DTV: Real World Channels Are Rayleighian

Following the earlier investigations of indoor and outdoor antenna reception in the United States and the problems that have been highlighted, I had an opportunity to repeat these investigations with the DVB-T COFDM system in the UK. These investigations took place with a new Nokia DVB-T receiver, which is NOT available in the UK, but which provides an opportunity to provide valuable REAL-WORLD insights into DTV receivability. Tests with this receiver are ongoing, using the off-air DTV channels, in the UK.

I am indebted to Nokia for making this remarkable consumer receiver available to me.

The key findings following real-world sampling and spectrum analyser plots are:

KEY FINDINGS:

In the real world virtually all DTV channels are TIME VARYING, and there is dynamic multipath in everyday situations with both OUTDOOR and INDOOR antennae. This means that DTV systems, which are optimised for AWGN channels, perform poorly in the field, as has been seen with the 8VSB system in the United States. (At the theoretical level this was highlighted in the recent UMas report commissioned by MSTV and which has been seen operationally in the US for the last eighteen months)

Contrary to the received wisdom, the COFDM decode threshold, is virtually identical in either simple or complex reception environments. This is IDENTICAL for both indoor and outdoor antenna reception. The baseband Signal-to-Noise ratio is 18.5dB for the DVB-T mode 64QAM FEC 2/3, GI, 1/32. This number is invariant from channel to channel.

Once the COFDM receiver attains the decode threshold it works in ALL receive environments irrespective of their complexity. (As has been seen in Brazil and the US this is NOT the case with the 8VSB DTV system.)

Separately this receiver was independently inspected by a number of well-known DTV solutions providers and silicon vendors who have access to significant hardware and software testbed facilities and are able to assess the receiver's performance in many modes including mobile DTV modes. Some of these results are disclosed, as this is a publicly available receiver in mainland Europe.

Nokia receiver

The receiver used in these tests is a new Nokia 9800 series receiver. It is a fully DVB-T compliant COFDM receiver able to decode all 2k/8k modes, all Guard Intervals, and all modulation standards (QPSK, 16QAM, 64QAM). Currently it is an 8Mhz UHF receiver.

The next version available in the summer will be a 6/7/8Mhz VHF/UHF receiver with full hierarchical decoding capabilities, able to provide analog PAL/SECAM/NTSC RF outputs as well as RGB, CVBS, digital and analog audio.

The current receiver supports mobile DTV as well as conventional roof-top/portable antenna DTV services. ABER/SET in Brazil, NBC in Philadelphia, and Sinclair in Baltimore used earlier versions of this receiver in trials in the Americas. The previous versions used in the Americas were modified for 6Mhz RF channel bandwidths.

The current receiver is considerably more sophisticated than the ONdigital receivers currently sold in the UK. In particular this receiver allow the expert user to investigate the DVB-T transmissions in some detail as well as operate it in its dual-function mode as a consumer receiver. In particular the receiver will:

1. Automatically scan all RF channels in all DVB-T modes
2. Permit direct manual entry of frequencies and modes
3. Allow the user to enter the offsets if the DTV network uses precision-offset working
4. An inspection of the MPEG2 TS is made available
5. Full and easy handling of DTV services transmitted from multiple markets
6. Direct baseband S/N read-out that can be superimposed on screen, which can be updated every second, whilst displaying DTV picture and sound in the background. This allows the S/N ratio to be tracked over time.
7. Interrogation of embedded/attached Conditional Access systems

The ability of the receiver to display information about the received RF channel, in particular the S/N ratio, is an invaluable diagnostic and measurement tool. In particular this facility is normally only available in professional DTV receivers, and is not available in the 'package' ONdigital receivers in the UK which are built to a cost-determined specification.

Receiver S/N threshold

It has proved possible, in every receive site, to determine the receiver S/N threshold to receive stable and reliable DTV pictures. This is accomplished by fitting variable and fixed attenuators in line with the antenna RF input until the picture just fails.

