KCSN primary booster coverage with F(50,50) 60dBu contours overlaying Longley-Rice field strength prediction. (Image from Kean/Madsen paper.)
Part of NPR's evaluation process of the KCSN booster design involved the development of software that was used to predict interference between the two digital sources. This software was used for these purposes:■ The propagation time from the main, and from the booster, on a grid across the area of interest;
■ Compared the calculated field strength of main and booster on the same grid:
■ Determined the field ratios and delay spread that could result in ISI, at each point on the grid;
■ Mapped the points on the grid in a color-coded fashion.
The predicted results are shown below. Delay spreads were noted if greater than 75µs.
Map showing locations of potentially high primary and booster multipath as colored dots. Reddish dots indicate where the signal from the primary transmitter is stronger and bluish dots indicate where the signal from the booster is stronger; the signal ratios of the gray dots are within 1.5dB. The rectangular box shows the study area of the measurement test map on the next page. (Image from Kean/Madsen paper.) (Click to enlarge.)
Clearly the predictions showed that, in a roughly oval shape centered about the KCSN-FM1 transmitter site, that the digital performance was expected to be good. NPR took measurements of the on-air system inside of the box shown in the figure. The northern part of the box is an area that encompasses the Hollywood hills so (based on my prior experience) I would expect to see trouble there just based on the terrain. However, of special interest is the southern part of the box, along I-10, just north of Los Angeles proper.
- continued on page 3
Measurement of HD Radio reception around the KCSN booster. A Longley-Rice field strength prediction underlay shows locations where the booster's field strength is 50-60dBu (blue) or greater. (Image from Kean/Madsen paper.) (Click to enlarge.)
This map shows a detail of the box referred to above. (The route taken in the NPR van is shown as the black/grey/white contiguous squares.) From this map of actual field measurements one can see along I-10 that, for the most part, the measurements indicate the digital signal was available between 97 and 100 percent of the time. As also predicted from the earlier map we can see that considerable interference was noted towards the eastern end of the measurement route.
As anyone who has constructed a booster knows, the goal is to have a net increase in the number of listeners that can get an easily useable signal; the hope is that there will be considerably more of those than are left with a diminished signal because of new interference. Measurement results would seem to indicate that with KCSN, a whole new area (south of the Beverly Hills and north of LA proper) were given access to KCSN's HD Radio signals.
The WD2XAB results (from the NAB Fastroad paper) are interesting in that they show before and after field measurements after an appropriate time delay was added in the booster's transmission path.
ComStudy was used to determine areas of predicted overlap between the two RF sources; that ended up being near Hartford road. As part of this experiment, iBiquity took RF spectrum measurements in the center of the predicted overlap region along Hartford Road.
- continued on page 4
20µs misalignment, 3.7 miles delay spread. (From iBiquity/Fastroad paper.)
The multipath effect shown in the spectrum above is caused by the difference in time between reception of the main, and that of the booster.
To minimize ISI at this location, the engineers from iBiquity added 20µs of delay in the transmission path of the booster (WD2XAB-FM1). Making note of the solid green, contiguous lines in both figures 7.2, and 7.3, one can see that there is an improvement in the after measurements.
HD Radio Signal Coverage before Time Alignment (From iBiquity/NAB Fastroad paper.)
HD Radio Signal Coverage after Time Alignment (From iBiquity/NAB Fastroad paper.)
Stepping back now for a moment to the Harris/NPR paper; the following conclusions are drawn at its end: "Digital radio lends itself naturally to SFN implementations. The main advantage is the potential of very flexible coverage and easy expansion - simply add more transmitters. Depending on the length of the guard interval, some care will be needed to avoid excessive multipath. The hybrid IBOC system warrants some further studies with respect to the analog part of the signal, before it is clear how well SFN will work in this case." You may have thought of this question as well: If one station were to use a digital-only booster, what will be the effect on analog-only receivers near the HD Radio booster, where the ratio of analog to digital will be far, far less than 20dB? The WKLB HD Radio booster experiment (NAB Fastroad) endeavored to answer that, along with looking at the booster results for HD Radio.
Interference Effects - Separate Test Runs on a Single Map (From iBiquity/NAB Fastroad paper.) (Click to enlarge.)
Like the WD2XAB booster experiment mentioned previously, a time alignment was performed in the case of the WKLB booster. Figure 8.7 shows measurement data taken before, and (slightly to the right) after the time alignment was performed on the booster transmitter. Digital signal augmentation and improvement was noted between the main and booster sites, but actually made worse as one went north beyond the booster location. This is because of delay spread that exceeded 75µs.
- continued on page 5
The effect of the digital-only boosters on analog radios was tested along I-93:
Test Route for Determining Impact of Digital-only Booster on Analog Reception. (From iBiquity/NAB Fastroad paper.)
As you can see, the route was taken driving towards the booster (digital only) and away from the main site (hybrid transmission). As iBiqity's test vehicle moved along this route, the digital-to-analog power ratio varied from -6dBc to +20dBc as the distance closed from 5.65 miles to 1.3 miles, and then increased to four miles as the route passed the booster site. The test van was outfitted with a newer Delphi radio, an older Chrysler OEM radio, and an analog clock radio. The newer Delphi radio was able to withstand the IBOC sidebands, and continued receiving the analog signal from the main site; the older Chrysler radio suffered, "exhibiting audible interference from the beginning of the run to end of the run. Close to the booster, the audio was all but inaudible." The analog clock radio's audio was not evaluated because of the receiver's extremely poor selectivity.
Another set of tests were performed during the WKLB booster experiment: a small amount of analog FM (synchronized to the main) was transmitted from the booster site, to see if that would mitigate the effect of the higher-than-usual IBOC carrier levels (with respect to analog) in proximity to the booster site. According to the report, the results were "mixed.” From the text: "Using this approach, with similar main and booster analog signal levels, all receivers will suffer from the multipath-like interference."
The results of these three experiments seem to indicate that while it is possible to build a digital-only SFN, that implementation of such systems is still a ways off. As time goes forward and more and more new analog receivers appear, the effects of digital only, on analog reception, near the digital booster transmitter itself, will diminish.
Irwin is transmission systems supervisor for Clear Channel NYC and chief engineer of WKTU, New York. Contact him at firstname.lastname@example.org.
Harris/NPR paper on SFN
NAB Fastroad paper on SFN