DVB-T2 (T2) MISO versus SISO Field Test

DVB-T2 (T2) MISO versus SISO Field Test Author: Bjørn Skog, M.Sc. E-mail: bjorn.skog@telenor.com Company: Telenor Broadcast, Norkring AS, Norway July 3rd 2013 @ LS telcom Summit 2013 V.2 2.7.13

The Case To investigate what coverage improvement (MISO gain) could be obtained in a SFN network using the Distributed Alamouti Scheme or Space-Time Block Coding MISO = Multiple Input, Single Output SISO = Single Input, Single Output

MISO Figures 1 & 2 Figures from page 5 of http://www.lstelcom.com/fileadmin/content/marketing/flyer/general/white_paper_dvb-t2_planning.pdf

MISO Background / advantages To avoid the zero db echo problem or increase in required C/N for decoding and to fully utilize the single frequency network (SFN) gain: a cost effective coverage improvement technique in DVB-T2 is the option of using multiple-input single-output (MISO) mode, using the Alamouti technique. One transmitter (group) transmits a slightly modified version of each pair of data subcarriers in the reverse order of the other neighbor transmitter (group)

MISO details 1 The T2 receiver uses only one receiving antenna, while the Alamouti scheme allows coverage improvements equivalent to that obtained from a two antenna receiver diversity system and a maximal ratio combining scheme. One transmitter (group 1) transmits an un-modified version of every constellation pair as in SISO (single-input single-output) and DVB-T, neither buffering nor negation or complex conjugation is applied. One pair of constellation cell is labeled C0 and C1

MISO details 2 The transmitter (group 2) transmits a slightly modified version of each constellation pair, and in reverse frequency order. Group 2 transmits -C1* and C0* where * denotes the complex conjugation operation. In practice co-located MISO SFN transmissions are too expensive for most commercial deployments: The number of transmitters and transmitter antennas would then (co-located) double and even if the transmitter power potentially could be lowered in a SFN, in average the MISO gain could not compensate for nearly doubling the transmitter network cost, Especially for stationary reception.

MIMO Multiple Input Multiple Output Figures explaining MIMO, MISO and SISO Distributed MIMO (network cost effective: no new transmitting antennas or additional transmitters are required)

MIMO advantages The more complex and advanced diversity technique called MIMO (multiple-input multiple-output), is mainly developed and used to overcome the limitations of the SISO (and MISO) Shannon capacity limit. In wireless communication systems, the path towards high spectral efficiency transmission techniques, has been through the use of the diversity, provided by the rich scattering wireless channels. Diversity is in either time, frequency, space (antenna) or polarization diversity. It has been used to combat the fading, by trying to stabilize the channel. But even if MIMO is technically advanced and attractive to the engineer; If the frequency cost is low as in Norway for Broadcasting:

MIMO disadvantages This means in practice that there is no network cost advantage using MIMO for broadcasting there. Using MIMO 2x2, for the transmitter antennas this means new cross polarized antennas and 2 transmitters per site (double power) and for the receiving antennas, this means new cross polarized dual direction antennas and new receivers.

Future use of MIMO for BC 1 Considering the DVB.org Technical Module: TM-MIMO Study Mission report to the 94th TM meeting @ EBU, Geneva, 4th and 5th June 2013: The possible lowered field strength requirement for a new DVB-Tx variant currently considered, simulated and possibly later developed may prove significant gains, using Non-Uniform QAMs (NU-QAM) possibly of ultra high order, example: 256-NU-QAM, code rate 3/5, SNR 16 db (256-QAM is today in DVB-T2, but not NU) 1k-NU-QAM, code rate 3/5, SNR 20 db 4k-NU-QAM, code rate 3/5, SNR 24 db, plus better error control coding (FEC) plus Time Frequency Slicing. These techniques are considered being used in combination with MIMO.

Future use of MIMO for BC 2 This interesting work within DVB.org Technical Module can potentially give 100% to 200% more capacity per RF (8 MHz) channel or a significant reduction in the required field strength, and this work might end up with two terrestrial system descriptions: A simpler variant not consisting of MIMO A more complex one with MIMO If only MIMO versus MISO gain is considered, my rough network cost estimate for coverage of rural areas stationary reception probably is valid, meaning MISO is the most cost effective, with the probable exception of Distributed MIMO (the more expensive end-user equipment can then be used only where needed for gaining coverage). Testing this would be interesting future work

MIMO directional receiving antennas High gain UHF receiving antennas for MIMO is less practical

Practical consequences for DVB-T2 network operators and end-users This is the focus in this presentation and the work behind, network coverage and cost is especially relevant for countries with wast areas to be covered and relatively low population density. The need for a high percentage population coverage in each country, often is required by Governments, broadcasters and advertizers.