This is 18.5dB as displayed by the receiver, and interested parties have calibrated the receiver's measurement accuracy against external AWGN tests in the laboratory. These measurements indicate that the AWGN for the receiver is 18.5dB, indicating that the receiver's internal S/N meter is accurate to within 1dB. The threshold performance of this new Nokia receiver is 3-4dB better than current ONdigital receivers and is just 2dB off the theoretical limit of 16.5dB for the DVB-T mode: 64QAM, FEC 2/3, GI 1/32, 2k.

This threshold, after numerous site investigations, with both indoor and outdoor antenna is almost constant. This is a somewhat surprising conclusion, since previous channel models had indicated that in heavily multipathed locations the DTV decode threshold would be of order 23-26dB. This is NOT the case and may be explained by the additive integration properties of COFDM in complex receive environments. This threshold has been seen in many complex locations as well as simple locations for receiving DTV service.

Picture and sound are still available in the range 16 - 18.5dB, as displayed onscreen by the receiver, but errors are very noticeable rendering the service unusable.

Once the receiver attains the required C/N threshold the service is stable and reliable. However it can be seen that time-varying channels are almost always the norm, contrary to the expectation of designers of other DTV systems. This is seen by calling up, overlaying on screen and running the S/N measurement menu, on top of the DTV picture. Very large fades are seen in a second, even with directional outdoor antenna, with fades up to 15dB in a second. (Individual COFDM carriers may fade even more as determined by spectrum analyzers)

The decode S/N has been determined to be almost identical at test points three and fifty miles from the DTV stations in a particular DTV service area highlighting the uniformity of COFDM DTV signal delivery.

Real-world receivability

A number of test sites have been measured, both in the near and far fields, with the intention to investigate both indoor and outdoor antenna performance.

The parameters of the DTV stations in the service area surveyed are:

Previous measurements, corroborated by earlier computer coverage predictions, have indicated that these DTV stations are available up to fifty miles away, the radio horizon, with suitable 'legacy' outdoor antenna. This has now been confirmed in practice. Note that all DTV stations are colocated and broadcast from the same transmitter mast. These stations share the same mast with 1MW analog PAL transmissions

However the current study sought to investigate the 'near-field' performance of COFDM DTV in the real world with this receiver. A number of test points were selected to investigate the performance of this receiver and, more generally, the COFDM DTV system in the real world. The following test site is typical of the receivability conditions attained.

Reception Site 12

Description:

This site is in a suburban neighborhood of London. It is a recognised analog TV 'blackspot' with primary service provided by a translator rather than the main analog TV station. The house made available is a four-story house and there is a rooftop Yagi antenna (multibeam, antighosting type), that feeds a TV distribution system serving multiple floors in the house. The houses are terraced in a row, which creates attenuation problems for indoor DTV reception as the direct path to the DTV station is almost directly through the other houses.

RF conditions

The analog PAL services from the local main station exhibit a pronounced ghost, although the service is usable. This is because the outdoor antenna is highly directional with good side-lobe rejection and has a very good front-to-back ratio, and this feeds all TV distribution points. Indoor analog TV reception is unusable from the main station. Measurements with the outdoor antenna and a rotor fitted for the purpose indicate a very congested RF spectrum with many translators and at least one out-of-area main analog TV station present at varying signal levels. The translator station targeting the area provided excellent analog TV signals via indoor antenna as well as outdoor antenna.

DTV services are also present from the main station and the out-of-area DTV station. Measurements were made on the local DTV station as it was felt most consumers would opt for the local DTV station. A check of the rooftops confirmed this, with the vast majority of installations being single antennas pointed at either the local main or translator station. It was noted that the majority of the outdoor antennae were high-gain anti-ghosting directional Yagis.

The location is TWELVE miles from the DTV station, through densely populated/concrete canyon terrain.

Results

Results have been tabulated floor-by-floor at this location for both indoor and outdoor antennae. The first measurement provides the results for the primary feed from the OUTDOOR Yagi prior to the house distribution system. The Nokia was connected to the first drop and a TV set for display purposes.