Test objectives Norkring as a terrestrial broadcasting network operator, wanted to better utilize the benefits of the T2 technology presently available, during T2 network planning of new or converted (from DVB-T) networks To investigate if using MISO versus SISO gave any coverage advantages or any technical problems.

Investigative method Investigative method was changing only one parameter (from SISO to MISO) and measure the required field strength for coverage or decoding of the video, in a real SFN environment with complex transfer channels between transmitters and receiver This change was executed at every measurement location and for several T2 modes. The transfer channel was always zero db echo from 2 transmitters, one in MISO Group 1, the other in MISO Group 2.

Statement of the problem MISO removes the RF spectrum ripples and notches that occur in a standard (SISO) SFN channel. This degrades the system s signal quality (Modulation Error Ratio, MER) and do not occur in a MISO SFN because the two transmitted signals are no longer identical, so destructive signal (for individual carriers) combination between transmitters in group A and B is avoided.

Measurement RF spectrum curves for MISO and SISO with zero db echo: Here the delay spread is nearly minimum (less than 1 µs), this gives a serious degradation of the SISO RF Spectrum (picture to the right) and none in the MISO RF Spectrum (picture to the left) At page 45 in this presentation, it is attached a Word report with all measurements points/locations details.

Why field test MISO? The testing of MISO versus SISO, was selected because this was the last unknown possible advantage with the current T2 equipment implementation. I write possible because there are two diverging conclusions concerning this, based on field measurements. This field test results comply favorably with the unpublished MISO results from Bavaria ( performed by Walter Fischer in R&S )

Results MISO Gain 1 For high capacity T2 modes, approximately 3 db gain or 3 db lowered field strength requirement for zero db echo receiving conditions was found. MISO can already be used with commercial T2 transmitters and receivers and has no additional network or equipment cost implications. This means MISO often gives a significant 3 db SFN coverage advantage, through reduction of practical zero db echo coverage problems, at zero network cost increase

MISO test details T2 Norway Mjøsa area SFN field test autumn 2012 test results (all modes used 32K number of (rotated) carriers) Conclusions from this report on the gain in using MISO vs. SISO for zero echo db receiving conditions between two neighboring T2 transmitters (received with equal amplitude at the antenna output): The gain here is defined as the required lowered input signal (or field strength) for decoding the video picture and audio for the T2 receiver (here using Sony s 2 nd generation T2 Chip), when one of the transmitters was in MISO Group 1 and the other transmitter received was in MISO Group 2. (when both transmitters are received with the same RF level at the receiver, here commonly defined as a zero db echo ).

Table: Statistical summary results for different code rates and QAM constellation levels Code Rate: 3/4 Constellation: 256 QAM Guard Interval: 1/16 Pilot Pattern 2 MISO versus SISO gain = 4,1 db Code Rate: 2/3 Constellation: 256 QAM Guard Interval: 1/16 Pilot Pattern 2 MISO versus SISO gain = 3,5 db Code Rate: 3/5 Constellation: 256 QAM Guard Interval: 1/16 Pilot Pattern 2 MISO versus SISO gain = 2,1 db Code Rate: 3/4 Constellation: 64 QAM Guard Interval: 1/16 Pilot Pattern 2 MISO versus SISO gain = 2,4 db Code Rate: 3/4 Constellation: 16 QAM Guard Interval: 1/16 Pilot Pattern 2 MISO versus SISO gain = 2,2 db MISO average Input signal requirement is 3,4 db lower than SISO for all modes/measurements using zero db echo between two transmitters, the gain is higher for 256 QAM and when less error protection is sent.

MISO Conclusion The MISO gain is significant and the only practical reason not using MISO in a SFN is because the modulation parameters wanted (for either having very large SFN or because of maximum bitrate capacity considerations) is not allowed to be used in combination with MISO. This may quite often be valid considerations, all the details concerning advantages/ disadvantages needs careful considerations which are outside the scope of this report. Limitations concerning guard interval size for MISO versus SISO in relation to SFN network planning after Digital Dividend 2, is discussed below under DD2.

Required RF input signal level in dbµv using Sony Bravia TV Figure 1: Required RF Input signal MISO (average 26,5 dbµv) and SISO (30,1 dbµv) CR for 2/3 256QAM, Sony Bravia TV 32 Zero db echo receiving DVB-T2 mode: Code Rate=2/3 256QAM Gl=1/16 PP2 31 30 29 28 27 26 25 24 1 2 3 4 5 6 7 8 9 10 11 Number of measurement locations MISO average=26.5 db SISO Average=30.1 db