The indoor antennae used were double bow ties with variable gain up to 36dB sourced from Philips and Altai, which are specifically sold as DTV ready antenna. These antennae have high variable gain and LNF (quoted under 3dB in one case). Both antennae had their gains set at maximum, because of the perceived indoor attenuation problems. The Altai costs around £18 and the Philips is £34, both have identical performance.

Outdoor antenna: Primary drop

Recall that the decode S/N is 18.5dB.

The most interesting aspect about the outdoor antenna's DTV performance is that very rapid time variation effects are seen with fades of up to 15dB in a second (individual carriers are worse), although the DTV picture is unimpaired. The other important fact highlighted by the Nokia display is that time varying effects are frequency selective, as DTV3 exhibits the classic characteristics of an AWGN channel, whereas the others did not.

This is all the more striking as all DTV services are radiated from the same station, and indicates that frequency selective effects are in play. Over a five day period DTV3 exhibited static AWGN channel characteristics whereas the others did not and constantly exhibited time varying S/N: This did not impact on DTV receivability as all fades were to values well above threshold (the lowest recorded S/N was 35dB).

From the primary drop the outdoor antenna feeds a conventional consumer grade distribution system (passive RF pass-through, no filtering, but with adjustable amplification.). This is used to compensate for the attenuation losses from the distribution amplifier caused by the very long cable runs within the dwelling.

Basement floor: Indoor and outdoor antenna DTV reception

The dwelling has a full distribution system with a basement outlet point. Analog indoor antenna reception (1MW ERP station) was unusable due to the attenuation. It proved possible to receive DTV services using the indoor antennae placed via the window, which looks out onto a road with passing traffic.

The following tables record the average S/N values for indoor and outdoor antenna reception, with comments.

The results for the basement were interesting.

The outdoor distribution system essentially levelled the signal strengths, as the distribution system was one floor above and the distribution system amplifier removed the large time varying variations. Indoor DTV reception relied on a reflected signal off a tall building clearly visible in the distance. This provided reliable indoor reception for four out of six DTV multiplexed channels and was unaffected by passing traffic.

As can be seen, given the station ERP's in the range 3.6kW to 10.0kW, the reflected signal was below or near threshold for two of the RF channels, even with the internal amplifiers on the indoor antenna set to maximum.

The contrast between the analog versions of the services carried in DTV1 and DTV2 could not be more striking: the analog services were unviewable and the DTV services were perfect. There were also a large number of consumer appliances generating impulse noise sources (the kitchen and other household appliance areas). All consumer appliances were activated: with the exception of DTV6 (at threshold), there was no discernible on-screen impact.

Ground floor: Indoor and outdoor antenna DTV reception

Outdoor antenna signal strengths have already been recorded, as this is the central ingress point for the rooftop antenna, prior to being fed into the distribution system. A measurement of the unamplified S/N via the indoor antennae was made by turning the gain to minimum, pointing in the TRUE direction towards the DTV stations, in the body of the house. The receiver recorded 16.5dB, below the threshold. Following the experience from the basement the indoor antennae were pointed towards the tall building to lock on the reflection, and the following results were obtained:

Note that via the indoor antenna, DTV3 was exhibiting very fast time varying fades, as recorded by the receiver. As the receiver was fading just above the threshold the service was unreliable with frequent break-ups.

Tantalisingly the direct S/N without amplified indoor antenna via 10kW ERP stations is 16.5dB, just 2dB short of the threshold. This suggests that if the UK DTV stations were increased in power by around 6dB to 50kW ERP the use of portable indoor antenna DTV reception would be widespread and would not require any indoor antennae with built-in low-noise amplifiers.

First floor: Indoor and outdoor antenna DTV reception

The results obtained were as follows:

Again the indoor antenna was pointed at the tall building, and it was noted that the path to the reflection was bisected by a major flightpath to the airport. It was noted that DTV services, via the indoor antenna, were not affected by passing aircraft confirming the reliability of the system.

Top-floor: Outdoor and indoor antenna DTV reception

The results obtained were as follows:

The results obtained via the distribution system feed on the fourth floor of the house indicate that the distribution system is inadequate for DTV at these ERP levels and with the very long cable runs in this house, and many RF outlets. The simple expedient of inserting a low-noise setback preamplifier restored DTV6 via the distribution system in a test

At this floor of the house it was possible to get a direct path towards the DTV station, albeit obstructed by terrain clutter and several very large buildings. The results were as follows:

Notice that by the top floor of the house, the indoor antenna outperforms the domestic distribution system (which clearly needs attention), by 5dB and is beginning to exhibit the time-varying Rayleighian behaviour. At this floor the outdoor Yagi antenna is clearly visible mast-mounted fifteen feet above. Note that the indoor antenna has a S/N exceeding 25.0dB, compared to the direct drop of the outdoor Yagi which is 50dB with time-varying fades of 15.0dB in a second. In both cases the receiver successfully and consistently decodes all DTV services as they are well above the system threshold of 18.5dB for this particular COFDM mode.

In addition the indoor antenna could be rotated through 120 degreees without loss of service, although the off-axis loss in S/N as measured by the receiver was 6dB down on the peak S/N.

Other results

The field-report from this site is typical of many others with this receiver taken both in near field and far-field sites. This receiver also handles the DTV overlapped transmission areas problem much more efficiently than the current packaged DTV receivers in the UK by the simple expedient of creating separate channel lists for each transmission area. The consumer then edits and manipulates the list, assuming that the RF inputs are diplexed, filtered and levelled.

Summary

Current results with this Nokia receiver indicate the following:

1. The S/N meter onboard the receiver demonstrates conclusively that DTV RF channels are time varying, even in the case of directional outdoor antennae. This indicates that the most appropriate channel model for DTV system designers is Rayleighian with time-varying multipath. COFDM DTV systems are able to track time-varying channels, attested to by the fact that there is no service discontinuity with time-varying multipath. This time-varying effect is also seen with indoor antennae. This has negative implications for other DTV systems, which are unable to track time-varying channels such as 8VSB.
2. For the DVB-T 2k 64QAM FEC 2/3, GI 1/32 mode the S/N threshold is 18.5dB almost irrespective of the receive environment. This runs counter to predictive models which indicated that in heavily multipathed areas the decode threshold would be 23-26dB. This indicates that COFDM DTV behaves in a very uniform way throughout the service area as 18.5dB is considered the norm for simple receive environments. Undoubtedly there is a system gain due to the ability of COFDM to additively integrate echos in the receiver in heavily multipathed areas.
3. Very high levels of indoor antenna reception are achievable with indoor antennae with internal high gain low noise amplifiers costing under $30. If the UK DTV stations were able to increase powers from 10kW ERP to, say, 50KW ERP the impact on low-cost indoor antenna reception would be very beneficial to broadcasters and consumers. In other countries where ERP constraints are less problematic COFDM DTV will lead to very high levels of reliable indoor antenna reception.
4. The results indicate that there is a gain of around 3dB in indoor antenna performance through each ascending floor in a home (corresponding to 10-15 feet height gain), in line with other measurements.
5. Operational experience indicated that, above threshold, DVB-T performs very well in the presence of time-varying effects.

Laboratory measurements

This receiver was measured by a number of DTV system integrators and semiconductor vendors at their facilities. Several vendors had the hardware and software measurement tools including full transmit chains to exercise this receiver in all DVB-T modes.

It was confirmed that the receiver AWGN is 18.5dB for the mode used in the operational UK DTV service, and one vendor was able to check the performance of the receiver with off-air pictures fed via a lab chain in a mobile DTV mode. This confirmed excellent performance, at very high speeds, in a channel supporting 8Mbit/sec (8Mhz channel) with a S/N ratio of 5dB. The receiver's performance in the 8k DVB-T modes was also tested, which is not possible with off-air DTV transmissions in the UK which are 2k only transmissions.

It was also shown that the receiver works very well with professional DVB encoder equipment enabling it to be used as a baseline receiver in future DTV trials around the world.

© TBS 2000