Gain in db lower required input signal by using MISO instead of SISO, for the Sony Bravia TV Figure 2: MISO vs. SISO gain for CR 2/3 256QAM. Average gain = 3,5 db 6,0 Zero db echo receiving DVB-T2 mode: Code Rate=2/3 256QAM Gl=1/16 PP2 5,0 4,0 3,0 2,0 1,0 0,0 1 2 3 4 5 6 7 8 9 10 11 Number of measurement locations Gain: Average=3.5 db

Gain in db lower required input signal by using MISO instead of SISO, for the Sony Bravia TV Figure 3: MISO vs. SISO gain CR 3/5 256QAM. Average gain = 2,1 db: Lower gain due to better error protection, which allows for correcting errors due to faded carriers in SISO 4,0 Zero db echo receiving DVB-T2 mode: Code Rate=3/5 256QAM Gl=1/16 PP2. 3,0 2,0 1,0 0,0 1 2 3 4 5 6 7 Number of measurement locations Gain: Average=2.1 db

Gain in db lower required input signal by using MISO instead of SISO, for the Sony Bravia TV Figure 4: MISO vs. SISO gain for CR 3/4 256QAM. Average gain = 4,1 db 8,0 Zero db echo receiving DVB-T2 mode: Code Rate=3/4 256QAM Gl=1/16 PP2. 7,0 6,0 5,0 4,0 3,0 2,0 1,0 0,0 1 2 3 4 5 6 7 Number of measurement locations Gain: Average=4.1 db

Gain in db lower required input signal by using MISO instead of SISO, for the Sony Bravia TV Figure 5: MISO vs. SISO gain for CR 3/4 64 QAM. Average gain = 2,4 db 4,0 Zero db echo receiving DVB-T2 mode: Code Rate=3/4 64QAM Gl=1/16 PP2. 3,0 2,0 1,0 0,0 1 2 3 4 5 6 7 Number of measurement locations Gain: Average=2.4 db

Gain in db lower required input signal by using MISO instead of SISO, for the Sony Bravia TV Figure 6: MISO vs. SISO gain for CR 3/4 16 QAM. Average gain = 2,2 db 4,0 Zero db echo receiving DVB-T2 mode: Code Rate=3/4 16QAM Gl=1/16 PP2. 3,0 2,0 1,0 0,0 1 2 3 4 5 6 7 Number of measurement locations Gain: Average=2.2 db

Discussion of MISO gain 1 The gain is much higher for Code Rate 3/4 than for 3/5, because the more robust mode (3/5), means bit errors caused by selective fading is less harmful, popularly this can be described as: many carriers can be lost in noise and still the signal be decoded.

Discussion of MISO gain 2 The MISO gain was achieved due to the physical condition that the RF spectrum is flatter and that the MER for each RF carrier received in the zero db transfer channel is of better quality. This physical phenomenon occurs due to the fact that for SISO a zero db echo, Meaning here that two T2 transmitters are received with equal amplitude at the TV antenna, Results in many RF carriers with low level due to 180 degree phase shift between the two carriers. The mathematical condition is 1 + (-1) = 0, resulting in selective fading and low carrier level at that particular frequency. This means in practice that the T2 receiver using channel estimation, amplifies this carrier and also the noise. The noise vector is large relatively the carrier vector, and the C/N or MER for that carrier is low.

Details about MISO and measurements 1 For visually most impressing demonstrating the MISO gain fast and effectively, I recommend using low delay spread between the transmitters and code rate ¾. The average or median MER value per carrier may not often be so much higher, But you avoid low MER values on a significantly percentage of RF carriers. This is of particular usefulness, when code rates using less Error Protection bits are sent. DVB-T2 receiver used during the measurements finding the thresholds for decoding: Sony Bravia TV Model # KDL-22EX553 Serial # 1000266 Program version PKG1.114EUA-0001 Instrument: R&S ETL with DVB-T2 option serial # 102502 Firmware version 2.53 was used to measure the input signal, impulse response, Modulation Error Ratio (MER) as a function of frequency (per carrier), RF spectrum.

Details about MISO and measurements 2 Both short and long echo delays (delays spread between the two transmitters relative the Guard Interval) were used and both gave good MISO gain. Concerning coverage, the MISO gain was always positive, meaning no negative coverage results were measured for MISO versus SISO.

Details about MISO and measurements 2 The transfer channel from each of the two transmitters was possible for the T2 receiver Sony Bravia to estimate, because each transmitter (MISO Group 1 and 2) has its own pilot pattern, generally the pilots are inverted in transmitter group 2. This enables channel estimation and correction of each individual transmitter (group) transfer channel. The advantage is less bit errors received for a given input signal RF level with MISO compared to SISO SFN, or the same reception stability at lower field strength levels, potentially meaning lower network cost or increased coverage.

DD2 1st DD2 implies only 224 MHz bandwidth will be available for terrestrial TV 470 to 694 MHz (down from 320 MHz today in use in Norway). WRC-15 cannot dictate the use of this band to ensure more effective use of it, meaning much larger SFN size, in practice requiring T2, EU can decide in a new frequency conference. Particularly after digital dividend 2 (DD2) in the 700 MHz band in Norway and similar countries, The following issues related to DD2 are in this report evaluated:

DD2 2nd - the possible benefits and limitations for using MISO versus SISO in T2 for surviving such a RF channel / bandwidth constraint - compared with the possible network and end-user cost increase (drawback) and possible benefits (larger capacity per MUX and RF channel used), when alternatively using future MIMO compatible equipment (Multiple Input Multiple Output).

DD2 3rd Selecting the most appropriate mode for T2 involves optimizing multiple parameters that affect coverage, network cost and network capacity This task is much more complicated and involves advanced analysis for T2 relative for T1, due to the large number of transmission mode parameters possible to select using T2. Selecting a mode that gives significant coverage advantages with the least possible negative consequences for net bitrate capacity, is the challenge.

DD2 4th The most cost effective strategy for DVB-Tx (T1 or T2) in Norway surviving DD2, depends partly on frequency coordination with neighboring countries, for Norway most importantly Sweden, for creating larger DVB-Tx SFN areas. Nationwide or large SFNs are the most capacity maximizing strategy, but demands that both Norway and Sweden is using this to a large extent and also many other neighboring countries

DD2 5th If not nationwide SFN areas is decided because of regionalized content or other constraints/considerations, then probably the SFN size will be compatible with T2 guard intervals that is supported by MISO.

DD2 6th A UHF frequency conference for European countries for re planning all TV channels at UHF frequencies between 470 to 694 MHz, without DD1 and DD2 frequencies, would be the best solution for optimizing net capacity by using large SFN areas. The UHF frequency use currently in Norway is sub-optimally based on the Stockholm 1961 frequency conference

DD2 7th For surviving DD2, the most important tools are in this order: re-planning of SFN areas to better suit T2 and DD2, Use MISO when SFN areas are of appropriate size to get ca 3 db better zero db echo performance for high capacity modes for free (no network cost) and use H.265 video codec for achieving ca double bitrate video efficiency (relative MPEG4 H.264 currently used in commercial T2 receivers).

DD2 8th The best strategy for surviving a post DD2 regulated environment, proposal for this strategy in more detail: If we after DD2, in a most cost effective way, shall maximize the terrestrial TV network capacity and coverage, I think that the first and most important step would be, to have a frequency planning conference, that in general increases the SFN network size. For Norway and Sweden this probably mean country wide SFN networks, for all MUXes except perhaps one regionalized MUX. This would maximize the number of multiplexes possible, in a post-dd2 frequency regulated environment. The re-use distance could also be reduced, meaning some areas with more interference, but when selecting large SFN areas, the interference areas will be smaller/fewer and in border areas.

DD2 9th Having many multiplexes available will enable a robust modulation per multiplex, giving cost effective coverage (low power consumption, lower frequency re-use distance). One cost challenge for terrestrial TV network operators, is that all DVB-T transmitters and all combiners must be replaced by new equipment. The most cost effective future DVB-T2 (or T3 ) transmitters would be that they enabled the transmission of many MUX in as many RF 8 MHz channels. To reach this cost effective situation in practice, what is needed (requirement 1), is that the frequency planning conference group all RF channels adjacently for each country.

DD2 10th Example for 7 MUX: Sweden gets RF UHF channels 30, 31, 32, 33, 34, 35, 36 and 37 Norway gets RF UHF channels 38, 39, 40, 41, 42, 43, 45 and 46 etc. If one MUX shall be regionalized, to save frequencies it could be required that it uses more robust modulation, alternatively also gets lower population coverage in the country. The second requirement (rolling out many new MUXes cost effectively) is that the new transmitters have new multi-mux -modulators (one modulator that can send many MUX in many RF 8 MHz channels, TFS [Time Frequency Slicing] would be included in the functionality). The antenna combiner cost is then zero and the space for large antenna combiners are also saved.

DD2 11th I have talked to one DVB-T2 transmitter manufacturer, and the only way to save the antenna combiner cost, is to make multi-mux -modulators. Trying to do this using today s modulators with one MUX per modulator and combining them before the amplification, Only leads to large intermodulation products. If you, noble audience, agree to the above text, I propose that some of you promotes that the next UHF frequency conference in Europe adopts these strategies, And that transmitter manufacturers are asked to make such DVB-T2 modulators/transmitters.

The End Q&A The report with details about the measurements is saved in the icon to the right: DVB-T2_Norkring_MISO_SISO_test_autumn2012_VersionJune26_2013.doc This report s author can be contacted like this: bjorn.skog@telenor.com Mobile Phone: +47 9006 1049